Systematic review and meta-analysis of the association between childhood overweight and obesity and primary school diet and physical activity policies

Abstract

Obesity is a major public health concern and there are increasing calls for policy intervention. As obesity and the related health conditions develop during childhood, schools are being seen as important locations for obesity prevention, including multifaceted interventions incorporating policy elements. The objective of this systematic review was to evaluate the effects of policies related to diet and physical activity in schools, either alone, or as part of an intervention programme on the weight status of children aged 4 to 11 years. A comprehensive and systematic search of medical, education, exercise science, and social science databases identified 21 studies which met the inclusion criteria. There were no date, location or language restrictions. The identified studies evaluated a range of either, or both, diet and physical activity related policies, or intervention programmes including such policies, using a variety of observational and experimental designs. The policies were clustered into those which sought to affect diet, those which sought to affect physical activity and those which sought to affect both diet and physical activity to undertake random effects meta-analysis. Within the diet cluster, studies of the United States of America National School Lunch and School Breakfast Programs were analysed separately; however there was significant heterogeneity in the pooled results. The pooled effects of the physical activity, and other diet related policies on BMI-SDS were non-significant. The multifaceted interventions tended to include policy elements related to both diet and physical activity (combined cluster), and although these interventions were too varied to pool their results, significant reductions in weight-related outcomes were demonstrated. The evidence from this review suggests that, when implemented alone, school diet and physical activity related policies appear insufficient to prevent or treat overweight or obesity in children, however, they do appear to have an effect when developed and implemented as part of a more extensive intervention programme. Additional evidence is required before recommendations regarding the focus of policies can be made and therefore, increased effort should be made to evaluate the effect of policies and policy containing intervention programmes upon weight status.

Full text

R E V I E W Open Access
Systematic review and meta-analysis of the
association between childhood overweight and
obesity and primary school diet and physical
activity policies
Andrew James Williams
1,2
, William E Henley
1
, Craig Anthony Williams
2
, Alison Jane Hurst
1
, Stuart Logan
1
and Katrina Mary Wyatt
1*
Abstract
Obesity is a major public health concern and there are increasing calls for policy intervention. As obesity and the
related health conditions develop during childhood, schools are being seen as important locations for obesity
prevention, including multifaceted interventions incorporating policy elements. The objective of this systematic
review was to evaluate the effects of policies related to diet and physical activity in schools, either alone, or as part
of an intervention programme on the weight status of children aged 4 to 11 years. A comprehensive and
systematic search of medical, education, exercise science, and social science databases identified 21 studies which
met the inclusion criteria. There were no date, location or language restrictions. The identified studies evaluated a
range of either, or both, diet and physical activity related policies, or intervention programmes including such
policies, using a variety of observational and experimental designs. The policies were clustered into those which
sought to affect diet, those which sought to affect physical activity and those which sought to affect both diet and
physical activity to undertake random effects meta-analysis. Within the diet cluster, studies of the United States of
America National School Lunch and School Breakfast Programs were analysed separately; however there was
significant heterogeneity in the pooled results. The pooled effects of the physical activity, and other diet related
policies on BMI-SDS were non-significant. The multifaceted interventions tended to include policy elements related
to both diet and physical activity (combined cluster), and although these interventions were too varied to pool
their results, significant reductions in weight-related outcomes were demonstrated. The evidence from this review
suggests that, when implemented alone, school diet and physical activity related policies appear insufficient to
prevent or treat overweight or obesity in children, however, they do appear to have an effect when developed and
implemented as part of a more extensive intervention programme. Additional evidence is required before
recommendations regarding the focus of policies can be made and therefore, increased effort should be made to
evaluate the effect of policies and policy containing intervention programmes upon weight status.
Keywords: Policy, Obese, Nutrition, Physical education
* Correspondence: k.m.wyatt@exeter.ac.uk
1
Institute of Health Services Research, University of Exeter Medical School
(formerly Peninsula College of Medicine and Dentistry), Veysey Building,
Salmon Pool Lane, EX2 4SG, Exeter, Devon, UK
Full list of author information is available at the end of the article
© 2013 Williams et al.; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative
Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and
reproduction in any medium, provided the original work is properly cited.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101
http://www.ijbnpa.org/content/10/1/101

Introduction
Obesity among children is associated with significant
psychological, social and health consequences including
insulin resistance, cardiovascular disease, low self-esteem
and poorer education and employment outcomes [1,2].
The rising prevalence of obese children combined with the
increased likelihood of obesity continuing into adulthood,
has resulted in childhood being seen as an important period
for interventions to prevent overweight and obesity [3-5].
There are increasing calls for governments to implement
policies which could halt the rise in obesity, similar to the
policies initiated to address smoking [6,7]. Within the
United States of American (USA) and the United Kingdom
(UK) there have been guidelines and policies introduced by
the governments to promote healthy behaviour in school
children, however, the impact of such guidelines and pol-
icies is rarely evaluated scientifically [8-10]. This system-
atic review was conducted to examine the effect of school
diet and physical activity related policies on anthropomet-
ric outcomes during primary education (primary or junior
school in the UK, elementary school in the USA).
To date the systematic reviews which have examined
the effect of obesity related school policies have evalu-
ated diet and physical activity outcomes rather than
weight status [11-16]. Jaime and Lock [11] and Van
Cauwenberghe [14] identified policy components which
appear to have a positive effect upon diet, including: nu-
trition guidelines; healthy food price interventions and
fruit and vegetable distribution or subscription schemes.
They found a lack of evidence for policies affecting
children’s breakfast or unhealthy food choices [11,14].
Lagarde and LeBlanc [15] identified a number of studies
which reported an increase in physical activity as a re-
sult of policy such as: improving the quality and variety
of physical education (PE); mandatory qualifications for
PE teachers and adequate facilities. This paper extends
existing work by systematically reviewing the evidence
for the effect of diet and physical activity policies on
children’s weight status [11,14-16].
Review
Methods
Guidance from The Cochrane Collaboration and the
National Health Service Centre for Reviews and Dissem-
ination informed the development of the review proto-
col, which is available upon request [17,18].
Search strategy
Two search strategies were developed for this systematic
review, one for diet related and one for physical activity
related policies (Additional file 1). Each search strategy
contained population terms, intervention terms and out-
come terms with only the intervention terms differing
between the two searches. Each set of terms included
thesaurus terms or Medical Subject Headings (MeSH) as
well as title and abstract text searches.
The following databases were searched from their earli-
est record to June 2011: Medline In-Process & Other
Non-Indexed Citations [Ovid], Medline [Ovid], EMBASE
[Ovid], PsychINFO [Ovid], SportDISCUS [Ebscohost],
Web of Science [ISI Web of Knowledge], Education
Resource Information Center (ERIC) [Dialog Datastar],
British Education Index [Dialog Datastar], Australian Edu-
cation Index [Dialog Datastar], Cumulative Index to Nurs-
ing and Allied Health Library (CINAHL Plus) [Ebscohost],
and The Cochrane Library [Wiley Online]. The search
strategy was developed in Medline (Additional file 1) prior
to adaptation for the other databases (the complete search
log is available upon request).
A grey literature search for unpublished and continuing
research was undertaken in July 2011 in the metaRegister
of Controlled Trials, Clinical Trials.gov and the Inter-
national Clinical Trials Registry Platform [18]. Similarly the
Robert Wood Johnson Foundation website was searched
for items not published within journals [19]. The following
search term was used ‘school and (physical activity or
physical education or nutrition or diet) and policy’with the
age limiter ‘child’where it was available. The references of
included studies and systematic reviews were inspected for
any additional studies.
Eligibility criteria
The eligibility criteria are outlined in Table 1. There was
no date, geographic or language restrictions. The popu-
lation of interest in this systematic review was children
aged 4 to 11 years participating in full time education.
The definition of policy utilised to identify whether an
intervention was eligible was that defined by Milio [20].
‘Policy is a guide to action to change what would
otherwise occur…Policy sets priorities and guides
resource allocation.’
p622, Milio [20]
Studies which evaluated national, regional and school
specific policies related to diet or physical activity
during primary education, including multifaceted inter-
ventions which included a policy component using an
anthropometric outcome were considered eligible.
Given that policies are unlikely to be introduced experi-
mentally with controls, controlled before and after
studies and interrupted time series, cohort and cross-
sectional studies were considered eligible as well as
randomised controlled trials. A minimum follow-up or
exposure to the policy was set at six months in line with
the National Institute for Health and Clinical Excellence
Obesityguidance[21].
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 2 of 22
http://www.ijbnpa.org/content/10/1/101

