scieee Science in your language
[en] (orig)
Effective and cost-effective strategies to prevent
overweight and obesity in South Australia
Vorgelegt von
Dipl.-Volkswirtin Helene Gräfin v. Luckner
aus Uetersen
Von der Fakultät VII – Wirtschaft und Management
der Technischen Universität Berlin
zur Erlangung des akademischen Grades
Doktor der Gesundheitswissenschaften / Public Health
- Dr. P.H. -
genehmigte Dissertation
Promotionsausschuss:
Vorsitzende: Prof. Dr. Leonie Sundmacher
Berichter: Prof. Dr. Christian Gericke
Berichter: Prof. Dr. Reinhard Busse
Tag der wissenschaftlichen Aussprache: 1. November 2012
Berlin 2012
D 83
II
Table of content
List of figures V!
List of tables VI!
Acknowledgements VII!
Abbreviations VIII!
Summary IX!
Zusammenfassung XIII!
1!Introduction 1!
PART I: REVIEW ON THE EFFECTIVENESS AND COST-EFFECTIVENESS OF
PREVENTING OVERWEIGHT AND OBESITY 7!
2!Effectiveness of interventions to promote healthy weight in general populations of
children and adults: a meta-analysis 7!
2.1!Introduction ................................................................................................................... 9!
2.2!Methods ......................................................................................................................... 9!
2.2.1!Literature search ..................................................................................................... 9!
2.2.2!Study selection ..................................................................................................... 10!
2.2.3!Data extraction ..................................................................................................... 10!
2.2.4!Assessment of quality ........................................................................................... 10!
2.2.5!Data synthesis and analysis .................................................................................. 11!
2.3!Results ......................................................................................................................... 12!
2.3.1!Literature search ................................................................................................... 12!
2.3.2!Interventions ......................................................................................................... 13!
2.3.3!Meta-analysis by type of intervention .................................................................. 14!
2.3.4!Sensitivity analyses .............................................................................................. 16!
2.3.5!Subgroup analyses ................................................................................................ 16!
2.4!Discussion ................................................................................................................... 17!
2.4.1!Summary of findings ............................................................................................ 17!
2.4.2!Limitations of the present study ........................................................................... 18!
2.4.3!Recommendations for the evaluation of future interventions to promote healthy
weight 19!
3!Synthesis of studies aimed at the reduction of TV-viewing 20!
3.1!Introduction ................................................................................................................. 20!
3.1.1!Objective .............................................................................................................. 20!
3.1.2!Review framework ............................................................................................... 21!
3.1.3!Development of a theory of change ..................................................................... 22!
3.2!Preliminary synthesis .................................................................................................. 23!
3.2.1!Study characteristics ............................................................................................. 23!
3.2.2!Study context ........................................................................................................ 24!
3.2.3!Intervention content .............................................................................................. 26!
3.2.4!Outcomes .............................................................................................................. 28!
3.2.5!Implementation and compliance .......................................................................... 32!
3.3!Relationship within and between the studies .............................................................. 32!
3.3.1!General characteristics and study context ............................................................ 32!
3.3.2!Intervention content .............................................................................................. 33!
3.3.3!Outcomes .............................................................................................................. 33!
3.3.4!Implementation and compliance .......................................................................... 34!
3.4!Robustness of the synthesis ......................................................................................... 34!
3.5!Research implications .................................................................................................. 35!
III
3.6!Conclusion ................................................................................................................... 36!
4!Critical appraisal of economic evaluations of obesity prevention interventions 37!
4.1!Objective ..................................................................................................................... 37!
4.2!Methods ....................................................................................................................... 37!
4.3!Results ......................................................................................................................... 38!
4.3.1!Literature search ................................................................................................... 38!
4.3.2!General study characteristics ................................................................................ 38!
4.3.3!Results of the appraisal ......................................................................................... 39!
4.3.4!Detailed examination of the appraised studies ..................................................... 41!
4.4!Discussion ................................................................................................................... 46!
4.4.1!Economic evidence ............................................................................................... 46!
4.4.2!Methodological issues .......................................................................................... 47!
4.4.3!Generalisability .................................................................................................... 49!
4.5!Conclusion ................................................................................................................... 49!
5!Systematic review of the effectiveness of school-based policy interventions to
promote healthy weight 50!
5.1!Background ................................................................................................................. 52!
5.2!Methods ....................................................................................................................... 53!
5.2.1!Search strategy ..................................................................................................... 53!
5.2.2!Eligibility criteria ................................................................................................. 53!
5.2.3!Study selection ..................................................................................................... 54!
5.2.4!Study quality ......................................................................................................... 54!
5.2.5!Review process ..................................................................................................... 54!
5.3!Results ......................................................................................................................... 55!
5.3.1!Literature search ................................................................................................... 55!
5.3.2!General study characteristics ................................................................................ 56!
5.3.3!Study quality ......................................................................................................... 59!
5.3.4!Interventions ......................................................................................................... 61!
5.3.5!Behavioural outcomes .......................................................................................... 62!
5.3.6!Anthropometric outcomes .................................................................................... 63!
5.3.7!Implementation ..................................................................................................... 63!
5.4!Discussion ................................................................................................................... 68!
PART II: ECONOMIC EVALUATION OF A HYPOTHETICAL SET OF
INTERVENTIONS TO PREVENT OVERWEIGHT AND OBESITY IN SOUTH
AUSTRALIA 71!
6!Economic issues in the planning of community-based obesity prevention 73!
6.1!Introduction ................................................................................................................. 75!
6.2!Model overview ........................................................................................................... 76!
6.3!Estimating costs ........................................................................................................... 76!
6.3.1!Intervention content and resource inventory ........................................................ 76!
6.3.2!Large scale implementation .................................................................................. 81!
6.3.3!Synergies .............................................................................................................. 82!
6.3.4!Resource mobilization .......................................................................................... 83!
6.4!Estimating consequences ............................................................................................. 83!
6.4.1!Intermediate outcomes ......................................................................................... 83!
6.4.2!Other outcomes ..................................................................................................... 84!
6.5!Discussion ................................................................................................................... 84!
6.6!Conclusion ................................................................................................................... 86!
7!Modeled Costs and Consequences of Multi-faceted Obesity Prevention in South
Australia 88!
7.1!Supplementary details on the calculation .................................................................... 88!
IV
7.1.1!Overview .............................................................................................................. 88!
7.1.2!Intervention content .............................................................................................. 89!
7.1.3!Effectiveness ........................................................................................................ 90!
7.1.4!Cost model ............................................................................................................ 91!
7.1.5!Resources mobilized ............................................................................................. 92!
7.1.6!Scenarios .............................................................................................................. 93!
7.1.7!Parameter uncertainty ........................................................................................... 93!
7.2!Results ......................................................................................................................... 94!
7.2.1!Intervention consequences ................................................................................... 94!
7.2.2!Intervention costs ................................................................................................. 94!
7.2.3!Resources mobilized ............................................................................................. 95!
7.2.4!Scenarios .............................................................................................................. 95!
7.2.5!Parameter uncertainty ........................................................................................... 96!
7.3!Discussion ................................................................................................................... 98!
7.4!Conclusion ................................................................................................................. 100!
8!Discussion 102!
References 109!
Appendix 1 124!
Appendix 2 126!
Appendix 3 127!
Appendix 4 130!
Appendix 5 136!
Appendix 6 147!
Appendix 7 150!
Appendix 8 152!
Appendix 9 167!
V
List of figures
Figure 1-1: Thesis structure ........................................................................................................ 6!
Figure 2-1: Flow chart of literature search ............................................................................... 13!
Figure 3-1: Hypothesized intervention pathway ....................................................................... 23!
Figure 5-1: Flow chart of the literature search ......................................................................... 55!
Figure 7-1: Structure of the indicative community-based intervention .................................... 89!
Figure 7-2: Shift in the BMI distribution as a result of the intervention .................................. 91!
Figure 7-3: Tornado diagram of selected parameter variations with notable impact on annual
intervention cost per child (AU $ in 2010) ............................................................................... 97!
VI
List of tables
Table 2-1: Meta-analyses of interventions to promote healthy weight in children (0-18 years)
from general populations measured as mean difference (MD) in change from baseline in
either BMI or %BF ................................................................................................................... 15!
Table 2-2: Meta-analyses of interventions to promote healthy weight in adults (19-65 years)
from general populations measured as mean difference (MD) in either BMI or %BF ............ 15!
Table 2-3: Selected sensitivity analyses for interventions in children (0-18 years) ................. 16!
Table 3-1: Overview of the main characteristics of studies aimed at the reduction of TV-
viewing in children ................................................................................................................... 24!
Table 3-2: Characteristics of the study populations ................................................................. 25!
Table 3-3: Overview of length and content of the interventions .............................................. 27!
Table 3-4: Common strategies employed in interventions targeted at school-aged children ... 28!
Table 3-5: Summary of the outcomes assessed with statistically significant changes indicated
in bold ....................................................................................................................................... 30!
Table 4-1: Summary of main study characteristics .................................................................. 39!
Table 4-2: Results of the critical appraisal ............................................................................... 39!
Table 4-3: Summary of answers given in the critical appraisal ................................................ 40!
Table 4-4: Overview of results in the economic evaluations ................................................... 45!
Table 5-1: Summary of studies on school-based policy interventions to promote healthy
weight ....................................................................................................................................... 57!
Table 5-2: Quality rating of studies included based on the Quality Assessment Tool for
Quantitative Studies (Thomas 2003) ........................................................................................ 60!
Table 5-3: Outcomes of studies on school-based policy interventions to promote healthy
weight ....................................................................................................................................... 65!
Table 6-1: Content of the hypothetical programme by intervention phase and by
implementation level ................................................................................................................ 78!
Table 6-2: Cost items in the hypothetical programme by implementation level and stakeholder
.................................................................................................................................................. 79!
Table 6-3: Overview of adjustments in each scenario for large-scale implementation ............ 82!
Table 7-1: Estimated costs of the indicative intervention by single strategies ......................... 95!
Table 7-2: Impact on costs and consequences under scenario assumptions ............................. 96!
VII
Acknowledgements
I extend grateful thanks to the Department of Health in South Australia who funded the
research that was conducted in this thesis. This has been a very interesting task to work on
and I appreciate that I was granted this opportunity.
Furthermore, I would also like to acknowledge my supervisor Professor Dr. med. Christian
Gericke. Without his initiative and his ongoing encouragement it would not have been
possible to write this thesis. I am also deeply indebted to Associate Professor John Moss at
the University of Adelaide for his thoughtful guidance and his kind support throughout the
research project.
The University of Adelaide was an inspiring work environment and I am grateful for the
support I received to attend workshops and conferences. It has also been a real pleasure to
work with the people at the Discipline of Public Health. Special thanks go to the members of
the Thesis Writing Group for their frank comments on my writing, which was an invaluable
help. I much enjoyed our comradery throughout the candidature.
Finally, I would like to thank my friends and my family for their support.
VIII
Abbreviations
ABS Australian Bureau of Statistics
AU $ Australian Dollar
BMI Body Mass Index
CI Confidence Interval
CDC Centres for Disease Prevention and Control
CBA Cost-Benefit Analysis
CEA Cost-Effectiveness Analysis
CRD Centre for Reviews and Dissemination
CUA Cost- Utility Analysis
DALY Disability-Adjusted Life Years
DARE Database of Abstracts and Reviews for Effectiveness
EDU Education
HUI Health Utility Index
IASO International Association for the Study on Obesity
IOTF International Obesity Taskforce
ICER Incremental Cost-Effectiveness Ratio
Kg Kilogram
MD Mean Difference
NUTR Nutrition
NZ $ New Zealand Dollar
PA Physical Activity
QALY Quality-Adjusted Life Years
RCT Randomised Controlled Trial
TV Television
%BF Percentage of Body Fat
USA United States of America
US $ United States Dollar
UK United Kingdom
IX
Summary
Background
Problems surrounding excess body weight, known as overweight and obesity, have reached
proportions of a global epidemic. Australia ranks amongst those countries with the highest
prevalence for an unhealthy weight. That signifies a paramount concern, as overweight and
obesity are associated with several chronic diseases, particularly during adulthood. Such
severe consequences for health and wellbeing also represent an enormous economic burden to
society through increased healthcare costs and a morbidity-related decrease in productivity.
Therefore, decision makers worldwide are seeking solutions to control overweight and
obesity. It has been widely advocated that population-based preventions should be given a
key role. There is also particular focus on children, so that the early onset of unhealthy weight
gain can be prevented. Since the present epidemic of overweight and obesity is believed to be
the response to a number of key environmental factors, this suggests that a multi-faceted
strategy is needed to promote a healthy nutrition and opportunities to be physically active.
Hence, interventions with multiple components may yield the most promising approach.
Nevertheless, the evidence base for interventions for the preventing of overweight and obesity
in children is often considered elusive.
In order to inform decision makers concerning the most promising interventions to effectively
prevent overweight and obesity in children, comprehensive evidence synthesis is required,
which markedly involves a rigorous appraisal of the evidence base through a systematic
literature search of published evaluation studies; this will help to identify the most effective
types of intervention. Furthermore, evidence synthesis in a public health decision-making
context also requires rich information concerning how a promising intervention works in a
specific context so that it can be implemented successfully in the target setting. Hence,
information is also needed surrounding the salient components of an intervention, as well as
how these influence the outcomes. In addition, evidence synthesis on obesity prevention
requires the consideration of observational studies, since these are more conducive to
population-based interventions. Eventually, decision makers will need to be able to prioritise
between different options, and thus also require a prior understanding of the costs and
consequences of an intervention.
Objective
The research in this thesis was conducted in order to inform the decision-making context in
the Australian state South Australia.
X
The objectives were firstly to identify interventions that are effective in preventing childhood
overweight and obesity, and secondly to estimate likely costs and consequences if these
interventions were to be implemented in South Australia.
Methods
This thesis has been organised into six main tasks for comprehensive evidence synthesis on
the prevention of overweight and obesity in children:
• The meta-analyses of randomised and non-randomised controlled studies to identify
the best evidence available for effective interventions with a particular focus on the
type and number of programme components
• A structured narrative synthesis of studies concerning interventions that included the
promotion of reduced TV-viewing in order to understand how this component
contributes to intervention effectiveness
• A systematic review of policy-based interventions in the school setting, based on a
wider range of study designs to address specific information needs for the South
Australian decision-making context
• A critical appraisal of economic evaluations on interventions to prevent overweight
and obesity in children in order to summarise evidence for cost-effectiveness,
methodological issues, as well as the generalisability of findings
• An analysis of conceptual issues in the conduct of economic evaluations of complex
public health interventions to prevent overweight and obesity in children.
• An economic evaluation to assess the costs and consequences of a hypothetical set of
school-based interventions implemented state-wide in South Australian communities.
Results
In total, 68 studies were included in the meta-analyses. Generally, heterogeneity between
studies was found to be high in most intervention types. However, there appeared to be
evidence for modest effectiveness in multi-component interventions and studies aimed at the
reduction of TV-viewing in combination with other strategies. The effects were expressed as
reductions in mean body mass index. Furthermore, sensitivity analysis suggested that the
heterogeneity observed could have been attributable to the fact that imputation was required
in many studies to order to derive outcome data.