Study identification
Having removed any duplicates using reference manage-
ment software all article titles were screened by AJW.
The resulting titles and abstracts were independently
assessed for eligibility by AJW and AJH and the full texts
of all potentially eligible articles were retrieved for inde-
pendent review. Any disagreements were resolved by
discussion. Articles deemed eligible went on to data ex-
traction and quality assessment.
Data extraction and quality assessment
The following data were extracted from each eligible art-
icle: study design; geographic location of study (country);
source of funding; ethics approval; recruitment; sum-
mary characteristics of the study population; details of
the intervention (policy name, target and any assessment
of uptake); treatment of any control group; definition of
obesity; duration of follow-up/exposure; and results.
Standard tools were used to assess the quality of the
studies [22-24]. The data extraction and quality assess-
ment tool was piloted for suitability on four papers by
AJW and AJH. Data extraction and quality assessment
were undertaken by AJW and checked by AJH or KMW,
any disagreement was resolved through discussion. In-
formation was also extracted on whether and which
stakeholders were involved in the development and im-
plementation and whether the policy engaged families.
Further details on three of the policies was sought (e.g.
manuals, policy criteria) to identify the policy compo-
nents [25-27].
Data analysis
As diet and physical activity are distinct concepts, an
overall meta-analysis was not considered to be appropri-
ate, instead policies which sought to affect similar behav-
iours, such as nutrition guidelines, were clustered and
analysed. Standardised mean difference (Cohen’sd)in
body mass index standard deviation score (BMI-SDS),
were calculated for each study using standard calcula-
tions and the R package MAd [28-30]. As there is a posi-
tive bias in Cohen’sd-values calculated from studies
with small sample sizes, effect sizes were adjusted into
Hedges’g-values [28-30].
Where studies reported multiple comparisons within
each cluster (i.e. for girls and boys or for multiple time-
points), we first calculated Hedges’gfor each compari-
son separately. Where a study did not report the com-
bined effect and variance, we calculated the weighted
mean of the multiple effects [28]. Where necessary the
covariate outcome correlation or multiple correlation
among studies using independent samples was assumed
to be 0.3 (r = 0.3), whereas the correlation between pre-
and post-scores was assumed to be 0.6 (r = 0.6). These as-
sumptions were tested with sensitivity analysis reported in
Additional file 2.
For studies that did not account for the potential clus-
tering within schools, prior to effect size calculation, we
divided the reported sample size by a ‘design factor’(1 +
[(m-1) × ICC]), where ‘m’is the average number of par-
ticipants in each school and ‘ICC’is the intra-cluster
correlation [17]. An ICC of 0.01 was chosen based on
the findings of Johnson, et al. [17,31]. Further details on
the calculation of Cohen’sd-values prior to adjustment
into Hedges’gand the combination of effect sizes can be
found in Additional file 3.
Random effects meta-analysis of each cluster was
undertaken in Stata [32] to obtain pooled estimates of
the effect of each policy cluster. We quantified the ex-
tent to which the between-study variability observed was
due to true between-study differences (rather than to
chance) using the I
2
statistic [28].
Results
Identified studies
The study identification process and reasons for exclusion
are illustrated in Figure 1 [33]. A total of 6894 unique
Table 1 Eligibility criteria
Inclusion criteria Exclusion criteria
Population: children undertaking primary education aged between 4
and 11 years
Population: people outside the specified age range and animal models
Intervention: diet or physical activity related school policies either alone
or as part of intervention programmes
Intervention: policy components which are insufficiently described to
enable replication.
Outcome: body mass index (using valid reference curves to define
overweight and obesity), body mass index z-score or standard
deviation score, percentage of body fat, waist circumference, waist-to
-hip ratio, waist-to-height ratio, skin pinch/skin fold thickness
Outcome: change in diet, physical activity or knowledge
Context: primary school or equivalent Context: clinical settings
Study design: any experimental or observational study design
(randomised controlled trial, controlled before and after study,
interrupted time series, cohort study or cross-sectional study)
Study design: narrative reviews, editorials, opinions and letters, reports
published as meeting abstracts only (where insufficient methodological
details are reported to allow critical appraisal of study quality)
Follow-up: ≥6 months [21]Follow-up: <6 months
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 3 of 22
http://www.ijbnpa.org/content/10/1/101

records were retrieved from the database and grey litera-
ture searches. Through the process of screening, title and
abstract review and full text assessment, 25 articles were
identified as potentially eligible for inclusion. Examining
the bibliographies of these articles identified one add-
itional paper [34]. These articles reported on 24 studies,
three of which had yet to publish any results and, conse-
quently, could not be included in the analysis [35-37]. The
remaining 21 eligible studies are summarised in Table 2.
Study characteristics and quality
Ten studies examined diet related policies, five physical
activity related policies, and six examined policies with
both diet and physical activity related components
(henceforth known as combined policies) (Table 2). Des-
pite the lack of time restrictions on the searches all the
included studies had been published since 2003. Sixteen
of the studies took place in the USA with the remaining
five taking place in: Australia, Canada, Italy, Mexico, and
the UK. The 21 studies employed the following study
designs: randomised controlled trial (2 studies), con-
trolled before and after study (3 studies), cohort study
(11 studies) and cross-sectional study (5 studies). Five of
the cohort studies analysed data from the Early Child-
hood Longitudinal Survey –Kindergarten (ECLS-K),
these studies are easily identifiable in the forest plots,
and no more than two studies using this cohort are ever
combined [38-44].
All the included studies examined BMI as an outcome
categorised as overweight or obese, or adjusted to
standard deviation scores (BMI-SDS), percentiles (BMI%),
growth rates or Healthy Fitness Zone (BMIHFZ) [45]. The
Healthy Fitness Zone is another categorisation of BMI-
SDS like overweight and obesity associated with body fat
and therefore could be analysed like overweight and
obesity [45]. Additional outcome measures included: fat
mass index (FMI), body fat percentage, waist-hip ratio and
waist-height ratio, however, these outcomes were only
reported by a small number of studies and therefore were
not meta-analysed. As overweight and obesity are cut-
points along the scale of BMI-SDS, odds ratios were
converted to effect sizes following the method detailed by
Chinn [46]. When studies reported BMI both continu-
ously and categorically, these results were combined using
the methods outlined by Borenstein, et al. [28]. One study
reported unadjusted BMI as an outcome as the subjects
were adults who had been exposed to the USA National
School Lunch Program (NSLP) as children, this study was
excluded from the meta-analysis [47].
Study quality is summarised in Table 3. Of the obser-
vational studies, the majority utilised a sample which
was representative or somewhat representative of the
population. Twelve of the studies adjusted the results for
socioeconomic status, ethnicity or additional factors. All
except one study assessed the outcome independently
from the assessment of exposure, and all studies had suf-
ficient follow-up duration. Seven observational studies
lost less than 20% of the sample during the study, how-
ever, five studies experienced loss to follow-up at a level
which may have introduced bias.
Records identified through
database searching
(n = 12144)
Additional records identified
through other sources
(n = 583)
Records after duplicates removed
(n = 6894)
Records screened
(n = 6894)
Title and abstract reviewed for
eligibility (n = 665)
Full-text articles assessed for
eligibility (n = 66)
Articles included in systematic
review (n = 26)
Intervention articles (n = 7
including 3 in progress)
Observational articles (n = 19)
Searching
bibliographies
(n=1)
Records excluded (n = 6229)
Population (n=1181)
Intervention (n=1427)
Outcome (n=715)
Context (n=1000)
Study design (n=1901)
Follow-up (n=5)
Records excluded (n = 599)
Population (n = 43)
Intervention (n = 99)
Outcome (n = 135)
Context (n = 40)
Study design (n = 279)
Follow-up (n = 3)
Full-text articles excluded (n = 41)
Population (n = 16)
Intervention (n = 11)
Outcome (n = 4)
Context (n = 0)
Study design (n = 9)
Follow-up (n = 1)
Figure 1 PRISMA Flow diagram [32] of the identification of literature for inclusion in this systematic review.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 4 of 22
http://www.ijbnpa.org/content/10/1/101

Table 2 Summary study characteristics
Study/ Location Study design/
Sample size
Policy Gender Age/ Follow-up
or exposure
duration
Ethnicity Socioeconomic
status
Baseline
weight status
Outcome
measure(s)/
Growth
reference
Diet policies
Foster, et al. 2008
[55]/ USA
Randomised
controlled
trial/ n = 844, I:
n = 479, C: n =
365
School nutrition
policy initiative
I: 45.0% males, 55.0%
females, C: 47.8%
males, 52.2% females
I: mean ± SD 11.1
± 1.0 years, C:
mean ± SD 11.2 ±
1.0 years/ 2 years
I: 44.3% black, 22.4%
Hispanic, 17.1% Asian,
10.7% white, 5.5%
other, C: 46.8% black,
27.7% Asian, 14.2%
white, 5.8% Hispanic,
5.5% other
Not described I: 17.2%
overweight,
25.34% obese,
C: 16.5%
overweight,
21.8% obese
BMI-SDS,
overweight,
obese/ CDC 2000
Baxter, et al. 2009
[53]/ USA
Cohort study/
n = 1,557
Location of School
Breakfast Program
consumption
Males and females 9–10 year olds/
4 years
90% black Not described Not provided BMI%
Henry, 2006 [50]/
USA
Cohort study/
n = 7,446
National School
Lunch Program
Males and females 4–10 year olds/
3 years
66% white, 26%
African American, 5%
Hispanic, 4%
American Indian, 3%
Asian
26% eligible for
FSM
Kindergarten:
4%
overweight,
4% obese, 3
rd
grade: 4%
overweight,
6% obese
Overweight from
BMI/ CDC2000
Hernandez,
Francis and Doyle,
2003 [41]/ USA
Cohort study/
n = 1,140
National School
Lunch Program
50% males, 50%
females
Mean ± SD 6.2 ±
0.4 years/ 9 years
54% white, 24%
Hispanic, 12% black,
10% other
37% household
income < $20,000
Mean BMI% ±
SD,
Kindergarten:
63.3 ± 28.0, 1
st
grade: 62.1 ±
29.8, 3
rd
grade:
66.6 ± 28.8, 5
th
grade: 69.4 ±
28.7
BMI/ CDC 2000
Hinrichs, 2010
[47]/ USA
Cohort study/
n = 130,353
National School
Lunch Program
47.4% males, 52.6%
females
Not provided
(studied adults
who had
participated in
policy during
childhood)
88.0% white, 10.3%
black, 1.6% other
Not described Males: 42.5%
overweight,
8.0% obese,
Females:
22.4%
overweight,
7.4% obese
BMI, overweight,
obese
Millimet, Tchernis
and Husain, 2008
[43] and 2010
[44]/ USA
Cohort study/
n = 13.531
National School
Lunch Program and
School Breakfast
Program
50.7% males, 49.3%
females
Mean ± SD 9.2 ±
0.4 years/ 3 years
57.9% white, 17.4%
Hispanic, 13.8% black,
4.5% Asian
Mothers
education: 19.8%
high school, 28.1%
some college,
14.4% bachelor’s
degree, 8.4%
advanced college
degree
Kindergarten:
25.8%
overweight,
11.4% obese,
3
rd
grade:
32.5%
overweight,
17.1% obese
BMI%, BMI growth
rate/ CDC 2000
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 5 of 22
http://www.ijbnpa.org/content/10/1/101