In the narrative synthesis, the main commonality between the eight interventions to promoting
a reduction in TV-viewing was the school-based education approach. In addition, the
involvement of parents may have been important in facilitating behavioural change in the
XI
home setting. On the other hand, the studies differed widely in terms of the intensity and
duration of the intervention, as well as by the number of other strategies included. No clear
pathway of behavioural change evolved across the studies so that effectiveness, with respect
to TV-viewing, could not fully be understood.
In the systematic review of policy-based interventions in the school setting, eleven studies
were eligible for inclusion. There was only one RCT, and six studies had a before/after
design. The majority of studies evaluated policies targeted at main school food sources.
Generally, evidence for policy effectiveness was uneven—either because studies employed a
number of prevention strategies alongside the policy, or otherwise because of secular trends
in observational designs. The role of the policy component in achieving outcomes often could
not be sufficiently discerned due to a lack of coherent reporting of process-related outcomes.
However, evidence for policies on nutritional standards as well as on the mandated provision
for physical education appeared to be promising.
The critical appraisal revealed that a major challenge for health economic evaluation was
intervention consequences. In order to capture meaningful prevention effects, the timeframe
of the evaluation is often extended until adulthood, where major health implications are most
likely to occur for obesity. However, such an extrapolation involves strong assumptions with
regard to the maintenance of long-term effectiveness. Moreover, the effect magnitudes of the
interventions evaluated are modest; hence, there is considerable uncertainty surrounding the
long-term outcomes. Conversely, economic evaluations constrained to the present time
horizon faced difficulties in expressing outcomes in terms of a generic utility measure, and
natural units of effectiveness in economic evaluations offered little relevance to decision
making.
Furthermore, several important issues emerged when conceptualising the economic
evaluation in this study. Cost issues amounted to the appropriate standardisation of a complex
setting-specific intervention in order to derive a credible resource inventory, and so the
question arose concerning how economies of scale and scope could be incorporated into the
economic evaluation. The main issues with respect to consequences were the translation of a
reduction in mean body mass index into a valid intermediate outcome indicator to inform
decision making, and the consideration of relevant system-level outcomes.
In the economic evaluation, the hypothetical set of interventions in communities and primary
schools would cost AU $415 (in 2010) per child per year; after three years, this was estimated
to prevent approximately 563 out of 75,000 exposed children from exceeding the healthy
weight range. Variations were explored with respect to economies of scale and scope, as well
as geographic and socioeconomic differences in schools. Furthermore, the resource
XII
mobilisation analysis found that 74% of the programme costs would be borne by the state and
the local governments, 16% by parents and other volunteers and 9% by schools.
Conclusion
The research conducted for this thesis found evidence to suggest that multi-component
interventions are able to reduce mean body mass index in children, and thus may be effective
in preventing overweight and obesity, despite the fact that the effect sizes remained modest.
Interventions that include the promotion of reduced TV-viewing are also promising; however,
the role of this component in achieving effectiveness requires more understanding. In
addition, the modelled community-based intervention was estimated to be demanding in
resources, and there may be contention concerning whether or not this provides value for
money.
There were also many impediments to deriving evidence-based recommendations for
effective and cost-effective interventions. It was more difficult to determine effectiveness of
policy-based interventions in the school setting due to weak study designs and a lack of body
composition outcomes. Nevertheless, it is of utmost importance that the evidence base
expands for population-based interventions directed at modifying the obesogenic
environment; therefore, evaluation studies need to enhance the reporting of individual and
process-related outcomes, as this may help to interpret findings.
Furthermore, economic evaluation methods are generally still underdeveloped for
preventative interventions in children. Future evaluation studies should assess the patterns of
resourced use in complex community-based interventions, and more research should be
directed towards developing suitable intervention endpoints.
XIII
Zusammenfassung
Hintergrund
Der Anstieg von Übergewicht und Adipositas hat die Ausmaße einer globalen Epidemie
erreicht. Australien gehört zu den Ländern mit einer der höchsten Prävalenzraten für die
Fettleibigkeit. Leider sind Übergewicht und Adipositas besonders bei Erwachsenen mit vielen
chronischen Erkrankungen assoziiert. Diese schwerwiegenden Folgen für Gesundheit und
Wohlbefinden haben auch drastische ökonomische Auswirkungen wegen erhöhter
Gesundheitsausgaben und einer morbiditätsbedingten Abnahme der Produktivität.
Daher suchen Entscheidungsträger weltweit nach Lösungen, um Übergewicht und Adipositas
Einhalt zu gebieten. Dabei wird der bevölkerungsbasierten Prävention eine zentrale Rolle
zugesprochen. Um schon frühzeitig einer ungesunden Gewichtszunahme entgegenzuwirken,
liegt außerdem ein besonderer Fokus auf der Prävention bei Kindern. Da die derzeitige
Epidemie von Übergewicht und Adipositas als eine Reaktion auf bestimmte Umweltfaktoren
verstanden wird, ist eine facettenreiche Strategie erforderlich, um eine gesunde Ernährung
und Möglichkeiten zur körperlichen Betätigung zu fördern. Daher gelten Interventionen mit
mehreren Komponenten als der vielversprechendste Ansatz. Insgesamt ist jedoch die
Evidenzlage bezüglich der Effektivität von präventiven Interventionen bei Kindern wenig
eindeutig.
Um Entscheidungsträger in Politik und Gesundheit über erfolgsversprechende Interventionen
zur Prävention informieren zu können wird eine umfassende Evidenzsynthese benötigt. Das
beinhaltet zum einen die systematische Literaturrecherche zu allen veröffentlichten
Evaluationsstudien, damit die effektivsten Interventionsarten identifiziert werden können. Zur
erfolgreichen Implementierung von Public-Health-Interventionen sind aber auch
umfangreiche Informationen dazu nötig, auf welche Art und Weise ein effektives Programm
funktioniert. Daher werden speziell auch detaillierte Informationen über die wesentlichen
Komponenten einer Intervention benötigt, insbesondere über deren Einfluss auf die
Ergebnisgrößen. Darüber hinaus verlangt eine Evidenzsynthese zur Adipositas-Prävention die
Berücksichtigung einer Vielzahl von teils schwächeren Studiendesigns, da oft nur solche
möglich sind, um bevölkerungsbasierte Interventionen zu evaluieren. Letztlich müssen
Entscheidungsträger aber auch in der Lage sein, zwischen verschiedenen Optionen
priorisieren zu können, was eine vorherige Kenntnis der Kosten-Effektivität einer
Intervention voraussetzt.
XIV
Ziel der Arbeit
Die Forschungsarbeit in dieser Dissertation wurde im Rahmen eines Projektes durchgeführt,
welches zum Ziel hatte, den Entscheidungsprozess im australischen Bundesstaat South
Australia zu informieren.
Das Hauptziel der Arbeit bestand darin, Interventionen zu identifizieren, die effektiv sind zur
Prävention von Übergewicht und Adipositas bei Kindern, und zweitens voraussichtliche
Kosten und Konsequenzen zu berechnen, wenn diese Interventionen in South Australia
implementiert werden würden.
Methoden
Diese Dissertation beinhaltet sechs Hauptaufgaben für eine umfassende Evidenzsynthese zur
Prävention von Übergewicht und Adipositas bei Kindern:
• Meta-Analysen von kontrollierten Studien (randomisiert und nicht-randomisiert) zur
Ermittlung der besten verfügbaren Evidenz zu wirksame Interventionen mit einem
besonderen Fokus auf die Art und Anzahl der Programm-Komponenten
• Strukturierte narrative Synthese von Studien über die Interventionen, welche gezielt
reduzierte Fernsehzeit fördern, um zu verstehen wie diese Komponente zur
Wirksamkeit der Intervention beiträgt
• Systematische Übersichtsarbeit von Politik-Interventionen in Schulen, basierend auf
einer breiteren Auswahl von Studiendesigns, um relevante Informationen für den
Entscheidungsprozess in South Australia beizutragen
• Kritische Beurteilung von gesundheitsökonomischen Evaluationsstudien zur
Prävention von Übergewicht und Adipositas bei Kindern mit dem Ziel, Evidenz für
die Wirtschaftlichkeit der Programme zusammenzutragen sowie methodische Fragen
und die Verallgemeinerbarkeit der Befunde zu analysieren
• Eine konzeptionelle Analyse zur Durchführung ökonomischer Evaluation bei
komplexen Public-Health-Interventionen, die zur Prävention von Übergewicht und
Adipositas bei Kindern bestimmt sind
• Analyse der Kosten und Konsequenzen einer hypothetischen Reihe von
schulbasierten Interventionsstrategien in den Kommunen von South Australia
Ergebnisse
Insgesamt wurden 68 Studien in die Meta-Analysen einbezogen. Dabei war für die meisten
Interventionstypen die Heterogenität zwischen den Studien sehr hoch. Allerdings ergab die
Meta-Analyse moderate Reduktionen des mittleren Body-Mass-Index für Mehrkomponenten-
XV
Interventionen in der Schule und für Interventionen, die eine Verringerung von Fernsehzeit
anstrebten. Darüber hinaus zeigte die Sensitivitätsanalyse, dass die in vielen Studien
notwendige Nachberechnung der Ergebnisdaten zu der beobachteten Heterogenität
beigetragen haben könnte.
In der narrativen Synthese war die wichtigste festgestellte Gemeinsamkeit zwischen den acht
Studien, dass die Interventionen eine Verringerung des Fernsehkonsums durch
Schulunterricht zu fördern versuchten. Außerdem erschien die Einbeziehung der Eltern
wichtig, um eine Änderung des Fernsehverhaltens im häuslichen Umfeld erfolgreich
umzusetzen. Andererseits unterscheiden sich die Studien stark hinsichtlich Intensität und
Dauer der Intervention und der Anzahl der zusätzlichen Präventionsstrategien. Des Weiteren
blieb der Effektmechanismus der Verhaltensänderung in den Studien kaum nachvollziehbar.
Daher konnte die Wirksamkeit der reduzierten Fernsehzeit als Präventionskomponente nicht
ausreichend geklärt werden.
In der systematischen Übersichtsarbeit zu Politikinterventionen in der Schule entsprachen elf
Studien den Suchkriterien. Darunter war nur eine der fünf kontrollierten Studien auch
randomisiert, während sechs weitere Studien auf einem Vorher/Nachher-Design basierten.
Die Mehrzahl der Studien evaluierte Interventionen, welche Regulierungen für die
wesentlichen Bezugsquellen von Nahrungsmitteln in Schulen beinhalteten. Allgemein war die
Evidenzlage für eine Effektivität der Politikinterventionen nicht eindeutig, entweder weil die
Studien eine Reihe von parallelen Präventionsstrategien implementiert hatten oder aufgrund
anderer säkularer Trends. Allerdings erschienen Nährwertstandards und verpflichtende
Teilnahme am Sportunterricht vielversprechend. Die Rolle der Politik-Komponente konnte
oft nicht ausreichend nachvollzogen werden, da prozessrelevante Effektivitätsmaße oft nur
vereinzelt evaluiert wurden.
Die kritische Bewertung der gesundheitsökonomischen Evaluation ergab, dass vor allem die
adäquate Darstellung der Interventionskonsequenzen problematisch ist. Bei einem Teil der
Studien reichte der zeitliche Rahmen der Evaluation bis ins Erwachsenenalter. Allerdings
basierte eine solche Extrapolation nur auf der Annahme, dass die Wirksamkeit der
Intervention langfristig anhält, was mit erheblicher Unsicherheit verbunden ist. Andrerseits
hatten die Studien, die sich lediglich auf den heutigen Zeithorizont beschränkten,
Schwierigkeiten, die Effektgrößen in Form eines vergleichbaren Maßes zur Lebensqualität
auszudrücken. Somit hatten die Ergebnisse dieser Studien auch weniger Aussagekraft für den
Entscheidungsfindungsprozess.
Des Weiteren wurden verschiedene konzeptionelle Aspekte der ökonomischen Evaluation
von komplexen Public-Health-Interventionen erörtert. Auf der Kostenseite handelte es sich
XVI
um die adäquate Standardisierung einer komplexen ortsbezogenen Intervention, um daraus
ein glaubwürdiges Ressourceninventar ableiten zu können, und um die Einbeziehung von
Skalen- und Verbundeffekten in die Analyse. Die wichtigsten Fragen in Bezug auf die
Interventionskonsequenzen waren die Wahl eines validen Effektivitätsmaßes sowie die
Berücksichtigung relevanter Erfolgsindikatoren auf Systemebene.
Die hypothetische Intervention in Kommunen und Grundschulen würde jährlich 415 AUD
pro Kind kosten (Basisjahr 2010). Außerdem würden durch die Intervention ca. 563 von
75.000 exponierten Kindern in einer gesunden Gewichtskategorie bleiben. Variationen in den
Kosten wurden in Bezug auf Skalen- und Verbundeffekte sowie für geografische und
sozioökonomische Unterschiede in den Schulen untersucht. Darüber hinaus ergab die
Ressourcen-Mobilisierungsanalyse, dass 74 % der Programmkosten auf den Staat und auf die
Kommunen fallen würden, während 16 % von den Eltern und anderen Freiwilligen sowie 9 %
durch die Schulen zu tragen wären.
Schlussfolgerungen/Fazit
Die Evidenzsynthese in dieser Dissertation ergab, dass Mehrkomponenten-Interventionen in
Schulen den Body-Mass-Index von Kindern reduzieren und damit wahrscheinlich eine
wirksame Strategie für die Prävention von Übergewicht und Adipositas darstellen,
wenngleich die Effektgröße eher gering ist. Interventionen, die eine Reduktion des
Fernsehkonsums fördern, sind wahrscheinlich auch effektiv, aber die Rolle dieser
Präventionskomponente erfordert weitere Klärung. Darüber hinaus wäre die modellierte
kommunenbasierte Intervention in South Australia recht ressourcenintensiv bei moderater
Effektivität.
Es gab auch viele Hindernisse beim Ableiten evidenzbasierter Empfehlungen zu effektiven
und kosteneffektiven Interventionen. Es war schwierig, die Effektivität der
Politikinterventionen in Schulen zu bestimmen aufgrund des schwachen Studiendesigns und
weil anthropometrische Effektivitätsmaße meist nicht evaluiert wurden. Jedoch ist es von
besonderer Wichtigkeit, dass die Evidenzlage zu solchen bevölkerungsbasierten
Interventionen erweitert wird. Daher müssen entsprechende Evaluationsstudien die Messung
von individuen- und prozessbezogenen Ergebnisgrößen verbessern, damit sich die Befunde
besser interpretieren lassen.
Darüber hinaus sind ökonomische Evaluationsmethoden bei präventiven Interventionen für
Kinder generell unterentwickelt. Zukünftige Studien sollten die Ressourcennutzung in
komplexen kommunenbasierten Interventionen möglichst umfangreich erfassen, sowie
geeignete Effektivitätsmaße entwickeln.
1 Introduction
Excess body weight—which is distinguished as when an individual is overweight, i.e. with a
BMI ≥ 25 kg/m2, or considered obese, i.e. with a BMI ≥ 30 kg/m21—is considered to be the
leading public health challenge of the 21st Century (Bassett & Perl, 2004; Lobstein et al.,
2004; WHO 2011). Obesity is now widely acknowledged as a global epidemic affecting
industrialised and developing countries alike (Caballero, 2007). For instance, Australia is
known to have one of the highest prevalence rates in the Western world, with 25.6% of males
and 24% females considered obese (ABS, 2008; IASO, 2012).