Table 2 Summary study characteristics (Continued)
Millimet and
Tchernis, 2009
[42]/ USA
Cohort study/
n = 7,824
School Breakfast
Program
51.3% males, 48.7%
females
Mean ± SD 9.1 ±
0.3 years/ 5 years
55.4% white, 19.1%
Hispanic, 13.7% black
Mean
socioeconomic
status index 0.06
± 0.77
3
rd
grade:
36.5%
overweight or
obese, 5
th
grade: 41.4%
overweight or
obese
BMI growth rate/
CDC 2000
Ramirez-Lopez,
et al. 2005 [54]/
Mexico
Cohort study/
n = 360, I: n =
254, C: n = 106
School Breakfast
Program
Males and females I: mean ± SD 8.6 ±
1.3 years, C: mean
± SD 8.4 ±
1.3 years/
9 months
Not described Not described I: 10.6%
overweight,
10.6% obese,
C: 8.5%
overweight,
11.3% obese
BMI, body fat%,
overweight,
obese/ CDC 2000
Fox, et al. 2009
[56]/ USA
Cross-sectional
study/ n = 706
Nutrition guidelines 51% males, 49%
females
Mean 8.8 years/ >
1 year
52% white, 24%
Hispanic, 17% black,
7% other
48.49% eligible for
FSM
Not described BMI-SDS, obese/
CDC 2000
Jones, et al. 2003
[34]/ USA
Cross-sectional
study/ n = 772
National School
Lunch Program and
School Breakfast
Program
50% males, 50%
females
50% aged 5–
8 years, 50% aged
9–12 years/ up to
7 years
58.2% Black, 25,8%
white, 10.4% Hispanic,
0.1% other
Head of
household has
<12 years
education 33.9%,
household food
insecure 24.0%
34.2%
overweight or
obese
Overweight and
obese from BMI%/
CDC 2000
Physical activity policies
Donnelly,
et al.2009 [62]/
USA
Randomised
controlled
trial/ n = 1,527,
I: n = 814, C:
n = 713
Physical activity across
the curriculum
48.8% males, 51.2%
females
7–9 year olds/
3 years
77.4% Caucasian, 10.1%
Hispanic, 6.2% African
American, 3.6% multi-
ethnic, 1.6% Native
American, 1.2% Asian
43% eligible for
FSM
Mean BMI ±
SD I: 17.9 ± 3.1,
C:18.0 ± 3.7
BMI/ CDC 2000
Heelan, et al. 2009
[61]/ USA
Controlled
before and
after study/
n = 324, I: n =
201, C: n = 123
Walking school bus
scheme
44.8% males, 55.2%
females
Mean ± SD I: 8.1 ±
1.7 years, C: 8.4 ±
1.6 years/ 2 years
90% white, 7%
Hispanic, 3% other
~30% eligible for
FSM
Mean BMI% ±
SD I:67.6 ±
22.3, C:61.6 ±
29.1
BMI-SDS,% body
fat/ CDC 2000
Chiodera,
et al.2008 [60]/
Italy
Cohort study/
n = 4,500
Professionally led PE 51.1% males, 48.9%
females
6–10 year olds/
8 months
Not described Not described Mean BMI ±
SD: grade 1
16.3 ± 2.3,
grade 2 16.9 ±
2.5, grade 3
17.2 ± 2.6,
grade 4 17.9 ±
3.1, grade 5
18.6 ± 3.1
BMI
Datar and Sturm,
2004 [38]/ USA
Cohort study/
n = 9,751, I:
n = 8,917, C:
n = 834
Increased PE duration
of 1 hour per week
50% males, 50%
females
4–6 year olds/
1 year
I: 61% white, 16%
Hispanic, 12% black,
11% other, C: 58%
white, 20% black, 15%
Hispanic, 8% other
I: 13% family
income < $15,000,
C: 16% family
income < $15,000
I: 15%
overweight,
11% obese, C:
15%
overweight,
12% obese
BMI/ CDC 2000
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 6 of 22
http://www.ijbnpa.org/content/10/1/101

Table 2 Summary study characteristics (Continued)
Fernandes, 2010
[39] and
Fernandes and
Sturm, 2011 [40]/
USA
Cohort study/
n = 8,246
Meeting the National
Association for Sport
and Physical
Education (NASPE)
guidelines
50.4% males, 49.6%
females
6–11 year olds/
5 years
61.2% white, 18.7%
Hispanic, 13.1% black,
7.0% other
11.3% below the
poverty threshold
Mean BMI% ±
CD 60.8 ± 28.3,
13.3% obese
BMI%/ CDC 2000
Combined policies
Johnson, et al.
2012 [31]/
Australia
Controlled
before and
after study/
n = 1318
Be Active Eat Well I: 46.3% males, 53.7%
females, C: 50.8%
males, 49.2% females
Baseline mean ±
SD –I: 8.16 ± 2.25,
C: 8.19 ± 2.15.
Follow-up mean
±SD–I: 11.1 ±
2.26 C:10.3 ± 2.14
Parents born overseas
[27]* I: 6% C: 12%
Mothers didn’t
complete high
school education
I: 47.1% C: 40.6%
Baseline mean
BMI-SDS ± SD
–I: 0.59 ± 0.92,
C: 0.60 ± 0.87.
Follow-up
mean BMI-SDS
±SD–I: 0.54
± 0.94 C:0.59
± 0.88
BMI-SDS/ CDC
2000
Jordan, et al. 2008
[64]/ USA
Controlled
before and
after study/
n = 577
Utah’s Gold Medal
Schools
I: 51% males, 49%
females, C:52% males,
48% females
Mean ± SD I: 9.0 ±
1.6 years, C: 9.0 ±
1.6 years/ 1 year
I: 85.8% white, 7.6%
Hispanic, 2.8%
Hawaiian, 0.7% Asian,
0.4% American Indian,
0.0% African
American, 2.8% other
C: 86.7% white, 7.0%
Hispanic, 2.1% African
American, 0.7%
American Indian, 0.7%
Asian, 0.4% Hawaiian,
2.5% other
Maternal
education: <high
school I: 1.7%, C:
4.9%, high school
graduate I: 19.9%,
C: 25.9%, some
college I: 41.9%, C:
41.6%, college
graduate I: 32.0%,
C: 25.5%, graduate
degree I: 4.5%, C:
2.1%
Not described BMI-SDS/ CDC
2000
Chomitz, et al.
2010 [63]/ USA
Cohort study/
n = 1,858
Healthy living
Cambridge kids
51.8% males, 48.2%
females
Mean ± SD 7.7 ±
1.8 years/ 3 years
37.3% white, 36.9%
black, 14.0% Hispanic,
10.2% Asian, 1.7%
other
43.3% from low
income families
Mean BMI-SDS
± SD 0.7 ± 1.1.
16.8%
overweight,
20.2% obese
BMI-SDS,
overweight,
obese/ CDC 2000
Harrison, et al.
2011 [57]/ UK
Cross-sectional
study/
n = 1,724
Variety of diet and
physical activity
related policies
44.4% males, 55.6%
females
Mean ± SD 10.3 ±
3.1 years/ 5 years
Not described Age parent left
full time
education:
<16 years 46.5%,
16–18 years
33.4%, >18 years
20.1%
16.8%
overweight,
5.2% obese
Fat mass index
(FMI)/ IOTF
Veugelers and
Fitzgerald, 2005
[58]/ Canada
Cross-sectional
study/ 279
schools
Nutrition policy and
Annapolis valley
health promoting
schools project
Males and females 10–11 year olds/
5 years
Not described Not described 32.8%
overweight,
9.9% obese
Overweight,
obese from BMI-
SDS/ IOTF
Zhu, et al. 2010
[59]/ USA
Cross-sectional
study/ 738
schools
Variety of diet and
physical activity
related policies
Males and females Not described/ up
to 6 years
Not described 53% eligible for
FSM
Mean ± SD
71.7% ± 12.6
within BMIHFZ
BMIHFZ
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 7 of 22
http://www.ijbnpa.org/content/10/1/101

Table 3 Summary of study quality
Randomised
controlled trials
Random allocation Baseline measurement Reliability of outcome
measure
Blinding Adequacy of follow-up Protection against
contamination
Donnelly, et al. 2009
[62]
Completed Completed Not clear Completed Adequate Completed
Foster, et al. 2008 [55] Completed Completed Sufficient Not blinded Significant loss to follow-
up
Completed
Controlled before-
after studies
Second site control Baseline measurement Reliability of outcome
measure
Blinding Adequacy of follow-up Protection against
contamination
Heelan, et al. 2009
[61]
Sufficient Completed Sufficient Not clear Significant loss to follow-
up
Completed
Johnson, et al. 2012
[31]
Sufficient Completed Sufficient Not done Adequate Completed
Jordan, et al. 2008
[64]
Sufficient Not clear Not clear Not clear Significant loss to follow-
up
Completed
Cohort Studies Representativeness of
the cohort/sample
Comparability of
cohorts
Ascertainment of
exposure
Assessment of
outcome
Duration of exposure Adequacy of exposure
Baxter, et al. 2009 [53] Not described Not indicated Measured as part of the
study
Independent of
exposure
Sufficient No statement
Chiodera, et al. 2008
[60]
Representative Did not control for
socioeconomic status or
ethnicity
Measured as part of the
study
Independent of
exposure
Sufficient Sufficient
Chomitz, et al. 2010
[63]
Somewhat representative Controlled for ethnicity,
socioeconomic status and
additional factors
Measured as part of the
study
Independent of
exposure
Sufficient Subjects lost to follow-up
unlikely to introduce bias
Datar and Sturm, 2004
[38]
Representative Controlled for ethnicity,
socioeconomic status and
additional factors
Measured as part of the
study
Independent of
exposure
Sufficient Subjects lost of follow-up
may have introduced bias
Fernandes, 2010 [39]
and Fernandes and
Sturm, 2011 [40]
Representative Controlled for ethnicity,
socioeconomic status and
additional factors
Structured interview Independent of
exposure
Sufficient Subjects lost to follow-up
may have introduced bias
Henry, 2006 [50] Somewhat representative Did not control for
socioeconomic status or
ethnicity
Measured as part of the
study
Independent of
exposure
Sufficient Sufficient
Hernandez. Francis
and Doyle, 2011 [41]
Representative Controlled for ethnicity,
socioeconomic status and
additional factors
Written self report Independent of
exposure
Sufficient Sufficient
Hinrichs, 2006 [47] Representative Controlled for ethnicity
and, socioeconomic
status
Measured as part of the
study
Independent of
exposure
Sufficient No statement
Millimet and Tchernis,
2009 [42]
Somewhat representative Controlled for ethnicity,
socioeconomic status and
additional factors
Measured as part of the
study
Independent of
exposure
Sufficient Subjects lost to follow-up
may have introduced bias
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 8 of 22
http://www.ijbnpa.org/content/10/1/101

Table 3 Summary of study quality (Continued)
Millimet, Tchernis and
Husain, 2008 [43]and
2010 [44]
Representative Controlled for ethnicity,
socioeconomic status and
additional factors
Measured as part of the
study
Independent of
exposure
Sufficient Sufficient
Ramirez-Lopez, et al.
2005 [54]
Somewhat representative Controlled for some
factors but not
socioeconomic status or
ethnicity
Measured as part of the
study
No description Sufficient Subjects lost to follow-up
may have introduced bias
Cross-sectional
Studies
Representativeness of
the cohort/sample
Comparability of
cohorts
Ascertainment of
exposure
Assessment of
outcome
Duration of exposure Adequacy of exposure
Fox, et al. 2009 [56] Somewhat representative Controlled for ethnicity,
socioeconomic status and
additional factors
Measured as part of the
study and structured
interviews
Independent of
exposure
Sufficient Sufficient
Jones, et al. 2003 [34] Somewhat representative Controlled for ethnicity
and, socioeconomic
status
Measured as part of the
study
Independent of
exposure
Sufficient Sufficient
Harrison, et al. 2011
[57]
Somewhat representative Controlled for
socioeconomic status
Measured as part of the
study
Independent of
exposure
Sufficient Subjects lost to follow-up
may have introduced bias
Veugelers and
Fitzgerald, 2005 [58]
Somewhat representative Controlled for
socioeconomic status and
additional factors
Measured as part of the
study
Independent of
exposure
Sufficient Sufficient
Zhu, et al. 2010 [59] At risk group Did not control for
socioeconomic status or
ethnicity
Measured as part of the
study
Independent of
exposure
Sufficient Subjects lost of follow-up
unlikely to introduce bias
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 9 of 22
http://www.ijbnpa.org/content/10/1/101

Table 4 Policy summaries and results
Study Involvement Policy components Impact†Results Sig
Stakeholder* Family BMI-SDS Ov Ob BF BMIHFZ Statistic (95% confidence interval)
unless otherwise stated
Diet policies
Foster, et al. 2008 [55]✓
a,c,e,f,h
✓School nutrition policy
initiative
--Adjusted change in BMI-SDS -0.01
(-0.08,0.06)
Adjust odds ratio overweight 0.65
(0.54,0.79)
Sig
Adjusted odds ratio obesity 1.09
(0.85,1.40)
Baxter, et al. 2009 [53] Location of School
Breakfast Program
consumption
--- - Δmean BMI% breakfast in classroom
compared to the cafeteria 2.64
(p=0.06)
Henry, 2006 [50] National School Lunch
Program
--- - Hedges’goverweight 1.39 (0.55,2.24) Sig
Hernandez, Francis and
Doyle, 2003 [41]
National School Lunch
Program
--- - Adjusted change in BMI
Kindergarten: 0.12 (-0.33,0.57)
Adjusted change in BMI 1
st
grade:
0.20 (-0.29,0.69)
Adjusted change in BMI 3
rd
grade:
0.36 (-0.25,0.97)
Adjusted change in BMI 5
th
grade:
0.52 (-0.24,1.28)
Hinrichs, 2010 [47] National School Lunch
Program
--Adjusted change in BMI ♂-0.02
(-0.06,0.02), ♀-0.02 (-0.07,0.03)
Change in prevalence of overweight
♂<-0.01 (-0.01,<0.01), ♀<-0.01
(-0.01, <0.01)
Change in prevalence of obesity ♂
<-0.01 (<-0.01, <0.01), ♀<-0.01
(<-0.01, <0.01)
Millimet, Tchernis and
Husain, 2008 [43]and
2010 [44]
Bivariate Probit results assuming
ρ=0.1
National School Lunch
Program,
---Change in probability of being
overweight 0.13 (0.07, 0.20)
Change in probability of being
obese 0.13 (0.05, 0.20)
School Breakfast Program ---Change in probability of being
overweight -0.07 (-0.14, <-0.01)
Change in probability of being
obese -0.05 (-0.13, 0.03)
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 10 of 22
http://www.ijbnpa.org/content/10/1/101

Table 4 Policy summaries and results (Continued)
Millimet and Tchernis,
2009 [42]
Bias corrected minimum bias
estimator assuming θ=0.25
School Breakfast Program --Change in BMI growth rate 3
rd
grade: -0.03 (-0.06, <-0.01)
Change in probability of overweight
3
rd
grade: -0.21 (-0.33, -0.03)
Change in probability of obesity 3
rd
grade: -0.17 (-0.26, -0.01)
Change in BMI growth rate 5
th
grade: -0.04 (-0.08, 0.01)
Change in probability of overweight
5
th
grade: -0.28 (-0.40, -0.09)
Change in probability of obesity 5
th
grade: -0.12 (-0.28, -0.04)
Ramirez-Lopez, et al. 2005
[54]
School Breakfast Program -Change in BMI Intervention: 0.1,
Control: -0.1
Change in BF% Intervention: -0.2,
Control: -0.5
Change in prevalence of overweight
or obesity Intervention: 1, Control: -1
Change prevalence of obesity
Intervention: 1, Control:-3
Fox, et al. 2009 [56] À la carte LNED food not
available
---Adjusted change in BMI-SDS -0.15
(-0.37,0.07)
Adjusted odds ratio obesity 1.09
(0.57,2.08)
Milk not available for
school lunch
---Adjusted change in BMI-SDS -0.13
(-0.33,0.07)
Adjusted odds ratio obesity 1.17
(0.75,1.82)
Fresh fruit/ raw
vegetables available
---Adjusted change in BMI-SDS 0.19
(0.01,0.37)
Adjusted odds ratio obesity 1.13
(0.73,1.75)
Fried potato products not
available
---Adjusted change in BMI-SDS 0.20
(<0.01,0.40)
Adjusted odds ratio obesity 2.70
(1.58,4.62)
Sig
Desserts offered ≤once a
week
---Adjusted change in BMI-SDS 0.08
(-0.08,0.24)
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 11 of 22
http://www.ijbnpa.org/content/10/1/101

Table 4 Policy summaries and results (Continued)
Adjusted odds ratio obesity 1.78
(1.13,2.80)
Sig
Jones, et al. 2003 [34]Adjusted odds ratio overweight and
obesity:
National School Lunch
Program
---Food secure ♂1.06 (0.53,2.08), ♀0.49
(0.22,1.10),
Food insecure ♂0.62 (0.25,1.54),
♀0.29 (0.11,0.80)
Sig♀
School Breakfast and
National School Lunch
Programs
---Food secure ♂1.33 (0.81,2.18), ♀0.66
(0.35,1.26)
Food insecure ♂0.85 (0.42,1.74),
♀0.42 (0.19,0.96)
Sig♀
Physical activity policies
Donnelly, et al.2009 [62]✓
h
Physical Activity Across
the Curriculum
--BMI Hedges’g0.01 (-0.09,0.11)
Heelan, et al. 2009 [61] Walking school bus
scheme
-- -Intervention vs. Control BMI-SDS
Hedges’g: -0.21 (-0.58,0.15)
Frequent v. passive BMI-SDS Hedges’
g: -0.49 (-0.94,-0.03)
Sig
Infrequent v. passive BMI-SDS
Hedges’g: -0.17 (-0.61,0.28)
Intervention vs. Control BF% Cohen’s
d: -0.25 (-0.61,0.11)
Frequent v. passive BF% Cohen’sd:
-0.59 (-1.05,-0.13)
Sig
Infrequent v. passive BF% Cohen’sd:
-0.28 (-0.72,0.17)
Chiodera, et al.2008 [60] Professionally led PE --- - Change in BMI grade 1: -0.21 Sig
Change in BMI grade 2: -0.05
Change in BMI grade 3: -0.06
Change in BMI grade 4: 0.04
Change in BMI grade 5: 0.02
Datar and Sturm, 2004
[38]
Increased PE duration of
1 hour per week
--- - Adjusted change in BMI, normal
weight ♂0.04 (-0.04,0.12)
Adjusted change in BMI, normal
weight ♀0.01 (-0.07,0.10)
Adjusted change in BMI, overweight
or obese ♂-0.07 (-0.19,0.05)
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 12 of 22
http://www.ijbnpa.org/content/10/1/101

Table 4 Policy summaries and results (Continued)
Adjusted change in BMI, overweight
or obese ♀-0.32 (-0.46,-0.17)
Sig
Fernandes, 2010 [39] and
Fernandes and Sturm,
2011 [40]
Meeting the National
Association for Sport and
Physical Education
(NASPE) guidelines
--- - PE duration Adjusted change in BMI
% -0.74 (-1.78,0.30), ♂-1.56 (-3.03,
-0.09), ♀0.05 (-1.40,1.50)
Sig♂
Break period duration: adjusted
change in BMI% -0.74 (-1.33,-0.15), ♂
-0.81 (-1.67,0.05), ♀-0.69 (-1.49,0.11)
Sig
Combined policies
Johnson, et al. 2012 [31]✓
e,f,g,h
✓Be Active Eat Well --- - Adjusted change in BMI-SDS -0.085
(-0.18,0.01)
HE policy --- - Adjusted change in BMI-SDS -0.008
(-0.06,0.04)
PA policy --- - Adjusted change in BMI-SDS -0.006
(-0.06,0.05)
Jordan, et al. 2008 [25,64]✓
b,c,d,e,f,h
✓Utah’s Gold Medal
Schools
--- - Change in BMI-SDS Intervention: 0.21
(-0.71,1.13), Control: 0.53 (-0.21,1.27)
Chomitz, et al. 2010 [63]✓
c,e,f,h
✓Healthy Living Cambridge
Kids
--Change in BMI-SDS -0.04 Sig
Change in prevalence of overweight
0.6% points
Change in prevalence of obesity
-2.2% points
Sig
Harrison, et al. 2011 [57] Cookery lessons --- -None of the policies were
significantly associated with FMI in
females, while only being able to eat
any food at break times and being
able to play 3-4 games during break
times where association with higher
FMI in males.
Foods permitted during
break periods
--- -
HE policy --- -
Sports allowed during
break periods
--- -
‘Park and stride’scheme --- -
PA policy --- -
PA and HE policy --- -
Veugelers and Fitzgerald,
2005 [26,58]
Nutrition policy ---Adjusted odds ratio overweight: 0.91
(0.77,1.09)
Adjusted odds ratio obesity: 0.85
(0.63,1.55)
✓
a,c,d,e,f,g,h
✓Annapolis Valley Health
Promoting Schools
Project
---Adjusted odds ratio overweight: 0.41
(0.32,0.53)
Sig
Adjusted odds ratio obesity: 0.28
(0.14,0.57)
Sig
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 13 of 22
http://www.ijbnpa.org/content/10/1/101