The consequences of overweight and obesity are known to become most severe during
adulthood owing to the elevated risks of developing cardiovascular disease, diabetes mellitus,
several forms of cancer, physical impairments, such as osteoarthritis, and other conditions that
impact on health and health-related quality of life (Dixon, 2010; Kolotkin et al., 2001). A
recent Australian study estimates that, in 2005, overweight and obese adults incurred
additional healthcare costs of AU $10.7 billion (Colagiuri et al., 2010). In addition, there are
considerable indirect costs to society due to lower productivity and morbidity-related absence
from work (Cawley et al., 2007; Ricci & Chee, 2005). Moreover, adverse effects associated
with overweight and obesity can already be found during earlier stages of life, such as
psychological problems (Dietz, 1998; Reilly et al., 2003). Excess weight during childhood is
also a predictor of obesity during adulthood (Reilly & Kelly, 2011). Hence, it is of particular
concern that the prevalence of overweight and obesity in children has been rising steadily
over the last decades (Olds et al., 2011).
Unfortunately, the options for the obese to lose weight are not only expensive and may bear
additional risks to health, but are also often of little sustainable success (Brownell, 2010).
Hence, treatment in itself would not suffice in terms of gaining control over the epidemic.
Conversely, prevention in children through population-based interventions has been widely
advocated as the key strategy to halt obesity (Gill et al., 2009; Jain, 2005; Swinburn & Egger,
2002). This means particularly universal prevention i.e. interventions that are directed at
everyone in the community, regardless of their weight status, as opposed to only targeting
specific groups at risk (WHO, 2000).2
1Cut-off points specific to age and gender apply in children (Cole et al. 2000).
2The categories of primary, secondary and tertiary prevention are considered to be misleading in the case of overweight and
obesity (WHO 2000). Hence, the terms prevention and promotion of healthy weight refer to universal prevention in this thesis.
Deriving effective prevention concepts requires an understanding of the underlying causes of
the obesity epidemic. The energy imbalance that induces excess weight gain is believed to be
the result of a complex interplay between genetic disposition in individuals, and
environmental factors (Egger & Swinburn, 1997). Although the genes do not appear to have
changed during recent decades, developments, such as globalisation, have led to an
obesogenic environment, which implies an abundance of energy-dense foods at low costs,
technological advances that reduced opportunities to be physically active, and a preference for
a sedentary lifestyle (Drewnowski, 2004; Hill & Melanson, 1999). The number of
environmental contributors to the obesity epidemic therefore also suggest that no single
solution is likely to be effective, but that several types of preventative interventions are
needed at various levels (Kumanyika et al., 2008).
Policy makers worldwide have started to feel the need for action, and thus require information
concerning successful obesity prevention in children. So far, however, it has been difficult to
derive general conclusions since evidence for effective interventions that impact children’s
body composition is limited (Summerbell et al., 2005; Thomas, 2006). Although there is
strong theory-driven support for multi-component interventions, it is also not clear how
effective such types of intervention are. Published systematic reviews have either not
specifically examined the relationship between outcomes and the number of intervention
components or did not establish coherent findings (Brown et al., 2007b; Connelly et al., 2007;
Sharma, 2006).
Adequately informing policy makers further involves evidence concerning how an
intervention works so that it can be implemented successfully (Rychetnik et al., 2004).
Moreover, in the light of scarce public resources, this also includes information surrounding
how alternative intervention options compare in terms of both costs and effects, thus enabling
the setting of priorities (Gold et al., 1996). Hence, formal health economic evaluation is
increasingly being used to guide decisions on public health interventions, despite these posing
a number of conceptual and methodological challenges (Leidl, 2008; Weatherly et al., 2009).
In the Australian state South Australia, approximately one fifth of all four-year-old children
are already found to be obese (Vaska & Volkmer, 2004). Therefore, policy-makers have set
ambitious targets for the reduction of the prevalence of overweight and obesity in the state
(SA Departement of Health, 2006). The research in this thesis was conducted with the
objective to inform the decision-making process in South Australia about promising
interventions based on the evidence for effectiveness, as well as cost-effectiveness.
Objective
The objectives were firstly to identify interventions that are effective in preventing childhood
overweight and obesity, and secondly to estimate likely costs and consequences if these
interventions were to be implemented in South Australia.
Scope of the Thesis
In order to pursue the objective of this thesis, systematic reviews were conducted of published
studies that have evaluated the effectiveness interventions aimed at preventing overweight
and obesity in children. Meta-analyses and narrative synthesis were used to summarise studies
representing the best evidence available for effective interventions. In addition, there was also
a separate synthesis of interventions that were of specific interest to the South Australian
decision-making context. Furthermore, an economic evaluation was conducted based on
effectiveness and cost data obtained from the studies included in the reviews. The analysis
simulated the state-wide implementation of a hypothetical set of interventions and
approximated the programme costs and intermediate consequences on body composition for
South Australia.
Methodological Background
The research in this thesis aimed to derive specific implications for the South Australian
context, but also sought to generate a broad understanding concerning the overall
effectiveness and cost-effectiveness of obesity prevention in children. The methods used refer
to common concepts for evidence-based synthesis, but notably stretch beyond identifying
merely what works, and thus further aims to address questions that add strength to the
generalisability of findings (how does it work?).
The first systematic review was a rigorous evidence synthesis based on systematic literature
searches to include studies with a robust design (randomised and non-randomised controlled
trials). Where possible, the meta-analytic pooling of body composition outcomes was
conducted so as to determine effect sizes of preventative interventions (what works?). There
was a particular focus on the relationship between the type and the number of intervention
components and the outcome, since that has not been specifically assessed by other
systematic reviews.
Whilst this first systematic review resulted in a comprehensive overview of what could
generally be considered the best evidence available on obesity prevention, there were
additional forms of evidence synthesis. In a subset of mainly multi-component intervention
studies, the thesis also aimed to understand the intervention mechanisms at play with respect
to reduced TV-viewing (how does it work?). This type of information is commonly
underrepresented when collating evidence relating to public health interventions, and
disentangling the contribution of single components is difficult. Therefore, the guidance
provided by a narrative review framework was used in order to elucidate the effect
mechanism in the studies.
In addition, there was also a systematic review of interventions concerning a policy-based
modification of the school environment, which was specifically relevant to South Australia.
Such interventions—which are most conducive to population-based prevention—are
commonly evaluated with weaker study designs, with causality for effectiveness more
difficult to determine. Since the evidence base for effective obesity prevention requires
expansion—particularly on this type of intervention—the second systematic review also
examined how the outcomes of these studies could be better understood through including
both individual-level as well as process-related outcomes.
Evidence synthesis was also conducted with the aim of summarising the current situation
concerning economic evaluations of interventions in order to prevent overweight and obesity.
This was done so as to facilitate the conceptualisation of an economic analysis through the
course of this thesis. Economic evaluations of interventions outside neatly defined clinical
settings often struggle with the multi-faceted nature of health promotion. There are currently
no evaluation frameworks that can serve as a blueprint, whilst taking into account all the
important dimensions of multi-component obesity prevention in children. In this respect, the
research conducted in this thesis aimed to provide a case study approach; it explored how a
multi-faceted intervention could be contextualised in a model, and how societal costs and
consequences could be estimated based on the information currently available in the
published literature.
Thesis Structure
The thesis was organised into two main parts: the first part includes all reviews of
effectiveness and cost-effectiveness, whilst the second part comprises the economic
evaluation. The chapters within these two parts were directed at specific research questions
(see outline below). Figure 1-1 also summarises the thesis structure.
The content of Chapter 2 was already published, whilst two chapters were prepared for
submission to a peer-reviewed journal (Chapter 5 and Chapter 6).
PART I: REVIEW
• Which interventions are effective in terms of preventing overweight and obesity in
general populations of children and adults based on the best evidence available?
(What works?)3—Chapter 2:
o Meta-analysis based on a systematic literature search of randomised
controlled trials (RCT) and non-randomised controlled trials
o Outcome of interest was a reduction in mean BMI in mean percentage of
body fat
o Publication: Luckner H, Moss JR, Gericke CA (2011) Effectiveness of
interventions to promote healthy weight in general populations of children
and adults: a meta-analysis. European Journal of Public Health, October 12,
2011: ckr141 [pii] 10.1093/eurpub/ckr141
• How does an intervention aimed at reduced TV-viewing provoke a decrease in mean
BMI in children? (How does it work?)—Chapter 3:
o Narrative synthesis of interventions aimed at the reduction of TV-viewing
that were identified in the meta-analyses.
• What is the evidence for the cost-effectiveness of interventions to prevent overweight
and obesity in children?—Chapter 4:
o Critical appraisal of economic evaluation studies identified alongside the
literature search for the meta-analyses.
• What is the effectiveness of policy-based interventions in terms of promoting healthy
nutrition and physical activity in the school setting? (Specific area of interest to the
South Australian decision-making context?)—Chapter 5:
o Systematic review of studies with experimental, as well as observational,
design
o Outcome of interest was change in body composition, change in behavioural
outcomes, as well as process-related outcomes
o Paper was prepared for submission to a journal for peer-review.
PART II: ECONOMIC EVALUATION
• What are the conceptual issues when conducting health economic evaluation of a
complex public health intervention?—Chapter 6:
3Interventions for adults were included in the meta-analysis, because the small number of relevant publications could be
accommodated without major effort in that study. However, interventions for adults were not further considered in the remaining
part of this thesis.
o Critical analysis of conceptual issues and a proposed evaluation framework
o Paper was prepared for submission to a journal for peer-review.
• What are the societal costs and consequences of a hypothetical state-wide multi-
faceted intervention in the community setting?—Chapter 7.
o Cost-consequence analysis.
Figure 1-1: Thesis structure
Evidence for the effectiveness of interventions
specific to the decision-making context in South
Australia
Best evidence available for the effectiveness of
interventions to prevent overweight and obesity
in children
Part I: Review
Chapter 5: Systematic review
of policy-based interventions
Chapter 2: Meta-analysis of
studies representing best
evidence available
Chapter 3: Narrative synthesis
of interventions aimed at the
reduction of TV-viewing
Chapter 4: Critical appraisal
of economic evaluation
studies
Part II: Economic evaluation
Chapter 6: Analysis of
conceptual issues
Chapter 7: Economic evaluation
PART I: REVIEW ON THE EFFECTIVENESS AND COST-
EFFECTIVENESS OF PREVENTING OVERWEIGHT AND
OBESITY
2 Effectiveness of interventions to promote healthy weight in
general populations of children and adults: a meta-analysis
Helene Luckner1, 2
John R. Moss2
Christian Gericke3
1 School of Population Health and Clinical Practice, University of Adelaide, Australia
2 Department of Health Care Management, Berlin University of Technology, Germany
3 Peninsula CLAHRC, National Institute for Health Research, Peninsula Medical School,
Universities of Exeter & Plymouth, Plymouth, UK
The European Journal of Public Health, Advance Access published October 12, 2011. DOI:
10.1093/eurpub/ckr141
Abstract
Background: Responding to the obesity epidemic requires robust evidence to help prioritize
the allocation of scarce resources to preventive interventions. The aim of this study was to
evaluate interventions that promote healthy weight [defined as reduction in body mass index
(BMI) or percentage body fat] in general populations (unselected by weight) using a
comprehensive meta-analysis. Interventions with both single and multiple components were
considered.
Methods: Studies were first identified through well-conducted systematic reviews
complemented by a search for single studies in five large medical databases up to 6
November 2008. Sixty-eight controlled studies were included. For each intervention type and
age group, all relevant studies were pooled in a random effects meta-analysis.
Results: In children, the highest reductions in mean BMI were achieved through promoting
reduced television viewing [-0.27 kg/m2 (95% CI -0.4 to -0.13 kg/m2)]. Programmes
combining physical activity, specifically themed or general health education and nutrition
achieved a lower reduction [-0.1 kg/m2 (95% CI -0.17 to -0.04 kg/m2)]. Other interventions
had high heterogeneity or showed no statistically significant reduction in outcomes. In adults,
single component interventions were found to reduce both outcome measures. Their mean
percentage body fat was reduced through education by -1.22% (95% CI -1.92 to -0.52).
Conclusion: The evidence for the effectiveness of promoting healthy weight in general
populations is limited, though multi-component interventions in schools and encouraging
reduced children’s television viewing are promising strategies. Improving the reporting of
outcomes is vital, as imputation of inadequately reported measures may have contributed to
the observed heterogeneity. Longer follow-up is essential for understanding policy relevance.
2.1 Introduction
Overweight and obesity represent a major public health burden by increasing the risk of type
2 diabetes, cardiovascular disease, osteoarthritis, various cancers and other severe chronic
conditions (WHO 2006). Given the rising prevalence of overweight and obesity in most
western countries, policy-makers are seeking to allocate the limited resources available for
implementing public health policy into effective preventative strategies (Swinburn et al.
2005). These strategies are being directed towards the general population and robust evidence
of successful prevention is therefore vital.
The best evidence available can generally be found in a well-conducted systematic review of
all relevant randomized controlled trials; however, evidence synthesis restricted to such
studies may be too narrow, because this study design is not always feasible in public health
(NHMRC 1999, Waters 2009). Body composition measured as the body mass index (BMI) or
percentage of body fat (%BF) is well recognised as a valid outcome for population studies on
prevention of overweight and obesity (Cole et al. 2005), but to date only a few meta-analyses
have addressed studies with these outcome measures, and this was mostly done through
pooling standardised effects (Beets et al. 2009, Gonzalez-Suarez et al. 2009, Harris et al.
2009, Kamath et al. 2008, Katz et al. 2008, Seo and Sa 2010, Stice et al. 2006). Two of these
meta-analyses were limited to physical activity interventions in the school setting (Beets et al.
2009, Harris et al. 2009), while three pooled all studies regardless of intervention type
(Gonzalez-Suarez et al. 2009, Kamath et al. 2008, Stice et al. 2006). Only two distinguished
between interventions with multiple components; however, one of these was restricted to the
school setting and the other focused on minority children in the USA (Katz et al. 2008, Seo
and Sa 2010). Useful though all these studies have been, more still needs to be done to address
the information requirements for evidence-based policy. This includes in particular an
understanding of the performance of single interventions compared with multi-component
strategies. The objective of this study was to evaluate interventions that promote healthy
weight (defined as a reduction in BMI or %BF) in general populations (unselected by weight)
from Western countries in a comprehensive meta-analysis.
2.2 Methods
2.2.1 Literature search
Articles were identified through well-conducted systematic reviews on overlapping body
weight reduction-related topics. These were obtained by searching the Cochrane Database of
Systematic Reviews and the Database of Abstracts and Reviews of Effects (DARE). The
10
MeSH terms ‘obesity’, ‘body composition’, ‘nutrition’, ‘physical activity’ and ‘health
promotion’ were used in combination with ‘prevention’. An additional search in PubMed was
performed with the search strategy ‘Overweight (MeSH) and prevention and systematic and
review’. Once the relevant systematic reviews were identified, all full-text articles on
interventions were retrieved from them and assessed for eligibility for meta-analysis. A
second search was conducted in the databases PubMed, Embase, Scopus, CINAHL and the
Cochrane Central Register of Controlled Trials to cross-check whether there were single
studies that were unidentified by the systematic reviews (see Appendix 1 for search
strategies). Eligible studies so identified were also included in the meta-analysis. All searches
were run up to 6 November 2008.
2.2.2 Study selection
Records identified in the search for systematic reviews were screened. Those that clearly
focused on the promotion of healthy weight in general populations (unselected by weight)
were included, but not those with a sole focus on treatment. Primary studies were included if
they reported on controlled, randomized or quasi-experimental projects to promote healthy
weight in general populations together with sufficient outcome statistics measured in BMI
and/or %BF. To be considered a general population, the mean BMI at baseline had to be
within either the normal or overweight ranges, but not the obese (BMI < 30 kg/m2; adjusted
for age and sex in children (Cole et al. 2000). Studies were excluded if the objective included
treatment of eating disorders or differed substantially from healthy weight promotion in a
general population, or if they were conducted in a non-Western country.
2.2.3 Data extraction
Detailed information on the characteristics of all studies reviewed was extracted and
summarized by one investigator. The results from the most recent date of data collection were
used. An attempt was made to contact the authors if follow-up values or measures of variation
were missing if outcomes were only presented graphically or if only a BMI z-score was
reported.
2.2.4 Assessment of quality
The criteria for assessing study quality in Harris et al., which were based on two established
assessment forms (Cochrane EPOC Group 2002, Harris et al. 2009, Jadad et al. 1996), were
considered to be the most appropriate for studies to promote healthy weight. Studies were
assessed for a clear description of inclusion/exclusion criteria and intervention content, an a
11
priori power calculation, blinding of outcome assessment, a method for randomization (where
applicable), reporting of the attrition rate, a reproducible description of statistical methods and
whether baseline characteristics were similar between intervention and control groups. The
criterion concerning the ‘theoretical basis for the intervention’ was not assessed as it was not
relevant for this review.
2.2.5 Data synthesis and analysis
Studies were clustered into groups according to the outcome measured (BMI or %BF), the
type of intervention they referred to and the target age. Meta-analysis was performed on those
groups that contained more than one study. Four different types of intervention were
identified. Regular exercise was classified as ‘physical activity’ (PA). Dissemination of
information or teaching on either general healthy behaviour or specifically related to nutrition,
physical activity or sedentary behaviour was classified as ‘education’ (EDU). If an
intervention consisted of a change in at least one major daily meal, it was classified as
‘nutrition’ (NUTR). Combinations of components were also possible (e.g. EDU + PA). In
addition, a second classification was formed for interventions that aimed at reducing TV-
viewing (TV) regardless of other components involved. One single study could contribute to
more than one intervention group if it consisted of several study arms with different
components. Age was categorized into two major groups: children and adolescents (0–18
years) or adults (>18 years).
In the meta-analysis, the generic inverse variance approach for continuous outcomes in
RevMan 5 (The Cochrane Collaboration 2008) was used with a random effects model as the
base case. An I2 > 50% was understood to indicate substantial heterogeneity. Funnel plots
were used to assess the possible risk of publication bias. No adjustments were made for
multiple statistical comparisons, it being understood that the reader would make due
allowance for this.
The outcome of interest for each intervention group was the mean difference (MD) and its
95% confidence interval (95% CI). This required each study to report the MD in change
scores compared with baseline between the intervention (∆i) and control group (∆c), namely
(∆i – ∆c). Where studies only reported point estimates of ∆i and ∆c, the continuous outcome
comparison in RevMan 5 was used to derive the CI of (∆i – ∆c). Moreover, if only mean
outcome values at baseline and follow-up were reported and not the MD, point estimates of ∆i
and ∆c, were calculated and the respective standard deviations were derived using single
imputation with a correlation coefficient of 0.97. This value was derived from those studies
within this systematic review having sufficient data (Bayne-Smith et al. 2004, Eliakim et al.
12
2007, Harrell et al. 1996, Howard et al. 2006, Proper et al. 2003, Sadowsky et al. 1999).
Subgroup results within one study were combined into one outcome estimate for males and
females together or for different study arms of the same intervention (see Appendix 2 for the
formula used).
To assess the extent of uncertainty, one-way sensitivity analyses were performed. First, a
fixed-effects model was applied to those intervention groups where the base case I2 ≤ 50%.
Second, an adjustment for cluster design was performed where this had not already been
accounted for in the original study. The intra-class correlation coefficient for BMI and %BF
needed for adjustment was taken from one study in this systematic review, which provided
both measures (Yin et al. 2005). Third, the correlation coefficient needed to impute standard
deviations was replaced by 0.9 and 0.999.
Interaction in subgroups was tested with a two-tailed z-test based on all studies for the same
outcome and age range, regardless of intervention. Studies were grouped according to
whether the following applied: randomization; school-based setting; family involvement;
active intervention for control group; participants described as being at risk or disadvantaged;
final measurement > 4 weeks after the intervention ended; data combined for study groups or
outcome statistics imputed. In addition, the following features were also taken into account
when grouping studies: the length of the follow-up (< 3 month, 3–6 months, > 6 months to 1
year, > 1–2 years and > 2 years) and differences within age groups (younger children vs.
children vs. adolescents). Whether the year of publication had an impact was also tested (after
1999; after 2004). Furthermore, the impact of removing each single study in turn from the
intervention group was assessed.
2.3 Results
2.3.1 Literature search
Figure 2-1 depicts the process of the literature search according to the PRISMA statement
(Moher et al. 2009). Twenty-three potentially relevant systematic reviews were identified (see
Appendix 3). Sixty-nine out of 227 full-text articles met the inclusion criteria. These reported
on 51 separate studies. Seventeen additional studies were identified in the second search.
Thus, 68 studies were included in the meta-analysis. Additional information was obtained
from the authors of four studies (Amaro et al. 2006, Burke et al. 1998, Levine et al. 2007,
Neumark-Sztainer et al. 2003).
13
Figure 2-1: Flow chart of literature search
2.3.2 Interventions
Details of the 68 studies included are provided in Appendices 4 and 5. They reported on 84
programmes across 7 intervention types: 16 as PA, 25 as EDU, 3 as NUTR, 22 as PA + EDU,
Total number of full-text articles included in quantitative synthesis after duplicates removed
(n = 103)
Screening
Included
Eligibility
Identification
Full-text articles excluded
(n = 228)
Reasons:
No randomized or controlled
clinical trial (n = 89)
Intervention is treatment of
obesity or participants were
obese at baseline (n = 58)
No data for quantitative
synthesis (n = 27)
Body Mass Index or percentage
body fat not assessed (n = 25)
Conducted in a non-western
country (n = 19)
Different study objective
(n = 10)
Records after duplicates removed
Search for single studies:
Records identified through database searching
(n = 3,972)
PubMed: n = 1,530
Scopus: n = 1,402
EMBASE: n = 547
Cochrane Central Register of Controlled Trials:
n = 387
Full-text articles included in quantitative synthesis
(n = 70)
Reporting on 50 studies
Records
screened
Records excluded
(n = 2,732)
Full-text
articles
assessed
(n = 298)
Search for systematic reviews:
Records identified through database searching
(n = 736)
Records screened after duplicates removed
(n = 517)
Full-text obtained (n = 78)
Systematic reviews excluded because no clear aim
regarding the promotion of healthy weight
(n = 55)
Records after duplicates removed
Full-text articles excluded
(n = 158)
Reasons:
Body Mass Index or percentage
body fat not assessed (n = 68)
Intervention is treatment of obesity
or participants were obese at
baseline (n = 61)
No data for quantitative synthesis
(n = 14)
No randomized or controlled
clinical trial (n = 8)
Conducted in a non-western
country (n = 5)
Full-text could not be retrieved
despite maximal effort (n = 2)
Full-text
articles
assessed
(n = 227)
Records
screened
Records excluded
(n = 0)
Full-text articles included in quantitative synthesis
(n = 69)
Reporting on 51 studies
14
3 as EDU + NUTR, 6 as PA + EDU+ NUTR and 9 as TV. Forty-seven studies used a cluster
as the unit of analysis and 54 were randomized. The length of the intervention varied from 1
month to > 7 years. Forty-seven studies measured only BMI as an outcome, while 7 measured
only %BF and 14 measured both outcomes. Assessment of %BF was performed by skin-fold
thickness in seven studies; another seven studies used bioelectrical impedance. Dual-energy
X-ray absorptiometry was used in five studies and two studies used underwater weighing.
Three studies were about younger children (<6 years), 34 were about children (6–12 years),
18 about adolescents (12–18 years) and 13 involved adults (>18 years).
Most studies scored well for providing a clear description of their intervention, the attrition
rate and the statistical analysis. Conversely, only a minority of studies reported information
on an a priori power calculation, a blinded outcome assessment or the method of
randomization.
2.3.3 Meta-analysis by type of intervention
For the children, meta-analysis was undertaken for the intervention groups PA, EDU, PA +
EDU and PA + EDU+ NUTR regarding both BMI and %BF, but for EDU+ NUTR and TV
regarding BMI only. Three intervention groups contained only one study: NUTR measured in
BMI as well as EDU+ NUTR and TV measured in %BF. In adults, meta-analysis was feasible
for PA and EDU in both outcome measures, but for NUTR only for BMI.
An imputation of the CI around the outcome estimates was needed in 22 studies. MD for each
intervention group in children is summarized in Table 2-1. In each situation where meta-
analysis could be performed, there was a reduction (negative MD) in BMI. Except for PA and
EDU+ NUTR, these MD were statistically significant. The reduction was highest in TV
followed by PA +EDU and EDU.
The children’s intervention group for PA also showed a statistically significant reduction in
%BF. In EDU and PA for adults, the reduction was statistically significant for both outcomes
(Table 2-2).
The level of heterogeneity was substantial in eight of the intervention groups. The respective
forest plots of the seven intervention groups with low heterogeneity can be found in Appendix
6. Due to small numbers of studies being included or high levels of heterogeneity, the results
of funnel plot analyses were not interpretable and thus not reported.
15
Table 2-1: Meta-analyses of interventions to promote healthy weight in children (0-18
years) from general populations measured as mean difference (MD) in change from
baseline in either BMI or %BF
Intervention group by component(s) and
outcome measure (change from baseline)
Sample
size
Number of studies
included (Number of
controlled but non -
randomised studies)
Mean Difference (MD)
[95% CI]
I²
PA (BMI)a
2,927
10 (1)
-0.15 kg/m2 [-0.33, 0.03]
87%
EDU (BMI)b
5,667
15 (6)
-0.15 kg/m2 [-0.24, -0.07]
65%
NUTR (BMI)c
103
1 (0)
-0.14 kg/m2 [-0.55, 0.27]
-
PA+EDU (BMI)d
8,399
21 (3)
-0.19 kg/m2 [-0.37, -0.02]
94%
EDU+NUTR (BMI)
1,695
2 (1)
-0.05 kg/m2 [-0.2, 0.1]
0%
PA+EDU+NUTR (BMI)e
10,257
6 (2)
-0.1 kg/m2 [-0.17, -0.04]
5%
TV and other (BMI)
3,962
8 (2)
-0.27 kg/m2 [-0.4, -0.13]
20%
PA (%BF)a
1,989
6 (1)
-0.7% [-1.05, -0.31]
59%
EDU (%BF)b
1,110
2 (0)
+0.13% [-0.04, 0.3]
0%
PA+EDU (%BF)d
1,915
6 (2)
-1.07% [-2.27, 0.13]
97%
EDU+NUTR (%BF)c
1,419
1 (0)
+0.18% [-1.75, 2.11]
-
PA+EDU+NUTR (%BF)e
1,517
2 (1)
+0.78% [-0.29, 1.86]
56%
TV and other (%BF)c
459
1 (0)
+0.08% [-0.68, 0.84]
-
BMI: Body Mass Index; %BF: Percentage of body fat; CI: Confidence interval; EDU: Education; MD: Mean
difference; NUTR: Nutrition; PA: Physical activity; TV and other: Promotion of reduced TV-viewing in
combination with other strategies.
aFour studies contribute to both sets of meta-analyses (BMI and %BF). bOne study contributes to both sets of
meta-analyses (BMI and %BF). cConsisted of a single study only; therefore no meta-analysis was performed. dFive
studies contribute to both sets of meta-analyses (BMI and %BF). eTwo studies contribute to both sets of meta-
analyses (BMI and %BF).
Table 2-2: Meta-analyses of interventions to promote healthy weight in adults (19-65
years) from general populations measured as mean difference (MD) in either BMI or
%BF
Intervention group by component and
outcome measure (change from baseline)
Sample
size
Number of studies
includeda
Mean Difference (MD)
95% CI
I²
PA (BMI)b
71
2
-1.24 kg/m2 [-1.62, -0.85]
0%
EDU (BMI)c
42,567
8
-0.41 kg/m2 [-0.63, -0.19]
73%
NUTR (BMI)
113
2
-1.40 kg/m2 [-3.66, 0.87]
94%
PA (%BF)b
130
2
-1.97% [-2.66, -1.29]
0%
EDU (%BF)c
769
4
-1.22% [-1.92, -0.52]
38%
BMI: Body Mass Index; %BF: Percentage of body fat; CI: Confidence interval; EDU: Education; MD: Mean
difference; NUTR: Nutrition; PA: Physical activity.
aAll studies were randomised. bOne study contributes to both sets of meta-analyses (BMI and %BF). cThree studies
contribute to both sets of meta-analyses (BMI and %BF).
16
2.3.4 Sensitivity analyses
Applying a fixed-effects model narrowed the CI in the group TV [MD -0.27 kg/m 2 (95% CI -
0.39 to -0.16 kg/m2)]. There was a higher MD for PA +EDU +NUTR [MD: -0.11 kg/m2 (95%
CI: -0.16 to -0.05 kg/m2)], but a lower MD measured in %BF in EDU for adults [MD: -1.17%
(95% CI: -1.62 to -0.71)].
Table 2-3 shows how additional cluster adjustment in 18 studies lowered the level of
heterogeneity in all intervention groups. Furthermore, I2 varied considerably when the
correlation coefficient for imputation of standard deviations was replaced by 0.9 or 0.999. In
adults, the I2 for these two replacements were 70% and 75% in the group EDU measured in
BMI, while these were 92% and 95% in NUTR.
Table 2-3: Selected sensitivity analyses for interventions in children (0-18 years)
Intervention group by
component(s) and outcome
measure
(change from baseline)
Mean Difference (MD) [95% CI] and I²
(adjustment for cluster design)
I2 (Correlation
coefficient = 0.9
used for
imputation)a
I2 (Correlation
coefficient =
0.999 used for
imputation)a
PA (BMI)
-0.15 kg/m2 [-0.34, 0.04], I2= 85%
46%
97%
EDU (BMI)
-0.18 kg/m2 [-0.3, -0.05], I2= 46%
50%
79%
NUTR (BMI)b
no adjustment required
-
-
PA+EDU (BMI)
-0.17 kg/m2 [-0.34, -0.00], I2= 86%
83%
99%
EDU+NUTR (BMI)c
-0.04 kg/m2 [-0.25, 0.17], I2= 0%
PA+EDU+NUTR (BMI)
-0.1 kg/m2 [-0,16 -0.04], I2= 3%
0%
34%
TV and other (BMI)c
-0.27 kg/m2 [-0.4, -0.13], I2= 20%
PA (%BF)
no adjustment required
58%
80%
EDU (%BF)
no adjustment required
0%
0%
PA+EDU (%BF)
-0.95% [-2.22, 0.32], I2= 90%
91%
99%
EDU+NUTR (%BF)b
no adjustment required
-
-
PA+EDU+NUTR (%BF)
+0.42% [-0.54, 1.38.], I2= 0%
18%
72%
TV and other (%BF)b
no adjustment required
-
-
BMI: Body Mass Index; %BF: Percentage of body fat; CI: Confidence interval; EDU: Education; MD: Mean
difference; NUTR: Nutrition; PA: Physical activity; TV and other: Promotion of reduced TV-viewing in
combination with other strategies.
aThe correlation coefficient used to impute the outcome estimate in the base case was 0.97. bConsisted of a single
study only; therefore no meta-analysis was performed. cStudies in this intervention group did not require
imputation.