Table 4 Policy summaries and results (Continued)
Zhu, et al. 2010 [59]✓
a,b,d,f,g,h
Professionally led PE ---- Adjusted change in BMIHFZ
achievement rate 0.62 (0.01,1.23)
Sig
Duration of PE periods ---- Adjusted change in BMIHFZ
achievement rate 0.05 (-0.03,0.13)
Number of PE periods ---- Adjusted change in BMIHFZ
achievement rate 1.06 (0.47,1.65)
Sig
Duration of Break periods ---- Adjusted change in BMIHFZ
achievement rate 2.71 (1.75,3.67)
Sig
Number of break periods ---- Adjusted change in BMIHFZ
achievement rate -2.25 (-3.86,-0.64)
Sig
Cancel due to weather ---- Adjusted change in BMIHFZ
achievement rate -1.26 (-3.73,1.21)
PE exemptions ---- Adjusted change in BMIHFZ
achievement rate -0.34 (-0.65,-0.03)
Sig
USDA ---- Adjusted change in BMIHFZ
achievement rate 0.02 (-1.49,1.53)
Wellness council ---- Adjusted change in BMIHFZ
achievement rate 0.41 (-0.04,0.86)
Abbreviations:♂male, ♀female, BF% body fat percentage, BMI body mass index, BMI% BMI percentile, BMIHFZ BMI Healthy Fitness Zone [45], BMI-SDS BMI standard deviation score, FSM free or reduced school meals,
HE healthy eating, LNED low-nutrient, energy-dense, PA physical activity, PE physical education, SD standard deviation, SE standard error, Sig p < 0.05, TV television, USDA United States Department of Agriculture
wellness program.
*Stakeholders: school administrators
a
, school board
b
, sports coaches
c
, food services
d
, health services
e
, parents
f
, pupils
g
, teachers
h
.
†Impact: a symbolic representation of the statistical results. ↑: positive association, ↓: negative association, ↕: mixed association, ↔: no association. Black arrows indicate significance (p < 0.05), while grey indicates non-
significance (p > 0.05) [51,52].
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 14 of 22
http://www.ijbnpa.org/content/10/1/101

Due to the nature of policy interventions, the ran-
domised controlled trials and controlled before and after
studies could not meet some of the quality criteria gen-
erally applied to these study designs. Blinding of the out-
come assessment may not always have been possible
and, for some studies, loss to follow up was greater than
20%. However, each study employed a valid design, in
terms of the use of second sites as controls, random al-
location and protection from contamination.
Participant characteristics
The demographics and baseline weight status of the par-
ticipants of each study are listed in Table 2. All the stud-
ies assessed both males and females and in those studies
that reported gender distribution there was approxi-
mately equal numbers of each sex. Thirteen of the stud-
ies examined children across the age span of primary
education, while four studies examined children towards
theendofprimaryeducation,onestudyexamined
children in the middle of primary education and three
studied those beginning primary education. Five of the
studies did not report ethnicity data. Of the sixteen
studies which did, eleven had a sample which was ma-
jority white, two studies had a majority black, two stud-
ies without a majority ethnic group had black as the
largest minority and one study only reported that the
majority of participants' parents were natives. Fifteen of
the studies reported the socioeconomic status of the par-
ticipants using a variety of measures which are reported in
Table 2 .
Fifteen studies utilised the Center for Disease Control
and Prevention (CDC) 2000 BMI reference categories
and an additional paper appears to have used this
categorisation but did not report it [48]. Two studies
utilised the International Obesity Task Force (IOTF)
BMI reference categories [49]. Of the three remaining
studies one categorised BMI according to the Healthy
Fitness Zone [45], one studied adults who had been ex-
posed to the NSLP as children and the remaining paper
did not report which reference was utilised. The major-
ity of studies reported that at baseline between 20% and
40% of the sample was overweight or obese, however,
Henry [50] reported a prevalence of overweight and
obesity below 10%. Three studies did not report the
baseline weight characteristics of the participants.
Study results
Key results from each study are presented in Table 4.
Alongside the quantitative results, the results have also
been depicted symbolically to aid understanding in a
similar way to that described by McCartney, et al. [51]
and Thomson [52]. Table 4 contains columns for each of
the outcomes assessed and within each column is a
symbol. If the symbol is a dash (−) that outcome was
not assessed by the study, otherwise the direction of the
arrow indicates the direction of the association (↑; posi-
tive, ↓; negative, ↕; mixed, ↔; no effect), black arrows in-
dicate significant (p < 0.05) results, while grey arrows are
non-significant.
Diet related policies
Thirteen studies evaluated diet related policies, including
five evaluating the NSLP [34,41,43,44,47,50] and five
School Breakfast Program (SBP) [34,42-44,53,54] (two
studies evaluated both the NSLP and SBP [34,43,44]).
The NSLP and SBP were developed and implemented to
improve the nutritional state and health of undernour-
ished children [41,47]. Subsequently, unlike the general
diet related policies, the NSLP and SBP are targeted at
specific pupils, their intention not being to benefit the
entire population of children, just an at risk group. With
improved nutrition it has become a concern that the
NSLP and SBP could be contributing to unhealthy weight
gain [41]. The remaining five studies evaluated policies re-
lated to the availability of foods within schools, one of
which did not present quantitative results and therefore
only four studies could be pooled [31,55-58]. Due to the
underlying conceptual differences in the intention and
population of the NSLP, SBP and other diet related pol-
icies, the studies were separated into three policy groups
for meta-analysis.
The pooled result of participation in the NSLP was a
small non-significant rise in BMI-SDS (0.038 BMI-SDS,
95% confidence interval (95% CI) -0.193 to 0.269) (Figure 2).
The study by Hinrichs [47] which could not be included
in the meta-analysis resulted in a similar non-significant
difference in BMI, overweight or obesity status between
adults who had and hadn’t participated in the NSLP.
The pooled result of the five studies that evaluated the
SBP was a significantly lower BMI-SDS among those who
participated in the SBP (−0.080 BMI-SDS, 95% CI −0.143
to −0.017) (Figure 3). However, it should be noted that
there was a significant degree of heterogeneity in both of
these clusters (Figures 2 and 3).
The other diet related policies evaluated included: re-
moving low nutrient, energy-dense foods, fried potato
products, desserts and whole or 2% milk from cafeterias,
ensuring fruits and vegetables are available in the cafe-
teria, children being prevented from eating any food at
break periods and attending a school with a nutrition
policy which enabled children to choose healthier foods
[31,56-58]. The pooled effect of these diet related pol-
icies was a small and non-significant reduction of −0.021
BMI-SDS (95% CI −0.066 to 0.023) (Figure 4).
Physical activity related policies
Eight studies examined physical activity related policies,
two of which could not be included in the meta-analysis
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 15 of 22
http://www.ijbnpa.org/content/10/1/101

as they did not provide sufficient information for the
calculation of effect sizes [31,38-40,57,59-62]. The pol-
icies evaluated included: having a general physical activ-
ity policy, the use of qualified PE teachers, PE and break
period duration and frequency, variety of activities
permitted during break periods, the number of valid
reasons for exemption from PE, cancelling PE due to
the weather, physical activity incorporated into lessons
and active commuting schemes [31,38-40,57,59-62].
The pooled effect of all policies related to physical activ-
ity was a small and non-significant reduction in BMI-
SDS (−0.011, 95% CI −0.036 to 0.013) (Figure 5).
Combined policies
Six studies evaluated policies with both diet and phys-
ical activity related components, one of which did not
reportthequantitativeresultsjustthedirectionand
significance and therefore effect sizes could not be cal-
culated [57-59,63,64]. Four of these studies also consid-
ered policy components which related to either diet or
physical activity separately and these components have
been included in the respective clusters discussed above
[31,57-59]. Five of the policies were components of
multifaceted intervention programmes, subsequently
there was great variability in the nature of the combined
policies leading to high heterogeneity (I
2
= 85.1%) and
the effects of these policies have not been combined.
However, the individual results and effect sizes are de-
tailed in Table 4 and Figure 6. Harrison, et al. [57]
reported a non-significant association between having
policies promoting both physical activity and healthy
eating and FMI. Similarly, the USDA wellness program
or wellness council was not significantly associated with
an improvement in the proportion of children within
NOTE: Weights are from random effects analysis
Overall (I−squared = 80.8%, p = 0.001)
Hernandez. Francis and Doyle, 2011 [41]*
Henry, 2006 [50]
Millimet, Tchernis and Husain, 2008 [43] and 2010 [44]*
Study
Jones, et al. 2003 [34]
ID
0.04 (−0.19, 0.27)
0.02 (−0.15, 0.18)
1.39 (0.55, 2.24)
0.05 (0.01, 0.09)
−0.32 (−0.63, −0.02)
ES (95% CI)
100.00
32.41
6.27
38.29
%
Weight
0.04 (−0.19, 0.27)
0.02 (−0.15, 0.18)
1.39 (0.55, 2.24)
0.05 (0.01, 0.09)
−0.32 (−0.63, −0.02)
ES (95% CI)
100.00
32.41
6.27
38.29
%
23.02
Weight
Favours polic
y
Favours no polic
y
0−1 −.5 0 .5 1 1.5
Figure 2 Forest plot showing body mass index standard deviation score effect sizes (Hedges’g) of studies evaluating participation in
the National School Lunch Program. *Study using the Early Childhood Longitudinal Study –Kindergarten (ECLS-K) cohort.
NOTE: Weights are from random effects analysis
Overall (I−squared = 71.3%, p = 0.007)
Millimet and Tchernis, 2009 [42]*
Jones, et al. 2003 [34]
Millimet, Tchernis and Husain, 2008 [43] and 2010 [44]*
Baxter, et al. 2009 [53]
Ramirez−Lopez, et al. 2005 [54]
ID
Study
−0.08 (−0.14, −0.02)
−0.12 (−0.14, −0.09)
−0.16 (−0.40, 0.08)
−0.03 (−0.07, 0.01)
−0.11 (−0.23, 0.00)
0.04 (−0.22, 0.29)
ES (95% CI)
100.00
37.94
5.78
34.13
16.84
5.31
Weight
%
−−
−−−
−−
−−
−−
−
Favours polic
y
Favours no polic
y
0−1 −.5 0 .5 1 1.5
Figure 3 Forest plot showing body mass index standard deviation score effect sizes (Hedges’g) of studies evaluating participation in
the School Breakfast Program. *Study using the Early Childhood Longitudinal Study –Kindergarten (ECLS-K) cohort.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 16 of 22
http://www.ijbnpa.org/content/10/1/101