2.3.5 Subgroup analyses
Only statistically significant interactions are reported here. For studies on children regardless
of intervention type, the MD was significantly larger in those aged younger than 6 years
17
compared with those between 6 and 12 years (P = 0.03). The test for interaction regarding
studies reporting %BF was significant for studies with a longer intervention duration showed
paradoxically a higher increase in children’s %BF (P = 0.05 for length > 6 months), although
why this should be so is unclear. Heterogeneity was high throughout the respective
subgroups. Across the adult interventions, there was a significantly lower MD in BMI for
studies not requiring imputation (P < 0.0001).
2.4 Discussion
2.4.1 Summary of findings
Although the two literature searches when combined revealed 453 (= 227 + 298 – 72
duplicates) potentially relevant articles on the promotion of healthy weight in general
populations, only 103 articles met the criteria for inclusion in the synthesis. One major reason
for this was that many studies reported only process-related measures rather than BMI or
%BF. The two separate searches proved necessary, because neither was exhaustive, with only
33 studies being found by both.
The relevant meta-analysis suggested that interventions with multiple components and those
that aimed to reduce TV-viewing in children led to a significant reduction in BMI. Two
parallel meta-analyses suggested that interventions where adults received lifestyle education
led to a statistically significant decrease in BMI or %BF, although the interpretation of this is
complicated by heterogeneity. However, PA directed at adults had a considerable effect on
both BMI and %BF and hence may be a promising strategy. There were also statistically
significant reductions for five of the other intervention groups, but substantial heterogeneity
limits the ability to interpret these results.
It cannot be ruled out that promoting a reduction in TV-viewing may have led concurrently to
reducing inactivity as well as lowering the energy intake derived from eating while watching
TV or reducing food advertisement exposure. Moreover, only one of these eight studies aimed
solely at reducing TV-viewing (Robinson 1999), while the others also incorporated physical
activity or other components, which may have had a separate impact on BMI. Evidence for
the success of this intervention is promising, but warrants further research.
PA+EDU+NUTR measured in BMI also appear promising. The six studies included in this
multi-component intervention group were all school based. Five of them lasted for at least 2
years and four also reported involvement of the family. Thus, this finding supports the current
18
opinion that these features characterize good practice for obesity prevention (Brown and
Summerbell 2009). Nevertheless, the reduction in BMI appeared rather small.
The heterogeneity commonly observed reflects differences in study design within each
intervention group. Despite the effort to pool similar studies, the programmes in each study
varied by content, intensity and length and so did the characteristics of the study population
by age in years, gender, ethnicity, socioeconomic status and initial weight at baseline. In
addition, the necessity to impute outcome statistics may have had a considerable impact on
the heterogeneity found. This was shown in the sensitivity analysis, where I2 decreased with
lower values for the correlation coefficient used for imputation. The value employed in the
base case (0.97) may have been high, but was calculated from the studies included in this
systematic review.
Even though more statistically significant reductions in BMI or %BF were found for different
intervention types than in the previous meta-analyses (Beets et al. 2009, Gonzalez-Suarez et
al. 2009, Harris et al. 2009, Kamath et al. 2008, Katz et al. 2008, Seo and Sa 2010, Stice et
al. 2006), the effectiveness of the interventions studied remains relatively modest in both age
groups. Since these studies mostly promoted individual behaviour change and targeted
children in the school setting, it may be reasonable to question the impact of this type of
intervention and to seek to compare it to what might be achievable by more comprehensive
population-based approaches.
Had there been less heterogeneity in eight of the intervention groups where a meta-analysis
was performed, potentially meaningful observations might have been derived from this
review regarding the generally modest reductions in outcomes and a possible interaction of
effects in multi-component interventions.
2.4.2 Limitations of the present study
There is a potential source of bias due to including controlled but non-randomized studies. In
addition, the degree of heterogeneity made the interpretation of funnel plots difficult. Another
limitation was that deriving missing CI around outcome estimates was approached by single
instead of multiple imputations. The latter would have increased the standard errors to
account for the uncertainty of imputation. Hence, the outcome estimates of those 22 studies
that needed imputation may be overly precise in the present study. It should be noted that MD
for different intervention types should be compared with caution since this is only an
informative indirect comparison; had there been less heterogeneity, a more complex multiple-
treatment analysis could have been conducted (Caldwell et al. 2005). Finally, there are known
19
shortcomings in the outcomes analyzed. BMI does not fully capture changes in body
composition. It is widely held that %BF may be a better indicator of this. However, %BF was
measured differently across the studies that reported it, which may have compromised the
comparability of their outcomes.
2.4.3 Recommendations for the evaluation of future interventions to promote healthy
weight
Kropski et al. listed the need for improvements such as control for covariates of the outcome,
appropriate adjustment for cluster design, subgroup analysis by gender and special assessment
of the impact on overweight and obese participants (Kropski et al. 2008). In addition, our
findings demonstrate that all studies should report both BMI and %BF and present outcome
estimates with a measure of variation derived from ANCOVA or equivalent approaches so
imputation can be avoided. These outcome estimates should also be reported separately for
each sex.
Measures of distribution should be reported because a reduction in mean BMI (or %BF) does
not indicate whether this was general or predominantly in a subgroup.
Improvements in the future design and reporting of overweight and obesity prevention trials
are required to ensure that changes in body composition remain comparable across studies
and compatible with a rigorous synthesis. In view of the modest weight reductions so far
achieved, long-term follow-up is needed to gain greater insight into their maintenance and
their relevance for healthy public policy.
Acknowledgements
The views expressed in this publication are those of the author(s) and not necessarily those of
the South Australian Department of Health, the NHS, the NIHR or the Department of Health
in England. An earlier draft of this work was presented as a poster at the International
Conference of Obesity in Stockholm, 11–15 July 2010.
Funding
A Strategic Health Research Programme Grant by the South Australian Department of Health
(SHRP 9881) and support from the National Institute for Health Research (NIHR) in England
for CAG’s contribution is also gratefully acknowledged.
20
3 Synthesis of Studies Aimed at the Reduction of TV-viewing
3.1 Introduction
3.1.1 Objective
The meta-analysis in Chapter 2 found a statistically significant reduction in mean BMI for a
group of eight studies that incorporated the promotion of reduced TV-viewing in their
intervention: Robinson (1999); Robinson et al. (2003); Dennison et al. (2004); Chavarro et al.
(2005) and Gortmaker et al. (1999)4; Fitzgibbon et al. (2002), Stolley et al. (2003),
Fitzgibbon et al. (2005)5; Harrison et al. (2006); Kipping et al. (2008); and Sanigorski et al.
(2008)6.
In comparison to other intervention types of the review, this finding stood out with a high
effect magnitude and little observed heterogeneity between studies, therefore suggesting
effectiveness. Nevertheless, this type of intervention was based on a subset of studies with
differing intervention components, and it was not clear whether the pooled estimate could be
attributed to the TV-reduction component only. Although this evidence can only be
interpreted with caution, it nevertheless offers a promising finding for overcoming issues
surrounding obesity, and thus warrants more research through future trials.
Nevertheless, the information currently available from the studies included in the meta-
analysis has the capacity to help explore the ways in which the intervention may work. This
means specifically a more thorough understanding is required about the underlying
intervention mechanism, i.e. a chain of variables mediating the effect in each study
(Baranowski et al., 2009). Once such mechanisms become sufficiently clear and appear to be
working, it can further be investigated whether or not a common pathway evolves across the
studies. Such an analysis is extremely useful in the context of public health policy, which
generally requires knowledge surrounding interventions, which stretches beyond merely
identifying what is effective, but which also requires rich information concerning why that
was achieved and how it should be implemented in the future (Rychetnik et al., 2004;
Shepperd et al., 2009). The reproducible elements of a programme can be particularly well
4For convenience, these will henceforth be referred to simply as Chavarro et al. 2005.
5For convenience, these will henceforth be referred to simply as Fitzgibbon et al. 2005.
6This study was supplemented by an unpublished report that described the details of the TV-reduction component. For
convenience, the study will henceforth be referred to simply as Sanigorski et al. 2008.
21
assessed where systematic reviews elucidate the intervention mechanism, whilst also
considering the context specific to an individual study (Shepperd et al. 2009).
Therefore, the aim of this chapter was to review the programme mechanism in these eight
primary studies in an attempt to understand whether reduced TV-viewing is an effective
intervention component to prevent overweight and obesity in children.
3.1.2 Review framework
The appraisal of public health interventions with multiple facets is often conducted through
narrative synthesis, which is the textual approach to integrating study findings (Dixon-Woods
et al., 2005; Popay et al., 2006). This is also frequently used in systematic reviews with
studies where data cannot be pooled but where such synthesis can be conducted alongside
meta-analyses; this is done so as to enrich the understanding of an intervention by offering
interpretative explanations about common patterns (Rodgers et al., 2009). Narrative synthesis
involves exploring relationships within and between the studies and by assessing synthesis
robustness (CRD 2008). Therefore, a narrative synthesis was used in this chapter to review
the intervention mechanisms of the eight studies.
However, few conventions exist on the ways in which the findings of a narrative synthesis
should be organised and presented. In contrast to quantitative pooling techniques, this type of
synthesis is commonly carried out more informally with some discretion by the analyst in
terms of which aspects to include. Hence, this leaves a potential risk of omitting important
information and introducing subjectivity into the review (Dixon-Woods et al., 2005). In order
to enhance the strength and transparency of this analysis, the approach supported by Popay et
al. (2006), who provide guidance on the conduct of narrative synthesis in systematic reviews,
was followed. Their generic framework evolved from a thorough review of the
methodological literature, and comprises four major steps (Arai et al., 2007; Popay et al.,
2006; Rodgers et al., 2009):
(i) Developing a theory of change
(ii) Preliminary synthesis
(iii) Exploring relationships within and between studies
(iv) Assessing robustness of the synthesis.
However, the authors of this framework acknowledge that exact synthesis techniques for this
type of research are inherently dependent on the context of the review, and have therefore
refrained from prescribing definite rules.
22
The following approach was taken in this chapter. Step (i) was conducted to identify the
variables likely to be part of the intervention mechanism, as well as contextual factors. The
preliminary synthesis (Step ii) reviewed the studies on general characteristics, the study
context, the intervention content, outcomes, and the implementation process, which was
followed by the presentation of the main commonalities and differences between the studies
(Step iii). Based on this synthesis, the possibility of establishing a global intervention
mechanism was explored in Step iv through the discussion of the strengths of the synthesis. In
addition, implications for future interventions resulting from this research synthesis were
derived.
3.1.3 Development of a theory of change
In order to organise the synthesis, a preliminary concept was derived concerning how the
intervention might work. The meta-analysis result provided two cornerstones for the
formulation of an ad-hoc theory of change for this purpose: the starting point is an
intervention aimed at the reduction of TV-viewing; the end point is a reduction in children’s
body composition (i.e. mean BMI) caused by an energy imbalance. Plausible options for
mechanisms that link these two points could then be hypothesised; this was broken down into
two parts, as depicted in the grey area in Figure 3-1. The first mechanism that needed to be
clarified was how the intervention reduces the risk factor, i.e. children’s time spent in front of
the TV. The second mechanism involved the channel through which reduced TV time impacts
on the energy balance.
In the case of the first mechanism, two major types of intervention could be used in the
studies to influence TV-viewing (Glanz & Bishop, 2010; Stokols, 1992). One is aimed
directly at individual behaviour, and involves a change either in attitudes through health
education or in cognitive skills through improved self-control. The alternative is to provoke a
reduction in TV-viewing time through an environmental modification. Given that obesity
prevention thus far has been dominated by behavioural and educational interventions, it could
be expected that the studies included would predominantly be found in the first intervention
category. Nevertheless, at least elements of environmental change could have also been a part
of the intervention.
Secondly, two main pathways have been suggested to link increased TV-viewing and energy
imbalance (Dietz & Gortmaker, 1985): one is the increased sedentariness in children that
lowers energy expenditure; the other is the higher intake of food and beverages due to more
between-meal-snacking whilst viewing TV. It is also possible that a higher exposure to
advertising intensifies the child’s demand for unhealthy products. Thus, it was plausible to
23
assume that a reduction in TV-viewing would decrease dietary intake and/or increase physical
activity levels. A further distinction could be made in regard to whether this would only mean
reduced sedentariness, or whether moderate to vigorous physical activity may have also
increased.
The task in the following steps of this synthesis was to clarify which of these suggested
pathways occurred in the studies, and whether there were dominant patterns.
Figure 3-1: Hypothesised intervention pathway
3.2 Preliminary synthesis
3.2.1 Study characteristics
Details of the eight studies included are displayed in Table 3-1. With the exception of one
study, all were both controlled and randomised, although the unit of randomisation was
mostly the school or the class (Sanigorski et al., 2008). The studies were published between
1999 and 2008, with the majority conducted in the USA, with only one study from England,
one from Ireland, and one from Australia. Only one study had a large sample size with more
than 1,000 participants in each group (Sanigorski et al., 2008). In contrast, two studies had a
very low statistical power (Dennison et al., 2004; Robinson et al., 2003).
Inter-
vention
Dietary intake
a) Frequency
of snacking
b) Exposure to
advertisement
Physical
activity (PA)
a) Sedentary
behaviour
b) Moderate to
vigorous PA
Energy
balance
BMI
Time spent
TV-
viewing
Cognition,
Attitudes,
Beliefs
Environment
First mechanism
Second mechanism
24
Four studies comprised of a school-based education component only (Chavarro et al., 2005;
Dennison et al., 2004; Harrison et al., 2006; Robinson 1999). In addition, three studies also
comprised a physical activity component (Fitzgibbon et al., 2005; Kipping et al., 2008;
Robinson et al., 2003). The remaining study focused on a wider range of health-promotion
activities in the school and community setting, and consisted of multiple components
(Sanigorski et al., 2008). Although all studies included an educational component, only two
studies employed this with the sole objective to reduce TV-viewing (Dennison et al., 2004;
Robinson 1999). All studies aimed to impact on obesity-related outcome variables.
Table 3-1: Overview of the main characteristics of studies aimed at the reduction of TV-
viewing in children
Author
Purpose and outline of the intervention
Study country
(state)
Design
Sample size at
baseline
Robinson
1999
School-based education that promoted screen time reduction
to lower adiposity levels in students
USA (California)
RCT
(cluster)
I: 95
C: 103
Robinson et
al. 2003
Home-based counselling to reduce TV-viewing and after-
school dancing classes at the community centre to prevent
excess weight gain in minority children
USA (California)
RCT
I: 28
C: 33
Dennison et
al. 2004
Day-care-based education that promoted TV-viewing
reduction to impact on growth variables in childrena
USA (New York)
RCT
(cluster)
I: 43
C: 34
Chavarro et
al. 2005
School-based education to reduce obesity by altering key
physical activity and dietary risk factors, including TV-
viewing
USA
(Massachusetts)
RCT
(cluster)
I: 259
C: 249
Fitzgibbon et
al. 2005
Pre-school-based programme that includes activity sessions
and health education to reduce increases in children’s
weight
USA (Illinois)
RCT
(cluster)
I: 197
C: 212
Harrison et al.
2006
School-based education targeted at sedentary behaviour and
physical activity to impact on students BMIa
Ireland
RCT
(cluster)
I: 182
C: 130
Kipping et al.
2008
School-based education aimed at improving knowledge
about physical activity and nutrition as well as reducing TV-
viewing to decrease student’s BMI
England
RCT
(cluster)
I: 249
C: 223
Sanigorski et
al. 2008
Community-based multi-faceted health promotion
programme to reduce increases in anthropometric measures
Australia
(Victoria)
CT
(cluster)
I: 1103
C: 1183
RCT: randomised controlled trial, CT: controlled trial, I: intervention, C: control.
aThe intervention was part of a general health promotion programme, but the evaluation is restricted to the TV-
viewing component.
3.2.2 Study context
Seven of the eight studies were institution-based and were conducted in schools, pre-schools,
and day-care settings. Conversely, the study of (Robinson et al., 2003) was carried out
through home visits. In this first group of seven, the TV-reduction component of the
intervention mainly comprised educational sessions. The interventions in these studies did not
appear to impact on the whole school or day-care setting policy or culture with regard to TV-
viewing other than a small-scale social marketing campaign in (Sanigorski et al., 2008).
Nevertheless, there were explicit attempts to involve parents in the intervention in six studies
(Dennison et al., 2004; Fitzgibbon et al., 2005; Harrison et al., 2006; Robinson, 1999;
25
Sanigorski et al., 2008). In these studies, parents received information through newsletters,
participated in the tasks that were to be carried out at home, or were otherwise asked to
provide a monitoring role. Parents were involved similarly in the only home-based study
(Robinson, 1999).
Studies focused either on younger children between two-and-a-half and five-and-a-half years
or on children between eight and twelve years (see Table 3-2). However, the intervention in
one study was targeted at a broader age group between four to twelve years (Sanigorski et al.,
2008). One study included girls only (Robinson et al., 2003), and one focused only on present
outcomes measured in a subset of females who had a pre-menarche status at baseline,
although the intervention was delivered to male students as well (Chavarro et al., 2005). The
other studies had a mixed-gender population. Participants belonged to an ethnic minority in
two studies (Fitzgibbon et al., 2005; Robinson et al., 2003). In addition, two studies targeted
students from a low-income background (Harrison et al., 2006; Robinson et al., 2003).
Six studies provided information concerning the baseline prevalence of either overweight or
overweight and obesity, although this was based on different cut-off point definitions. Three
studies based this on the growth charts of the Centers for Disease Control and Prevention
(CDC) (Kuczmarski et al., 2002), two studies used the International Obesity Task Force
(IOTF) cut-off points (Cole et al., 2000), and one study referred to the UK National BMI
classification (Cole et al., 1995). Hence, the direct comparability of these prevalence rates
was limited; however, it appeared that the combined prevalence of overweight and obesity
was broadly similar in four studies (between 25% and 35%), but notably lower in two studies
(Chavarro et al., 2005; Dennison et al., 2004). Furthermore, in one study in which prevalence
was not reported, cross-checking against the CDC growth charts showed that the majority of
participants in that study were presumably overweight (Robinson et al., 2003).
Table 3-2: Characteristics of the study populations
Author
Age group
(years)
Special profile study
population
Mean BMI (SD)
at baseline
Prevalence of overweight/obesity at
baseline
Robinson
1999
8-9
I: 18.38 (3.67)
C: 18.10 (3.77)
Not reported (presumably normal weighta)
Robinson et
al. 2003
9-10
African-American or
Black; BMI ≥ 50th
percentile and/or at least
one overweight
parent/guardian
I: 20.95 (5.39)
C: 21.57 (5.26)
Not reported (presumably overweighta)
Dennison et
al. 2004
2.5-5.5
Children from low-
income households
I: 15.9 (1.96)
C: 15.9 (1.16)
I and C: 17.6% in at least 85th percentile
(CDC growth charts)
Chavarro et
al. 2005
10-12
I: 19.8 (4.4)
C: 19.9 (4.1)
I: 10.8% and C: 10.45%
> 85th percentile (CDC growth charts)
Fitzgibbon et
al. 2005
3-5
Majority of children
Black or Hispanic
I: 16.5 (1.5)
C: 16.7 (2.0)
I: 31.5% and C: 36.3%
≥ 85th percentile (CDC growth charts)
Harrison et
al. 2006
10
Schools in areas of high
social disadvantage
I: 19.0 (2.7)
C: 19.2 (4.56)
I: 35% and C: 32% overweight (IOTF cut-
off points)
26
Author
Age group
(years)
Special profile study
population
Mean BMI (SD)
at baseline
Prevalence of overweight/obesity at
baseline
Kipping et al.
2008
8-10
I: 17.9 (3.0)
C: 17.9 (3.0)
I: 10.4% and C: 14.8% overweight
I: 15.3% and C: 12.6% obese (UK
National BMI classification with a 1990
reference population)
Sanigorski et
al. 2008
4-12
I: 18.0 (3.0)
C: 17.9 (2.9)
I: 18.76% and C: 19.73% overweight
I: 8.55% and C: 6.77% obese (IOTF cut-
off points)
CDC: Centres for Disease Control and Prevention; IOTF: International Obesity Task Force.
aAssumption was based on a cross-check of the mean BMI reported in the study against the age-and sex-specific
CDC growth charts.
3.2.3 Intervention content
This scope of this review section was restricted to that part of the intervention aimed at the
reduction of TV-viewing. As shown in Table 3-3, education to reduce TV-viewing was
delivered with a different duration and intensity across all studies. Four studies dedicated
between two and six months to the component, but in the other four studies, this was only a
small number of sessions. Programme staff delivered the intervention in three of the studies
(Fitzgibbon et al., 2005; Robinson, 1999; Robinson et al., 2003), whilst the other studies
utilised regular teachers to deliver the intervention; these had either received brief training or,
in the case of one study, were provided with curriculum manuals only (Sanigorski et al.,
2008).
Regarding the content of the intervention, a distinction could be made between the six studies
that targeted school-aged children (Chavarro et al., 2005; Harrison et al., 2006; Kipping et al.,
2008; Robinson, 1999; Robinson et al., 2003; Sanigorski et al., 2008), and the two studies
that were directed at pre-school children (Dennison et al., 2004; Fitzgibbon et al., 2005).
In the first group, five studies were classroom-based (Chavarro et al., 2005; Harrison et al.,
2006; Kipping et al., 2008; Robinson, 1999; Sanigorski et al.,2008). One was conducted in
the home setting (Robinson et al., 2003). Social cognitive theory was universally reported as
the intervention’s theoretical underpinning (Bandura, 1986)7. These studies also employed
similar strategies, as listed in Table 3-4. Conversely, one study reported very limited
information concerning the TV-reduction component (Kipping et al., 2008); however, it was
stated that the concept was adapted from a programme similar to Planet Health (Chavarro et
al., 2005). TV-reduction was the content of one lesson only, during which children were
encouraged to identify alternative leisure time activities.
7Social cognitive theory is characterized by the interaction of behaviour, personal factors and environmental influences.
According to this theory, the key constructs of behaviour change are observational learning, reinforcement, self-control and self-
efficacy (Bandura 1986).
27
Only one of the two studies targeted at pre-school children provided detailed information
about the intervention content (Dennison et al., 2004). The programme focused on a general
reduction in TV time, and placed no emphasis on self-management or budgeting. Moreover,
no theoretical framework was referred to in the paper. However, there were also two TV-
turnoff periods; one was in conjunction with the “National TV-Turnoff Week”. The major
messages of the intervention were reading to enhance literacy skills and family meal times
without a TV being switched on. These were directed at the children, as well as at parents and
pre-school or day-care staff. The only information provided in Fitzgibbon et al. (2005) was
that the theme of the three sessions was “Instead of TV, I could…”, and the fact that puppets
were used to drive the health promotion curriculum.
A common feature in six studies was that the parents were updated in the form of newsletters.
Diaries were used to monitor self-budgeting either by the children or by the parents
(Dennison et al., 2004; Harrison et al., 2006; Sanigorski et al., 2008). A time manager device
was used in two studies (Robinson, 1999; Robinson et al., 2003), which was provided free of
charge to families, and was locked onto the power plug of the TVs at home to help children in
‘budgeting’ through establishing individual viewing accounts.
In general, studies differed in the activities suggested to replace screen time, as this also
depended on the study objective. For example, since the intervention of Robinson (1999) was
the first experimental study concerned with testing the effect of promoting reduced TV time
on body composition, there was an interest to minimise other influences, and no replacement
activities were proposed. Conversely, Harrison et al. (2006) used a clear message of replacing
screen time with moderate physical activity as a major component of the intervention. The
other studies only stated examples such as parents reading books to children e.g. in Dennison
et al. (2004), carrying out a moderate physical activity of their choice (Chavarro et al., 2005;
Robinson et al., 2003), or creating artwork (Sanigorski et al., 2008).
Table 3-3: Overview of length and content of the interventions
Study
Intervention length
Number of education sessions
about the TV component (time
per session, where reported)
Robinson (1999)
6 months
18 (30-50 min.)
Robinson et al. (2003)
3 months
5-6
Dennison et al. (2004)
2 months
7 (20 min.)
Fitzgibbon et al. (2005)
4 months; TV component: 1 week
3 (40 min.)
Harrison et al. (2006)
4 months
10 (30 min.)
Chavarro et al. (2005)
2 years; TV component: 2 weeks
1 (45 min.)
Kipping et al. (2008)
5 months; TV component: 1 lesson
1
Sanigorski et al. (2008)
3 years; TV component: 2 weeks
Not reported
28
Table 3-4: Common strategies employed in interventions targeted at school-aged
children
Strategy
Description
Studies that included the strategy
Self-monitoring
Encouraging children to monitor and report
their time spent viewing TV, videotapes or
playing videogames to enhance motivation
for a change
Robinson (1999)
Robinson et al. (2003)
Chavarro et al. (2005)
Harrison et al. (2006)
Sanigorski et al. (2008)
A voluntary TV-turnoff
Asking Children to not view TV,
videotapes or videogames to enhance self-
efficacy for a certain amount of time to
enhance self-efficacy
Robinson et al. (2003) (two weeks)
Chavarro et al. (2005) (two weeks)
Robinson (1999) (10 days)
Sanigorski et al. (2008) (one week)
Harrison et al. (2006) (1 night)
Budgeting of viewing time
Encourage children to adhere to a weekly
viewing budget or use an activity point
system that rewards time spent with
physical activity (demerits points for screen
time)
Robinson (1999)
Robinson et al. (2003)
Harrison et al. (2006)
Selective viewing
Encourage children to become ‘intelligent
viewers’ by learning to use their screen
time selectively
Robinson (1999)
Robinson et al. (2003)
Harrison et al. (2006)
Sanigorski et al. (2008)
Becoming an advocate of a
life-style with less TV-
viewing
Encourage children themselves to actively
promote screen time reduction
Robinson (1999)
Harrison et al. (2006)
3.2.4 Outcomes
Outcomes presented and reviewed in this section were selected from the studies in accordance
with the hypothesised theory of change that suggests a change on cognitive measures, the
time spent watching TV or on other screen-related activities, dietary intake and/or physical
activity, and eventually the impact on body composition. An overview of the study outcomes
is provided in Table 3-5.
None of the studies’ assessed outcomes related to cognition, such as self-efficacy.8 The time
spent viewing TV or other screen-based activities decreased in all studies compared with
controls, except in one study, where TV time increased slightly but video game time
decreased (Sanigorski et al., 2008); however, statistically significant decreases were only
found in four studies (Chavarro et al., 2005; Dennison et al., 2004; Robinson, 1999; Robinson
et al., 2003). The effect magnitude for screen time reduction was difficult to compare between
studies owing to a difference in the ways in which this measurement was obtained and
expressed (i.e. previous day vs. average day; one day vs. > one day or whole week; including
or excluding weekend days; reported by children vs. reported by parents or both; including
video tape and computer games vs. TV time only).
8Two studies assessed cognitive outcomes that were not related to the TV intervention component (Harrison et al. 2006,
Robinson et al. 2003).
29
Change in dietary intake was evaluated in four studies (Fitzgibbon et al., 2005; Robinson
1999; Robinson et al., 2003; Sanigorski et al., 2008), although little impact was found with
respect to overall intake. However, the frequency of meals and snacks in front of the TV were
assessed in two studies, both of which reported a statistically significant reduction (Robinson,
1999; Robinson et al., 2003).
Outcomes with regard to physical activity were reported in all studies with the exception of
one (Dennison et al., 2004). Time undertaking moderate and vigorous physical activity
increased in two studies (Chavarro et al., 2005; Harrison et al., 2006); however, where
children’s fitness levels were assessed, no statistically significant change was found.
The mean BMI decreased in seven of the eight studies, although this was only statistically
significant in the case of three studies (Chavarro et al., 2005; Fitzgibbon et al., 2005;
Robinson, 1999).
Some studies noted challenges in assessing behavioural outcomes. Dennison et al. (2004), for
example, stated that the reporting of TV time was perhaps biased due to the pressure on
parents to display socially desirable behaviours. Moreover, participation in outdoor activities
was thought to be dependent on the seasons in one study (Harrison et al., 2006). Kipping et
al. (2008) noted that the questionnaire could have lacked the validity and sensitivity to detect
changes with respect to the short time frame of the evaluation. Furthermore, the authors in
one study (Fitzgibbon et al., 2005) mentioned incomplete assessment of dietary intake, since
24 hour recalls were no longer feasible at follow-up. In addition, their instrument for
measuring changes in physical activity was not validated.