the BMIHFZ in the study by Zhu, et al. [59]. Whereas,
exposure to Healthy Living Cambridge Kids was signifi-
cantly associated with lower BMI-SDS and prevalence of
obesity, but not the prevalence of overweight which
resulted in a non-significant combined effect size [63].
Chomitz, et al. [63] considered this result to be expected
as obese children become overweight before reaching a
healthy weight and therefore the prevalence of overweight
might not change significantly. Participants in Gold Medal
Schools programme in Utah, USA and Be Active Eat Well
in Victoria, Australia gained less weight than control
participants (Figure 6) [31,64]. The Annapolis Valley
Health Promoting Schools Program (AVHPSP), evalu-
ated by Veugelers and Fitzgerald [58], was significantly
associated with reduced odds of both overweight and
obesity. All four of these policies included significant
stakeholder involvement within the development and
implementation of the policy and engaged families.
Discussion
The aim of this systematic review was to examine the ef-
fect of school diet and physical activity related policies
upon anthropometric outcomes among children aged 4–
11 years. Twenty-one studies were identified which
examined a range of policies which were clustered as
either diet related or physical activity related or both
(combined policies) for analysis. Within the diet related
policies cluster, eight studies evaluated the NSLP and
SBP and as these policies target a subset of the popula-
tion they were analysed separately from the other diet
NOTE: Weights are from random effects analysis
Overall (I−squared = 0.0%, p = 0.954)
Johnson, et al. 2012 [31]
Fox, et al. 2009 [56]
Veugelers and Fitzgerald, 2005 [58]
Foster, et al. 2008 [55]
ID
Study
−0.02 (−0.07, 0.02)
−0.04 (−0.13, 0.05)
−0.03 (−0.13, 0.07)
−0.03 (−0.14, 0.08)
−0.01 (−0.08, 0.07)
ES (95% CI)
100.00
25.41
19.89
16.93
37.77
Weight
%
−−
−−
−−
−−
−−
%
Favours polic
y
Favours no polic
y
0−1 −.5 0 .5 1 1.5
Figure 4 Forest plot showing body mass index standard deviation score effect sizes (Hedges’g) of studies evaluating other diet
related policies.
NOTE: Weights are from random effects analysis
Overall (I−squared = 0.0%, p = 0.977)
Chiodera, et al. 2008 [60]
ID
Study
Zhu, et al. 2010 [59]
Heelan, et al. 2009 [61]
Fernandes, 2010 [39] and Fernandes and Sturm, 2011 [40]*
Donnelly, et al. 2009 [62]
Johnson, et al. 2012 [31]
−0.01 (−0.04, 0.01)
−0.01 (−0.04, 0.02)
ES (95% CI)
−0.01 (−0.27, 0.24)
−0.07 (−0.32, 0.18)
−0.01 (−0.06, 0.04)
0.01 (−0.09, 0.11)
−0.04 (−0.14, 0.06)
100.00
62.18
Weight
%
0.95
0.99
23.66
5.92
6.30
−−
−−
−−
−−
−−
−
−−
Favours polic
y
Favours no polic
y
0−1 −.5 0.5 1 1.5
Figure 5 Forest plot showing body mass index standard deviation score effect sizes (Hedges’g) from studies evaluating physical
activity related policies. *Study using the Early Childhood Longitudinal Study –Kindergarten (ECLS-K) cohort.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 17 of 22
http://www.ijbnpa.org/content/10/1/101

related policies. The NSLP was associated with a non-
significant rise in BMI-SDS results, whereas the SBP was
associated with a significant decrease in BMI-SDS and
the other diet related policies were associated with a
non-significant decrease in BMI-SDS, however, signifi-
cant heterogeneity remained in the NSLP and SBP sub-
clusters reducing the validity of these results (Figures 2,
3 and 4). Physical activity related policies were not asso-
ciated with significant changes in BMI-SDS (Figure 5).
Among the combined policies there was significant het-
erogeneity preventing meta-analysis, yet the combined
policies demonstrated promising results in particular
Gold Medal Schools, AVHPSP and Be Active Eat Well
(Figure 6) [31,58,64]. These were multifaceted interven-
tion programmes, which had wider health promotion
aims, as well as improving diet and increasing physical
activity. As well as utilising policy these programmes in-
cluded stakeholder involvement and family engagement,
methods recommended by Khambalia, et al. [16] as im-
portant components in school based obesity prevention
interventions [25,26,31]. Gold Medal Schools also in-
cluded health surveys and promotion among the school
staff [25]. Five of the studies also evaluated the effect of
the policy upon prevalence of underweight and none of
the policies were found to have a negative impact, with
some reporting a reduced prevalence of underweight
among those exposed to the policy [34,42,47,55,63].
A strength of this review was the broad search strat-
egy. School policy evaluation may be reported by a
variety of disciplines and inside and outside of peer-
reviewed journals and therefore through the variety of
databases searched, the grey literature search and the
inclusion of literature such as dissertations all the rele-
vant studies were sought. Primarily, this demonstrated
that there is a paucity of scientific evaluations of school
policies as only 21 eligible studies were identified from
the 6,894 retrieved, yet among the eligible studies where
were a variety of designs, quality and policies which im-
pingeuponthereview.Amongthe21studiesreviewed
only five utilised experimental study designs which
prevented the consideration of causal pathways in this
review (Table 3). Loss to follow-up which may have led
to bias was a significant concern for a number of the in-
cluded studies as overweight or obese children may have
been more likely to avoid follow-up (Table 3). As well as
these differences in terms of quality and design, even
when only the studies which evaluated similar policies
were pooled for analysis there still remained significant
heterogeneity. The length of follow-up/exposure within
the included studies ranged from 8 months to more
than 9 years (Table 2). The results of those studies with
shorter follow-up/exposure duration may reflect the
novelty of the policy or that insufficient time had passed
for changes in body mass to be observed. The results from
studies with longer follow-up/exposure reflect whether the
policy prompted maintained behaviour change, or had only
produced short lived changes in behaviour, which might
also have contributed to the heterogeneity. Heterogeneity
in the combined policies cluster was expected as there
were differences in the policy each study evaluated, but the
high heterogeneity in the NSLP and SBP clusters is unex-
pected and may be due to the differences in the sample
characteristics or analytical methods (Table 2). Henry [50]
reported a low baseline prevalence of overweight and obes-
ity and produced an unusually large effect size, however,
removing this result from the NSLP cluster only reduced
the heterogeneity to I
2
= 65.6%. Millimet and Tchernis [42]
and Millimet, Tchernis and Husain [43] used complex ana-
lytical methods to account for non-random selection into
the NSLP and SBP which may have produced greater
differences between the studies. It is therefore more appro-
priate to understand the results of the meta-analyses
presented as averages of the individual study effects rather
than estimates of the common policy effect [65].
Chomitz, et al. 2010 [63]
Johnson, et al. 2012 [31]
Jordan, et al. 2008 [64]
Veugelers and Fitzgerald, 2005 [58]
Zhu, et al. 2010 [59]
Study
Healthy Living Cambridge Kids
Be Active Eat Well
Gold Medal Schools Utah
Annapolic Valley Health Promoting Schools Project
USDA wellness program or wellness council
title
Policy
−0.01 (−0.08, 0.06)
−0.09 (−0.18, −0.00)
−0.75 (−1.03, −0.46)
−0.25 (−0.47, −0.02)
−0.00 (−0.47, 0.47)
ES (95% CI)
−−
−−−
−−−
−−−
−−
Favours policy Favours no policy
0−1 −.5 0 .5 1 1.5
Figure 6 Forest plot showing body mass index standard deviation score effect sizes (Hedges’g) from studies evaluating the
combined policies.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 18 of 22
http://www.ijbnpa.org/content/10/1/101

In order to calculate effect sizes, assumptions about the
outcome correlations in studies using independent and
non-independent samples were made; these assumptions
were relaxed in a sensitivity analysis, reported in Additional
file 2. However, there were no significant changes in the
results. Combining continuous and categorical BMI-SDS
outcomes also require some discussion. Foster, et al. [55]
and Chomitz, et al. [63] both found the effect of the policy
they evaluated to be inconsistent across weight categories
which resulted in non-significant effect sizes. However,
they found conflicting differences, Foster, et al. [55] found
a significant effect in the overweight but not obese while
Chomitz, et al. [63] found the opposite. Foster, et al. [55]
argue that obesity is more intractable than overweight,
supporting the need for early intervention to improve
the weight status of overweight pupils before they be-
come obese. While Chomitz, et al. [63] argue that the
number of obese pupils becoming overweight may equal
the number of overweight pupils obtaining a healthy
weight resulting in no significant change in the prevalence
of overweight. Rappaport, Daskalakis and Sendecki [66]
recently re-evaluated the School Nutrition Policy Initiative
evaluated by Foster, et al. [55] using routinely collected
data and found the policy to no longer have an effect on
either overweight or obesity. Repeating the meta-analysis
replacing the results of Foster, et al. [55] with those of
Rappaport, Daskalakis and Sendacki [66] did not signifi-
cantly alter the results (Additional file 4). Ideally, policies
would result in lowering the prevalence of both over-
weight and obesity which is likely to result in reduced
mean BMI-SDS, suggesting that combining the results
was appropriate [31,67].
This review evaluated the effect of school policies
upon an objective measure of weight status (BMI-SDS)
unlike previous reviews which have evaluated physical
activity and diet outcomes, which may be more subject-
ive [11-16]. Therefore, the positive effects of school
policies upon diet identified by Jaime and Lock [11] and
Van Cauwenberghe [14], were not found to extend to
improved weight status in this review most likely due to
the difficulties in accurately assessing diet. Nutrition
guidelines formed a component in each of the combined
policies which may indicate that diet related policies are
beneficial when used in combination with physical activ-
ity policies. More evidence was found to support the
introduction of physical activity policies to affect weight
status with some evidence found to support the im-
provement of the quality and variety of PE identified by
Lagarde and LeBlanc [15] which were also components
in the AVHPSP [58], Be Active Eat Well [27,31], Gold
Medal Schools [64] and Healthy Living Cambridge Kids
[63]. However, results relating to professionally led PE
and the duration and frequency of PE and break periods
were mixed. Although there was a lack of significant
findings for diet and physical activity policies by them-
selves (Figures 4 and 5) the overall result of the Be
Active Eat Well programme (which encouraged the de-
velopment of healthy eating and physical activity pol-
icies) was a significant reduction in BMI-SDS (Figure 6).
This suggests that the process of policy development,
engagement and broader activities may be more import-
ant than the presence or absence of a policy, supporting
the need for policies to be implemented as part of a
multifaceted intervention programme. The overall effect
of each of the included multifaceted intervention pro-
grammes, was less than one BMI-SDS which is only
equivalent to a change in weight of around 2.0 kg in Re-
ception or 6.4 kg in Year 6 aged children. Rose and Day
[67] have demonstrated that small changes in population
mean values like those observed produce significant re-
ductions in the prevalence of conditions like overweight
and obesity. More recently, Kolsgaard, et al. [68] found
significant physiological improvements (lower insulin
and cholesterol) among obese children and adolescents
from very small changes in BMI-SDS (<0.1). There has
been discussion regarding shifting the focus from weight
loss to improving health and fitness which may not
require or result in weight loss through initiatives like
Health At Every Size (HAES) as it is possible to be fit
and fat [69]. Subsequently, further discussion is required
upon what constitutes an important or clinically signifi-
cant effect of obesity prevention or health promotion
interventions.
Conclusion
The evidence from this systematic review suggests that
diet and physical activity related policies need to be lo-
cated within more complex approaches to preventing
childhood obesity which focus on multiple factors (e.g.
diet, physical activity, sedentary behaviour, self-esteem)
and at multiple levels of influence (e.g. home, school,
neighbourhood) as advocated by the Centers for Disease
Control and Prevention guidelines [10]. No policies which
guided choice through disincentives, or eliminated choice
were identified during the review, which may be pertinent
as these policy actions have been effectively employed in
campaigns to reduce the prevalence of smoking [7]. Al-
though there are calls for similar policy actions to prevent
further increases in the prevalence of obesity, the policy
would need to extend outside of schools [6].
The complex web of factors which influence weight
have been illustrated in the obesity systems map which
also highlights the range of levels of influence from mi-
cro to macro [1]. Within this systematic review insuffi-
cient evidence was found to make recommendations
upon the use of policies which aim to influence only one
factor related to weight status (diet or physical activity)
and at one level of influence (school). However, these
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 19 of 22
http://www.ijbnpa.org/content/10/1/101