30
Table 3-5: Summary of the outcomes assessed with statistically significant changes indicated in bold
Study
TV-viewing and other screen-
based activities
Dietary intake
Physical activity/Fitness
Body composition
Robinson
(1999)
CHILDREN
TV-viewing (hours/week):
−5.53 (−8.64 to −2.42)
Video tapes (hours/week):
−1.53 (−3.39 to 0.33)
Video games (hours/week):
−2.54 (−4.48 to −0.60)
PARENT REPORTS
TV-viewing (hours/week):
−4.29 (−5.89 to −2.70)
Video tapes (hours/week):
−0.25 (−1.19 to 0.69)
Video games viewing
(hours/week):
−0.76 (−1.75 to 0.22)
Overall household television use,
0-16 scale:
−0.77 (−1.69 to 0.14)
CHILDREN
Meals in front of TV (0-3 scale):
−0.54 (−0.98 to −0.12)
Snacking in front of the TV (1-3 scale):
−0.11 (−0.27 to 0.04)
Daily servings of high-fat foods:
−0.82 (−1.87 to 0.23)
Daily servings of highly advertised foods:
0.06 (−0.24 to 0.36)
PARENT REPORTS
Number of children’s meals in front of TV:
−1.07 (−1.96 to −0.18)
Percentage of children’s viewing when snacking
(%):
−1.94 (−9.06 to 5.17)
CHILDREN
Other sedentary behaviours, (hours/day):
−0.34 (−1.21 to 0.52)
Physical activity (MET, min./week):
−16.7 (−78.6 to 45.3)
20-m shuttle test (laps):
0.87 (−1.41 to 3.15)
PARENT REPORTS
Children’s physical activity (hours/week):
−2.00 (−4.58 to 0.59)
BMI (kg/m2):
−0.45 (−0.73 to −0.17)
Triceps skinfold (mm):
−1.47 (−2.41 to −0.54)
Waist circumference (cm):
−2.30 (−3.27 to −1.33)
Hip circumference (cm):
−0.27 (−1.08 to 0.53)
Waist-to-hip-ratio:
−0.02 (−0.03 to −0.01)
Robinson et
al. (2003)
TV, videotape and video games
(hours/week):
-4.96 (-11.41 to 1.49)
Total household TV use (0-4
scale):
−0.56 (−0.95 to −0.17)
Ate breakfast with the TV on (days/week):
-0.09 (-1.52 to 1.34)
Ate dinner with TV on (days/week):
-1.6 (-2.99 to -0.21)
Total dietary calorie intake/day:
84.3 (-201.5 to 370.1)
Percent of dietary kilocalories from fat (%):
-0.3 (-3.6 to 3.0)
Physical activity noon-6 PM (accelerometer
counts/min.):
55.1 (-115.6 to 225.8)
Moderate-to-vigorous physical activity noon-6
PM (average min.):
7.3 (-25.8, 40.4)
Self-reported moderate-to-vigorous physical
activity (previous day/min.):
9.2 (-11.2 to 29.6)
BMI (kg/m2):
-0.32 (-0.77 to 0.12)
Waist circumference (cm):
−0.63 (−1.92 to 0.67)
Dennison et
al. (2004)
PARENT REPORTS
TV/video viewing (hours/week):
−4.7 (−8.4 to −1.0)
% viewing ≥ 2 hours/day (%):
−21.5 (−42.5 to −0.5)
-
-
BMI (kg/m2) per year:
−0.36 (−1.22 to 0.50)
Standardised BMI, year:
−0.19 (−0.83 to 0.46)
Triceps skinfold (mm/year):
−0.41 (−3.52 to 2.70)
31
Study
TV-viewing and other screen-
based activities
Dietary intake
Physical activity/Fitness
Body composition
Chavarro et
al. (2005)
TV-viewing (hours/day):
−0.6 (−0.8 to −0.4)
!
Moderate and vigorous physical activity MET-
(hours/week):
3.1 (0.3 to 5.9)
BMI (kg/m2):
-0.3 (-0.5 to -0.1)
Triceps skinfold (mm):
-1.5 (-2.4 to 0.5)
Fitzgibbon et
al. (2005)
PARENT REPORTS
TV-viewing (hours/day):
-0.11 (-0.60 to 0.38)
PARENT REPORTS
Total fat (% kcal):
0.58 (-2.00 to 3.16)
Saturated fat acid (% kcal):
0.27 (-0.89 to 1.43)
Fibre (g/1000 kcal):
-0.58 (-1.61 to 0.44)
PARENT REPORTS
Exercise frequency (% ≥7 x /week):
0.79 (-15.97 to 17.55)
Exercise intensity (Borg scale):
-0.62 (-1.77 to 0.53)
Adjusted BMI (kg/m2):
-0.54 (-0.98 to -0.10)
Adjusted BMI z-score:
-0.18 (-0.31 to -0.04)
Harrison et
al. (2006)
Screen time (30 min. blocks /day):
−0.41 (−0.93 to 0.12)
-
Moderate and vigorous physical activity (30 min.
blocks/day):
0.84 (0.11 to 1.57)
Aerobic fitness (laps):
1.7 (−3.5 to 6.9)
BMI (kg/m2):
−0.08 (−0.38 to 0.22)
Kipping et
al. (2008)
Screen viewing weekdays (min.):
-11.6 (-42.7 to 19.4)
Screen viewing Saturdays (min.):
-15.4 (-57.5 to 26.8)
-
Walks/cycles to and from school (odds ratio):
0.45 (0.27 to 0.75)
BMI (kg/m2):
0.10 (-0.27 to 0.46)
Obesity prevalence (odds
ratio):
0.68 (0.24 to 1.94)
Sanigorski et
al. (2008)
Data only presented graphically.
PARENT REPORTS
TV-viewing (previous day):
5 min. increase for intervention, 20
min. decrease for controls
Video game time (previous day):
<5 min. decrease for intervention,
10 min. increase for controls
Data only presented graphically.
PARENT REPORTS
Significant increase in fruit intake at home
Data only presented graphically.
PARENT REPORTS
Non-significant increase in time playing outside
after school.
Body weight (Kg):
-0.92 (-1.74 to -0.11)
Waist circumference (cm):
-3.14 (-5.07 to -1.22)
BMI (kg/m2):
-0.28 (-0.7 to 0.15)
Waist/height:
-0.02 (-0.03 to -0.004)
BMI z-score:
-0.11 (-0.21 to -0.01)
32
3.2.5 Implementation and compliance
The studies varied in the degree to which process-related information was reported.
Information regarding compliance with the TV component was provided in only three studies.
One study (Robinson, 1999) stated that 67% completed the full TV-turnoff period, and 90%
of the children participated on at least some of the days; however, only 55% of children
regularly participated in budgeting periods. In Robinson et al. (2003), 82% of the families
participated in all the sessions that were offered to them, and 61% of the parents read three or
more of the newsletters, who accordingly gave uniformly positive feedback about the
intervention. Sanigorski et al. (2008) also found that all schools participated in the Power-
Down Week, although distribution of resources to parents failed to be implemented due to
time constraints. Nevertheless, 70% of parents surveyed were aware of the programme and its
main messages. Furthermore, based on the conclusion in that study, it could be assumed that
teachers found it easy to incorporate the programme within the curriculum. Conversely, one
study stated that, although enthusiastic about the programme, teachers were unable to deliver
all lessons of the health curriculum (Kipping et al., 2008), with the same was reported in
Chavarro et al. (2005), although the extent to which the TV component was affected was not
clear in both studies.
In addition, the authors of one study (Harrison et al. 2006) also acknowledged that easy
access to TV-viewing might have been a general barrier to the intervention, especially where
children had a TV in their own bedroom.
3.3 Relationship within and between the studies
3.3.1 General characteristics and study context
The eight studies reviewed were similar with respect to the general context of the intervention
as they were almost all conducted in either the school or the day-care setting. All of them
aimed to induce a reduction in TV-viewing through education, meaning they primarily sought
to impact children’s cognitive ability to reduce screen time. In addition, the use of campaigns
to promote TV turn-off periods, as well as the time manager devices, represented elements of
environmental modifications. However, in general, changes to the environment were rarely
seen. For example, the interventions did not address aspects such as the elimination of TVs
from bedrooms.
Most studies noted a form of parental involvement in the intervention: whilst the stimulus for
behaviour change was set in the sessions through education and reflection, the success of the
33
intervention would have been crucially determined by what transpired at home. Therefore,
although no study examined this issue closely, it is plausible to suggest that the involvement
of parents was vital through their monitoring of screen time, organising family meals without
TV, or participating in TV-turn-off periods.
3.3.2 Intervention content
A general difference between the studies was the intensity by which the TV-reduction
component was delivered. Similarly, both the number and the characteristics of other
components that were part of the intervention differed widely.
In most studies, emphasis was placed on activities to enhance self-efficacy i.e. self-
monitoring and voluntary TV-turn-off periods, even though these varied in length and reach
of promotion. It would be plausible to assume that, in studies that adopted such turn-off
periods, this may have also been an awareness raiser of the message to reduce TV-viewing.
Strategies based on social cognitive theory represented a major commonality of the six studies
targeted at school-aged children. Conversely, this was less clear in the two studies for
younger children, where no such methodological underpinning was stated (Dennison et al.,
2004; Fitzgibbon et al., 2005). In addition, where reported, the strategies used to replace
screen time appeared to vary widely. A message to decrease meals in front of the TV was
only included in one study (Dennison et al., 2004), even though this was assessed as an
outcome in two other studies (Robinson, 1999; Robinson et al., 2003).
3.3.3 Outcomes
Given the educational nature of the interventions—and particularly the theoretical
underpinning—it was surprising that none of the studies assessed cognitive outcomes. With
respect to TV-viewing or other screen-related activities, the following observation could be
made: all four studies with statistically significant reductions in viewing time included a turn-
off period of at least one week, whereas this was not the case for three of the other studies,
without significant changes experienced in terms of the outcome. However, the effect
magnitude on viewing time was generally difficult to compare.
The role of other secondary outcomes on behaviour change for a mediating role was less
clear: no single study found statistically significant changes across all the variables assessed
that would indicate an effect mechanism along any of the proposed pathways. Vigorous
physical activity did probably not increase given that no improvement in fitness levels was
found in studies that measured this outcome. It could not be ruled out that moderate physical
34
activity increased and that sedentary behaviour was reduced, which is a likely consequence of
reduced TV-viewing. However, this was not directly measured in the studies.
With regard to dietary intake, only two studies assessed this in relation to TV-viewing,
although there was an indication for a lower frequency of meals in front of the TV.
Generally, it appeared that studies focused more attention to measuring physical activity
variables rather than dietary intake. In addition, those variables measured were sometimes
unlikely to be linked to a reduction in screen time, such as active transport to and from school
(Kipping et al., 2008), for example. The role of a reduced exposure to advertisement was not
assessed, and thus remained unclear.
Body composition outcomes were astoundingly coherent, despite the apparent difference in
age, baseline risk status, and socio-economic background across the populations studied.
However, it cannot be ruled out that those studies comprising multiple components were
subject to effect interactions resulting from other intervention activities unrelated to the
reduction of TV-viewing. For example, the physical activity components in some of the
studies are likely to have contributed to the impact on body composition.
3.3.4 Implementation and compliance
Unfortunately, it was difficult to compare implementation and compliance as limited
information was given; however, the few studies that reported on this suggested that
participation and awareness appeared sufficient. The degree to which schools could
accommodate the intervention in the routine curriculum was less clear.
3.4 Robustness of the synthesis
The robustness of this synthesis was assessed with reference to the comparability of studies.
Furthermore, this required a reflection on the synthesis process and on the limitations in
single studies.
This research synthesis found major commonalities between the eight studies reviewed. One
main overlap was that the interventions comprised a series of educational sessions delivered
in schools and in pre-schools, with the exception of one home-based study. The content of the
interventions was very similar, particularly in the group of studies targeting school-aged
children. The strategies employed were all based on social cognitive theory with the ultimate
aim to teach children to use screen time selectively. Studies differed with respect to the
intervention dose of the education sessions. Comparison was further made difficult where
studies provided little detail on the intervention content; however, based on the broadly
35
consistent intervention patterns across the eight studies, it remained plausible that the effect
on body composition may have been induced through a similar mechanism. Nevertheless, the
exact pathways could not be identified when analysing the outcomes in the eight studies.
Reduction in TV-viewing time was found to be statistically significant only in half of the
studies, and no general statement could be derived regarding the effect mediation through
increased physical activity levels and/or lower dietary intake. The pathway was further
obscured by the relative complexity of the programmes, i.e. the number of components not
related to TV-viewing. The influence of these components on body composition remains
inherently difficult to disentangle. Markedly, had the studies generally assessed mediating
variables more closely in relation to TV-viewing, it is possible that this could have helped to
shed light on this issue.
A limitation in this synthesis was that no distinction could be made between TV-viewing time
and other screen-based activities, such as video and computer games. The studies used
different definitions for screen time, and so, for practical reasons, these were handled as one
screen category in the present study. Furthermore, it was also not clear in some studies
whether or not reduced TV-viewing was compensated by other screen activities such as
playing computer games.
Several other shortcomings in the studies limited the robustness of findings. The small sample
sizes in most studies limited the ability to detect statistically significant changes. Furthermore,
all studies stated difficulties in terms of assessing behavioural outcomes, which was
particularly relevant where self-reporting was utilised, e.g. through error-prone recalls of TV-
viewing time. Social desirability biases for TV-viewing time may have also occurred when
obtaining reports from parents. A further limitation in many studies was the short duration
between experiment completion and follow-up, which leaves doubt concerning whether the
mechanism of behaviour change and the translation into body composition could have fully
come into play.
3.5 Research implications
A number of implications evolved from this narrative synthesis, all of which are relevant for
future interventions to reduce TV-viewing in children.
The interventions in this synthesis were limited to classroom education and messages to take
home, whilst children’s broader environment, including access to TVs, remained almost
unchanged. In one study, the authors concluded the following: “Reducing TV-viewing will be
a real health promotion challenge because of low levels of awareness of the problem and
36
difficulty in influencing an activity that takes place inside homes” (Sanigorski et al., 2008).9
Therefore, one could argue that more emphasis should be placed on the home setting. The
home-based counselling approach taken in Robinson et al. (2003) was suitable for girls with a
particular risk profile, but is possibly too resource-intense to serve as a population-based
approach. Future interventions aiming to reduce TV-viewing in children should be innovative
in considering the home environment, and particularly involving the parents.
Furthermore, the measurement of behavioural outcomes needs to be improved; this is
particularly the case for the validity of self-reported TV-viewing time. The combination of
self-reported and objective measures has been suggested as optimal for assessing sedentary
behaviour (Biddle et al., 2011). In addition, the outcome TV-reduction time should be
expressed in a format that standardises the screen types as well as the time slots, thus enabling
this to be compared between studies.
In order to enhance the understanding of the intervention mechanism, future studies should
consider assessing more outcomes able to explain the effect on body composition; therefore,
such behavioural change outcomes that are specifically related to a reduction in TV-viewing
time should be measured. Furthermore, assessing dietary intake over a whole day may not
suffice to explain the impact of the intervention; rather, an outcome such as ‘snacking whilst
viewing’ may be more appropriate. In addition, the nature of the activities displacing TV-
viewing require a deeper understanding; this implies that outcomes on compensatory
behaviour need to be assessed, as well as the level of physical activity. Future interventions
should also address the fact that children are increasingly gaining access to a wide range of
screen-based communication technologies (Maniccia et al., 2011; Schmidt et al., 2012). This
may require different strategies than those solely aimed at the reduction of TV-viewing.
3.6 Conclusion
The narrative synthesis found that the studies followed similar intervention logic when
striving to reduce TV-viewing in children; however, a generalisable intervention mechanism
could not be confirmed; therefore, uncertainty remains concerning whether or not the
statistically significant decrease in BMI actually resulted from the TV-reduction component
of the intervention. With this in mind, future programmes should focus more on modifications
in the home environment, and their evaluations should add to the understanding of effect
mechanisms, as well as the assessment of the intervention’s long-term sustainability.
9This was quoted in a supplementary report (WHO Collaborating Centre for Obesity Prevention 2008), page 46.