results suggest that policies need to be located within
wider health promotion intervention programmes in
order to have an effect [10]. Further research is going to
be crucial to the development and commissioning of
evidence based policy and therefore, policy makers and
researchers should work in partnership to consider the
evaluation of new policies prior to implementation.
Although there are difficulties in implementing new
policies experimentally, such as blinding of outcome
assessment and loss to follow-up, making use of the nat-
ural variation in uptake of policies to research the effects
on weight status, so-called natural experiments (e.g. con-
trolled before and after studies, interrupted time series
studies) could be used to evaluate new policies [70,71].
The difficulties encountered in this review highlight the
need for future studies to be comprehensively reported
and have a duration of years rather than months, in order
to inform future systematic reviews and meta-analyses.
Additional files
Additional file 1: Search strategy.
Additional file 2: Sensitivity analysis.
Additional file 3: Effect size calculations.
Additional file 4: Diet related policies meta-analysis with
Rappaport, Daskalakis and Sendacki [66] replacing Foster, et al.
[55].
Abbreviations
95% CI: 95% confidence interval; AVHPSP: Annapolis valley health promoting
schools program; BMI: Body mass index; BMI%: Body mass index percentile;
BMIHFZ: Body mass index healthy fitness zone; BMI-SDS: Body mass index
standard deviation score; CDC: Centers for disease control and prevention;
ECLS-K: Early childhood longitudinal study –kindergarten cohort; FMI: Fat
mass index; HAES: Health at every size; ICC: Intra-cluster correlation;
IOTF: International obesity task force; MeSH: Medical subject headings;
NSLP: National school lunch program; PE: Physical education; SBP: School
breakfast program; UK: United Kingdom; USA: United States of America;
USDA: United States Department of Agriculture.
Competing interest
The authors declare that they have no competing interests.
Authors’contributions
AJW was involved with the conception and design of the review, undertook
the searches and participated in the study identification, data extraction and
quality assessment, he then undertook the analysis and drafted the
manuscript. WEH was involved with the conception and design of the
review, advised on and supervised the analysis and assisted with drafting the
manuscript. CAW was involved with the conception of the study and had
input into the final manuscript. AJH participated in the study identification,
data extraction and quality assessment and proofread the final manuscript.
SL contributed to the drafting of the final manuscript and interpretation of
the results. KMW was involved with the conception and design of the
review, participated in the data extraction and quality assessment, assisted
with the interpretation of results and drafting of the final manuscript. All
authors read and approved the final manuscript.
Acknowledgements
The authors would like to acknowledge the contributions of Mary Reece,
Kate Boddy and the Systematic Review group at the Peninsula College of
Medicine and Dentistry. The authors would like to thank the anonymous
reviewers for the constructive comments in improving the manuscript. AJW
is funded by a Medical Research Council Doctoral Training Grant and Sport
and Health Sciences, University of Exeter. SL, KMW, WEH and AJH are
partially supported by the National Institute for Health Research (NIHR)
Collaboration for Leadership in Applied Health Research and Care (CLAHRC)
for the South West Peninsula. The views expressed in this publication are
those of the author(s) and not necessarily those of the NHS, the NIHR or the
Department of Health in England.
Author details
1
Institute of Health Services Research, University of Exeter Medical School
(formerly Peninsula College of Medicine and Dentistry), Veysey Building,
Salmon Pool Lane, EX2 4SG, Exeter, Devon, UK.
2
Children’s Health and
Exercise Research Centre, Sport and Health Sciences, College of Life and
Environmental Sciences, University of Exeter, St. Luke’s Campus, Heavitree
Road, EX1 2LU, Exeter, Devon, UK.
Received: 13 March 2013 Accepted: 16 August 2013
Published: 22 August 2013
References
1. Butland B, Jebb S, Kopelman P, McPherson K, Thomas S, Mardell J, Parry V:
Tackling obesities: future choices –project report. 2nd edition. London:
Department of Innovation, Universities and Skills; 2007.
2. Reilly JJ, Methven E, McDowell ZC, Hacking B, Alexander D, Stewart L, Kelnar
CJ: Health consequences of obesity. Arch Dis Child 2003, 88:748–752.
3. Procter KL: The aetiology of childhood obesity: a review. Nutr Res Rev
2007, 20:29–45.
4. Ridler C, Townsend N, Dinsdale H, Mulhall C, Rutter H, Ridler C, Townsend
N, Dinsdale H, Mulhall C, Rutter H: National Child Measurement Programme:
detailed analysis of the 2007/08 National dataset. Oxford: National Obesity
Observatory; 2009. [http://www.noo.org.uk/uploads/
doc168_2_noo_NCMPreport1_110509.pdf] (Accessed 13 June 2013).
5. Strauss RS, Pollack HA: Epidemic increase in childhood overweight, 1986–
1998. JAMA 2001, 286:2845–2848.
6. Gortmaker SL, Swinburn BA, Levy D, Carter R, Mabry PL, Finegood DT,
Huang T, Marsh T, Moodie ML: Changing the future of obesity: science,
policy, and action. Lancet 2011, 378:838–847.
7. Nuffield Council on Bioethics: Public health: ethical issues. London: Nuffield
Council on Bioethics; 2007. [http://www.nuffieldbioethics.org/sites/default/files/
Public%20health%20-%20ethical%20issues.pdf] (Accessed 13 June 2013).
8. Office for Standards in Education: The school sports partnership programme:
evaluation of phases 3 and 4 2003/04. London: Office for Standards in
Education; 2004. [http://www.ofsted.gov.uk/resources/school-sports-
partnerships-programme-evaluation-of-phases-3-and-4-200304] (Accessed
13 June 2013).
9. Department of Health: National Healthy School status: a guide for schools.
London: Department of Health; 2005. [http://www.salisbury.anglican.org/
resources-library/schools/schools-every-child-matters/be-healthy] (Accessed
13 June 2013).
10. Centers for Disease Control and Prevention: School health guidelines to
promote healthy eating and physical activity. MMWR Recomm Rep 2011,
60:1–76.
11. Jaime PC, Lock K: Do school based food and nutrition policies improve
diet and reduce obesity? Prev Med 2009, 48:45–53.
12. Sharma M: School-based interventions for childhood and adolescent
obesity. Obes Rev 2006, 7:261–269.
13. Sharma M: International school-based interventions for preventing
obesity in children. Obes Rev 2007, 8:155–167.
14. Van Cauwenberghe E, Maes L, Spittaels H, van Lenthe FJ, Brug J, Oppert J-M,
De Bourdeaudhuij I: Effectiveness of school-based interventions in Europe
to promote healthy nutrition in children and adolescents: systematic
review of published and 'grey' literature. Br J Nutr 2010, 103:781–797.
15. Lagarde F, LeBlanc C: Policy options to support physical activity in
schools. Can J Public Health 2010, 101(Suppl 2):S9–S13.
16. Khambalia AZ, Dickinson S, Hardy LL, Gill T, Baur LA: A synthesis of existing
systematic reviews and meta-analyses of school-based behavioural
interventions for controlling and preventing obesity. Obes Rev 2012,
13:214–233.
17. Higgins JPT, Green S (Eds): Cochrane handbook for systematic reviews of
interventions. Version 5.1.0 [updated March 2011]. The Cochrane
Collaboration; 2011. Available from www.cochrane-handbook.org.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 20 of 22
http://www.ijbnpa.org/content/10/1/101

18. NHS Centre for Reviews and Dissemination: Systematic reviews: CRD’s
guidance for undertaking reviews in health care. 3rd edition. York: Centre for
Reviews and Dissemination, University of York; 2009.
19. Publications and research. [http://www.rwjf.org/pr/]
20. Milio N: Glossary: healthy public policy. J Epidemiol Community Health
2001, 55:622–623.
21. National Institute for Health and Clinical Excellence: Obesity: guidance on the
prevention, identification, assessment and management of overweight and
obesity in adults and children. London: National Institute for Health and
Clinical Excellence; 2006. [http://guidance.nice.org.uk/nicemedia/live/11000/
30365/30365.pdf] (Accessed 13 June 2013).
22. The Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised
studies in meta-analyses. [http://www.ohri.ca/programs/
clinical_epidemiology/oxford.asp]
23. Cochrane Effective Practice and Organisation of Care Review Group: Data
collection checklist. Ontario: University of Ottawa; 2002. [http://epoc.
cochrane.org/epoc-resources] (Accessed 13 June 2013).
24. Cochrane Effective Practice and Organisation of Care Review Group: Data
abstraction form. Ontario: University of Ottawa; 2002. [http://epoc.cochrane.
org/epoc-resources] (Accessed 13 June 2013).
25. The Utah Department of Health: Gold Medal Schools: 2011–12 GMS guide.
Salt Lake City: The Utal Department of Health; n.d. [http://health.utah.gov/
obesity/gms/guide/Guide.pdf] (Accessed 13 June 2013).
26. Annapolis Valley Health Promoting Schools: Making the healthy choice the easy
choice. Nova Scotia: Annapolis Valley Regional School Board; n.d. [http://www.
avrsb.ca/sites/default/files/forms/avhpsp.pdf] (Accessed 13 June 2013).
27. Sanigorski AM, Bell AC, Kremer PJ, Cuttler R, Swinburn BA: Reducing
unhealthy weight gain in children through community capacity-building:
results of a quasi-experimental intervention program, Be Active Eat Well.
Int J Obes (Lond) 2008, 32:1060–1067.
28. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR: Introduction to meta-
analysis. Oxford: Wiley; 2009.
29. R Development Core Team: R: A language and environment for statistical
computing. Vienna, Austria; 2011. URL http://www.R-project.org/. ISBN 3-
900051-07-0.
30. Del Re AC, Hoyt WT: MAd: Meta-Analysis with Mean Differences; 2010. URL
http://CRAN.R-project.org/package=MAd.
31. Johnson BA, Kremer PJ, Swinburn BA, de Silva-Sanigorski AM: Multilevel
analysis of the Be Active Eat Well intervention: environmental and
behavioural influences on reductions in child obesity risk. Int J Obes
(Lond) 2012, 36:901–907.
32. StataCorp: Stata statistical software: release 11. College Station, TX: StataCorp
LP; 2009.
33. Moher D, Liberati A, Tetzlaff J, Altman DG: Preferred reporting items for
systematic reviews and meta-analyses: the PRISMA statement. J Clin
Epidemiol 2009, 62:1006–1012.
34. Jones SJ, Jahns L, Laraia BA, Haughton B: Lower risk of overweight in
school-aged food insecure girls who participate in food assistance:
results from the panel study of income dynamics child development
supplement. Arch Pediatr Adolesc Med 2003, 157:780–784.
35. Evaluation of California's SB 19 Pupil Nutrition Act. [http://clinicaltrials.gov/
show/NCT00067847]
36. Evaluation of Eat Well Be Active (EWBA) community programs. http://apps.
who.int/trialsearch/Trial.aspx?TrialID=ACTRN12607000414415]
37. Fun ’n healthy in Moreland! A 5-year school-community-based health
promotion and obesity prevention study for primary school children. fun ‘n
healthy in Moreland!. [http://apps.who.int/trialsearch/Trial.aspx?
TrialID=ACTRN12607000385448]
38. Datar A, Sturm R: Physical education in elementary school and body
mass index: evidence from the early childhood longitudinal study. Am J
Public Health 2004, 94:1501–1506.
39. Fernandes MM: Evaluating the impacts of school nutrition and physical
activity policies on child health. RAND Graduate School: PRGS Dissertation.
Doctoral; 2010.
40. Fernandes MM, Sturm R: The role of school physical activity programs in
child body mass trajectory. J Phys Act Health 2011, 8:174–181.
41. Hernandez DC, Francis LA, Doyle EA: National school lunch program
participation and sex differences in body mass index trajectories of children
from low-income families. Arch Pediatr Adolesc Med 2011, 165:346–353.
42. Millimet DL, Tchernis R: Estimation of treatment effects without an exclusion
restriction: with an application to the analysis of the School Breakfast Program.
NBER Working Paper No. 15539. Cambridge, MA: National Bureau of
Economic Research; 2009.
43. Millimet DL, Tchernis R, Husain M: School nutrition programs and the
incidence of childhood obesity. NBER Working Paper No. 14297. Cambridge,
MA: National Bureau of Economic Research; 2008.
44. Millimet DL, Tchernis R, Husain M: School nutrition programs and the
incidence of childhood obesity. J Hum Resour 2010, 45:640–654.
45. Going SB, Lohman TG, Falls HB: Body composition assessment. Dallas. TX: The
Cooper Institute; 2008. [http://www.cooperinstitute.org/pub/file.cfm?
item_type=xm_file&id=662] (Accessed 13 June 2013).
46. Chinn S: A simple method for converting an odds ratio to effect size for
use in meta-analysis. Stat Med 2000, 19:3127–3131.
47. Hinrichs P: The effects of the National School Lunch Program on
education and health. J Policy Anal Manage 2010, 29:479–505.
48. Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, Mei Z, Wei
R, Curtin LR, Roche AF, Johnson CL: 2000 CDC growth charts for the United
States: methods and development. Vital Health Stat 2002, 11:1–190.
49. Cole TJ, Bellizzi MC, Flegal KM, Dietz WH: Establishing a standard definition
for child overweight and obesity worldwide: international survey.
BMJ 2000, 320:1240–1243.
50. Henry LL: The influence of elementary-school programs on childhood
obesity. Dissertation. Doctoral George Mason University; 2006.
51. McCartney G, Thomas S, Thomson H, Scott J, Hamilton V, Hanlon P,
Morrison DS, Bond L: The health and socioeconomic impacts of major
multi-sport events: systematic review (1978–2008). BMJ 2010, 340:c2369.
52. Thomson H: Housing improvements and their health effects. In
Environmental burden of disease associated with inadequate housing. Edited
by Braubach M, Jacobs DE, Ormandy D. Copenhagen: WHO Regional office
for Europe; 2011:179–195. [http://www.euro.who.int/__data/assets/pdf_file/
0003/142077/e95004.pdf] (Accessed 26 February 2013).
53. Baxter SD, Royer JA, Hardin JW, Guinn CH, Mackelprang AJ: Daily
participation in school meals and 4th-grade children's age/sex body
mass index percentile. FASEB J 2009, 23:735.15.
54. Ramirez-Lopez E, Grijalva-Haro MI, Valencia ME, Ponce JA, Artalejo E: Effect of a
School Breakfast Program on the prevalence of obesity and cardiovascular
risk factors in children [Spanish]. Salud Publica Mex 2005, 47:126–133.
55. Foster GD, Sherman S, Borradaile KE, Grundy KM, Vander Veur SS, Nachmani
J, Karpyn A, Kumanyika S, Shults J: A policy-based school intervention to
prevent overweight and obesity. Pediatrics 2008, 121:e794–e802.
56. Fox MK, Dodd AH, Wilson A, Gleason PM: Association between school
food environment and practices and body mass index of US public
school children. J Am Diet Assoc 2009, 109:S108–S117.
57. Harrison F, Bentham G, Jones AP, Cassidy A, van Sluijs EM, Griffin SJ: School
level correlates with adiposity in 9–10 year old children. Health Place
2011, 17:710–716.
58. Veugelers PJ, Fitzgerald AL: Effectiveness of school programs in
preventing childhood obesity: a multilevel comparison. Am J Public
Health 2005, 95:432–435.
59. Zhu W, Boiarskaia EA, Welk GJ, Meredith MD: Physical education and
school contextual factors relating to students' achievement and cross-
grade differences in aerobic fitness and obesity. Res Q Exerc Sport 2010,
81:S53–S64.
60. Chiodera P, Volta E, Gobbi G, Milioli MA, Mirandola P, Bonetti A, Delsignore
R, Bernasconi S, Anedda A, Vitale M: Specifically designed physical
exercise programs improve children's motor abilities. Scand J Med Sci
Sports 2008, 18:179–187.
61. Heelan KA, Abbey BM, Donnelly JE, Mayo MS, Welk GJ: Evaluation of a
walking school bus for promoting physical activity in youth. J Phys Act
Health 2009, 6:560–567.
62. Donnelly JE, Greene JL, Gibson CA, Smith BK, Washburn RA, Sullivan DK,
DuBose K, Mayo MS, Schmelzle KH, Ryan JJ, et al:Physical Activity Across
the Curriculum (PAAC): a randomized controlled trial to promote
physical activity and diminish overweight and obesity in elementary
school children. Prev Med 2009, 49:336–341.
63. Chomitz VR, McGowan RJ, Wendel JM, Williams SA, Cabral HJ, King SE,
Olcott DB, Cappello M, Breen S, Hacker KA: Healthy living Cambridge kids:
a community-based participatory effort to promote healthy weight and
fitness. Obesity (Silver Spring) 2010, 18(Suppl 1):S45–S53.
64. Jordan KC, Erickson ED, Cox R, Carlson EC, Heap E, Friedrichs M, Moyer-
Mileur LJ, Shen S, Mihalopoulos NL: Evaluation of the gold medal schools
program. J Am Diet Assoc 2008, 108:1916–1920.
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 21 of 22
http://www.ijbnpa.org/content/10/1/101

65. Riley RD, Higgins JP, Deeks JJ: Interpretation of random effects meta-
analyses. BMJ 2011, 342:d549.
66. Rappaport EB, Daskalakis C, Sendecki JA: Using routinely collected growth
data to assess a school-based obesity prevention strategy. Int J Obes
(Lond) 2013, 37:79–85.
67. Rose G, Day S: The population mean predicts the number of deviant
individuals. BMJ 1990, 301:1031–1034.
68. Kolsgaard ML, Joner G, Brunborg C, Anderssen SA, Tonstad S, Andersen LF:
Reduction in BMI z-score and improvement in cardiometabolic risk
factors in obese children and adolescents. The Oslo adiposity
intervention study - a hospital/public health nurse combined treatment.
BMC Pediatr 2011, 11:47.
69. Bacon L, Aphramor L: Weight science: evaluating the evidence for a
paradigm shift. Nutr J 2011, 10:9.
70. Ramanathan S, Allison KR, Faulkner G, Dwyer JJ: Challenges in assessing
the implementation and effectiveness of physical activity and nutrition
policy interventions as natural experiments. Health Promot Int 2008,
23:290–297.
71. Sandy R, Liu G, Ottensmann J, Tchernis R, Wilson J, Ford OT: Studying the
child obesity epidemic with natural experiments. Cambridge, MA: National
Bureau of Economic Research; 2009. [http://www.nber.org/papers/w14989.
pdf] (Accessed 13 June 2013).
doi:10.1186/1479-5868-10-101
Cite this article as: Williams et al.:Systematic review and meta-analysis
of the association between childhood overweight and obesity and
primary school diet and physical activity policies. International Journal of
Behavioral Nutrition and Physical Activity 2013 10:101.
Submit your next manuscript to BioMed Central
and take full advantage of:
• Convenient online submission
• Thorough peer review
• No space constraints or color figure charges
• Immediate publication on acceptance
• Inclusion in PubMed, CAS, Scopus and Google Scholar
• Research which is freely available for redistribution
Submit your manuscript at
www.biomedcentral.com/submit
Williams et al. International Journal of Behavioral Nutrition and Physical Activity 2013, 10:101 Page 22 of 22
http://www.ijbnpa.org/content/10/1/101