37
4 Critical Appraisal of Economic Evaluations of Obesity
Prevention Interventions
4.1 Objective
In addition to the syntheses in the previous two chapters, a thorough review concerning the
prevention of overweight and obesity also warrants the consideration of evidence generated
by economic evaluations (Rychetnik et al., 2002). First of all, this provides an overview
concerning the current evidence for the economic efficiency of prevention programmes in
children. Furthermore, such a review helps to gain understanding concerning the most
suitable evaluation approaches, but also outlines pertinent methodological issues. In addition,
the generalisability of economic findings can be assessed; thus, relevant information can be
collated that may be used in the economic evaluation model for South Australia in Part II of
this thesis.
The present study refers to economic evaluations in accordance with the common definitions
(Drummond et al., 2005; Gold et al., 1996): Cost-effectiveness analysis (CEA), where
outcomes are expressed as clinical effect measures; cost-utility analysis (CUA), where
outcomes are expressed as a utility measure; and cost-benefit analysis (CBA), where
outcomes are expressed in monetary terms. However, the term ‘cost-effective’ was used more
broadly to describe efficiency in intervention, regardless of the type of analyses involved.
The aim of this chapter was to critically appraise studies that evaluated the cost-effectiveness
of obesity prevention programmes in the context of children.
4.2 Methods
Studies for this appraisal were identified through the same literature search that was
conducted in Chapter 2. All records screened for inclusion in the meta-analysis were searched
for economic content in the primary studies, or online databases were checked for related
publications; this included studies published up to May 2009. A crosscheck of recently
published systematic reviews with a similar research question showed that these authors had
identified no other eligible papers for this period (John et al., 2012; Konig et al., 2011).
Eligible studies to be included in the critical appraisal had to be full economic evaluations, i.e.
they comprised costs and consequences. The eligible interventions were primary prevention
programmes targeting children. Modelling studies were not considered.
38
The critical appraisal was conducted according to the checklist suggested by the Centre of
Reviews and Dissemination (CRD), which was based on an established guideline (CRD,
2008; Drummond & Jefferson, 1996). This questionnaire comprises 36 items that could be
answered as ‘Yes’, ‘No’, ‘Can’t Tell’, or ‘Not Appropriate’.
The result of the appraisal was presented in tables, although the studies were also examined in
detail according to five main categories, namely method of economic evaluations,
intervention costs, consequences, study results, and comparability and transferability. Where
relevant, reference was made to the items of the appraisal checklist.
4.3 Results
4.3.1 Literature search
Three economic evaluations were found where the primary study was part of the meta-
analysis:
• Planet Health (Wang et al., 2003)
• MCG FitKid (Wang et al., 2008)
• APPLE (McAuley et al., 2010).
One additional economic evaluation was found in a primary study that was excluded from the
meta-analysis due to a lack of suitable outcome data:
• CATCH (Brown et al., 2007a).
For convenience, studies will henceforth be referred to by their programme name.
4.3.2 General study characteristics
The main characteristics of the four studies are briefly described here, and are also
summarised in Table 4-1.
Three primary studies were RCTs conducted in the USA, but the APPLE study was a non-
randomised controlled trial from New Zealand. Study length ranged between one and two
years. The age of participants was between 7 and 12 years. CATCH and MCG FitKid had high
proportions of minority groups. The economic evaluation in Planet Health was limited to the
subgroup of female participants only. The reduction in prevalence of overweight was not
statistically significant for the boys in the intervention (Gortmaker et al., 1999).
39
All four studies evaluated interventions delivered in the school setting. In addition, the
APPLE programme also involved activities in the wider school community. APPLE and
CATCH could be described as multi-faceted interventions. Conversely, Planet Health only
consisted of education, although this was an interdisciplinary curriculum that involved all
teachers. MCG FitKid was an after-school programme that included physical activity sessions
and a healthy snack break. The control arms did not receive treatment in any of the studies.
Table 4-1: Summary of main study characteristics
Study,
Country
Planet Health, USA
CATCH, USA
MCG FitKid, USA
APPLE, New
Zealand
Design
RCT
RCT
RCT
Controlled trial
Intervention
School-based,
interdisciplinary health
curriculum
Length: Two years
School-based, multi-
faceted health
promotion (nutrition,
physical activity,
health curriculum)
Length: Two years
After-school care
sessions (physical
activity, nutrition)
Length: First year
results, total length was
three years
School-based and
community-based,
health promotion
(physical activity)
Length: Two years
Participantsa
School children,
11.7 years
Approx. 70% white
ethnicity
nI= 1,203
School children
8.3 years
93% Hispanic
nI= 423
School children
8.7 years
Approx. 60% black
ethnicity
nI= 312
School children,
7.7 years
Approx. 82% white
ethnicity
nI= 279
Type of
economic
evaluation
CUA and CBA
CUA and CBA
CEA
CEA
aMean age and ethnic composition at baseline.
nI: Number of participants for whom programme costs were estimated.
4.3.3 Results of the appraisal
Table 4-2 lists all 36 appraisal questions and the answers given in each study. Table 4-3
provides an overview of the appraisal results.
In all four studies, the vast majority of the questions could be answered positively. More than
30 items were answered with ‘Yes’ in the studies on Planet Health and CATCH.
A small number of items were ‘Not Appropriate’ for MCG FitKid and APPLE, which was
mainly due to the absence of effect valuation and modelling techniques in these two studies.
The APPLE study delivered the most negative answers. Five items were answered with ‘No’,
with all issues that arose surrounding items answered with ‘No’ or ‘Can’t Tell’ examined
more thoroughly in the next section.
Table 4-2: Results of the critical appraisal
Appraisal items according to CRD 2008, Drummond and Jefferson 1996
Planet
Health
CATCH
MCG
FitKid
APPLE
1. Was the research question stated?
YES
YES
YES
YES
2. Was the economic importance of the research question stated?
YES
YES
YES
YES
3. Was/were the viewpoint(s) of the analysis clearly stated and justified?
YES
YES
YES
YES
4. Was a rationale reported for the choice of the alternative programmes
or interventions compared?
YES
YES
YES
YES
40
Appraisal items according to CRD 2008, Drummond and Jefferson 1996
Planet
Health
CATCH
MCG
FitKid
APPLE
5. Were the alternatives being compared clearly described?
YES
YES
YES
NO
6. Was the form of economic evaluation stated?
YES
YES
YES
YES
7. Was the choice of form of economic evaluation justified in relation to
the questions addressed?
YES
YES
YES
YES
8. Was/were the source(s) of effectiveness estimates used stated?
YES
YES
YES
YES
9. Were details of the design and results of the effectiveness study given
(if based on a single study)?
YES
YES
YES
YES
10. Were details of the methods of synthesis or meta-analysis of estimates
given (if based on an overview of several effectiveness studies)?
N. A.
N. A.
N. A.
N. A.
11. Were the primary outcome measure(s) for the economic evaluation
clearly stated?
YES
YES
YES
YES
12. Were the methods used to value health states and other benefits
stated?
YES
YES
N. A.
YES
13. Were the details of the subjects from whom valuations were obtained
given?
YES
YES
N. A.
NO
14. Were productivity changes (if included) reported separately?
YES
YES
N. A.
N. A.
15. Was the relevance of productivity changes to the study question
discussed?
YES
YES
N. A.
N. A.
16. Were quantities of resources reported separately from their unit
cost?
YES
YES
YES
YES
17. Were the methods for the estimation of quantities and unit costs
described?
YES
YES
YES
YES
18. Were currency and price data recorded?
YES
YES
YES
YES
19. Were details of price adjustments for inflation or currency
conversion given?
NO
NO
YES
YES
20. Were details of any model used given?
YES
YES
N. A.
N. A.
21. Was there a justification for the choice of model used and the key
parameters on which it was based?
YES
YES
N. A.
N. A.
22. Was time horizon of cost and benefits stated?
YES
YES
YES
YES
23. Was the discount rate stated?
YES
YES
N. A.
YES
24. Was the choice of rate justified?
NO
YES
N. A.
NO
25. Was an explanation given if cost or benefits were not discounted?
N. A.
N. A.
N. A.
NO
26. Were the details of statistical test(s) and confidence intervals given for
stochastic data?
YES
YES
YES
YES
27. Was the approach to sensitivity analysis described?
YES
YES
YES
YES
28. Was the choice of variables for sensitivity analysis justified?
NO
NO
NO
NO
29. Were the ranges over which the parameters were varied stated?
YES
YES
YES
YES
30. Were relevant alternatives compared? (i.e. Were appropriate
comparisons made when conducting the incremental analysis?)
YES
YES
YES
CAN’T
TELL
31. Was an incremental analysis reported?
YES
YES
YES
CAN’T
TELL
32. Were major outcomes presented in a disaggregated as well as
aggregated form?
YES
YES
YES
YES
33. Was the answer to the study question given?
YES
YES
YES
YES
34. Did conclusions follow from the data reported?
YES
YES
YES
YES
35. Were conclusions accompanied by the appropriate caveats?
YES
YES
YES
YES
36. Were generalisability issues addressed?
YES
YES
YES
YES
N.A.: NOT APPROPRIATE
Table 4-3: Summary of answers given in the critical appraisal
Possible answer
Planet
Health
CATCH
MCG
FitKid
APPLE
YES
31
32
25
24
NO
3
2
1
5
CAN'T TELL
0
0
0
2
NOT APPROPRIATE
2
2
10
5
41
4.3.4 Detailed examination of the appraised studies
Economic evaluation method
All four studies based the economic evaluation on the societal perspective. MCG Fit KID and
APPLE were CEA, whilst the other two studies assessed the overall cost-utility and cost-
benefit of Planet Health and CATCH.
The school-based education programme Planet Health was the first intervention concerning
obesity prevention for which an economic evaluation was conducted. A lifetime projection
model was used to estimate the impacts of the reduced prevalence of overweight in adolescent
girls on health-related cost offsets and Quality-adjusted Life Years (QALYs) saved during
adulthood. Furthermore, the authors of the CATCH study replicated this projection model in
order to facilitate comparison with Planet Health. Both studies also included a CBA based on
a human capital approach.
The two other studies did not project long-term consequences. The time horizon in the CEA
of the after-school care programme MCG FitKid was restricted to one year, with effects only
expressed as %BF.
In the evaluation of the community-controlled trial APPLE, health-related quality of life was
approximated through interviewing parents on behalf of their children. In the absence of a
statistically significant difference in the Health Utility Index (HUI) score, the authors chose to
present the cost per kg of weight-gain prevented per child per year. The critical appraisal
revealed that no detail was provided concerning the activities in the comparator (Item 5). It
appeared that the comparator was only used to determine net effects in the primary studies,
but no incremental approach was undertaken to estimate costs (Item 30 and Item 31). In
actual fact, the cost calculation was described as marginal rather than incremental (McAuley
et al., 2010).
In addition, no justification was provided for the choice of discount rate in the studies on
Planet Health and APPLE (Item 24), although these appeared to be within a common range
(3% and 5%, respectively). Furthermore, the authors in the APPLE study chose to report the
non-discounted costs per year as the main finding as well as non-discounted health effects,
although these were obtained from a two-year follow-up after intervention ending (Item 25).
The reasons for that were not further explained.
Intervention costs
Resource use reporting was generally transparent in all four studies. They all referred to the
number of participants at baseline to estimate the intervention costs. These were reported in a
42
disaggregated manner, outlining each cost component and separating quantities and unit
prices.
The most detailed and comprehensive description of costs was provided in the evaluation of
APPLE. Nevertheless, not all programme content was included in the calculation. For
example, a free fruit distribution programme was mentioned in the intervention description,
but was not listed as a cost item.
Wang et al. acknowledged that intervention costs in Planet Health were estimated
retrospectively; hence, there remains some uncertainty concerning the accuracy of the
estimates derived (Wang et al., 2003). Furthermore, there was no mention in the CATCH
study of when resource data were collected. Given the lack of detail in reporting, this may
have also been a retrospective analysis. For example, the cost calculation only included the
training of personnel, but there was no inclusion of materials.
Only the MCG FitKid study considered the costs in the comparator. This was the rate of usual
after-school care that would have been inflicted on parents without the after-school
programme being in place. However, the authors stated that it was difficult to obtain a
generalisable estimate for parents’ time, and therefore assumed that this would equal the
monetary value of caring for sick relatives (Wang et al., 2008).
In all four studies, the costs occurred mainly in the education sector, with personnel-related
resources recognised as having the highest share of programme costs. In Planet Health and
CATCH, this was mainly the time that teachers and other school staff spent in training
sessions. Conversely, these personnel costs in MCG FitKid and APPLE were mainly due to
the employment of new staff to deliver the intervention.
Intervention consequences
A distinction can be made between two types of intervention consequences; these were either
health outcomes, expressed in terms of clinical effects, or quality-of-life measures. The other
possible consequence was offsets in healthcare costs as a result of the intervention.
The four primary studies reported statistically significant reduction in at least one of the
obesity indicators assessed (see Table 4-4). The types of clinical effects reported differed
between all four evaluations. The body composition outcome in Planet Health and CATCH
was the change in prevalence of overweight and/or obesity, although the definitions for
weight status also differed between these two studies. The outcome in the evaluation of MCG
FitKid was percentage of %BF measured in those students who attended more than 40% of
the sessions. In the primary study on the APPLE programme, the BMI z-score was the main
43
outcome, although this was translated into prevented weight gain per child per year in the
economic evaluation.
Health-related quality of life was included as an outcome in the evaluations of Planet Health
and CATCH. This was estimated based on the projected BMI in adulthood through data from
the National Health Interview Survey.
In order to project the long-term consequences of Planet Health and CATCH, a number of
estimates had to be derived. Weight status following intervention ending could not be linked
directly with the assumed impact on morbidity during adulthood as there were no data for this
purpose; therefore, in both studies, the probability of participants being overweight in the age
group 21–29 was derived first based on available data (Whitaker et al., 1997). Based on
estimated weight status in young adulthood, the probabilities of being overweight for Planet
Health, or being obese for CATCH respectively, at the age of 40, were derived.10 In addition,
the life expectancy of those aged 40–65 was estimated according to weight status.
Conversely, children’s HUI scores in the APPLE programme were already obtained during
the evaluation, although this was done through a proxy by interviewing the parents. As no
statistically significant difference in these scores was detected, the authors suggested that the
instrument used to obtain the HUI scores was perhaps not specific enough for obesity
prevention. Furthermore, the relatively small differences in weight after the intervention were
considered unlikely to be able to provoke noticeable impacts on health-related quality of life
in the short-run. Nevertheless, uncertainty remained in the study concerning the potential bias
in the interviews, as no further detail was provided about the parents (Item 13). Moreover,
there was also no effect valuation in the CEA of MCG FitKid; Wang et al. (2003) stated that
this was due to limited research published on the translation of %BF into a measure for
quality of life.
The cost offsets induced by the intervention differed between the two studies, where this was
included. Only the studies on Planet Health and CATCH included the averted excess medical
costs due to reduced morbidity during adulthood, but these were based on different sources in
each study, subsequently resulting in different diseases being considered. In both studies, little
detail was provided on the price adjustment of excess medical costs averted. Wang et al.
10This was done through linking National Health and Nutrition Examination Survey I (NHANES) and NHANES I Epidemiologic
Follow-up Study.
44
(2003) at least stated that these were adjusted for inflation, but no further information was
given about the rate used (Item 19).
The studies on Planet Health and CATCH also included estimates for averted loss of labour
time during adulthood as a result of the intervention. However, this was only incorporated
into the CBA, which was conducted as a separate analysis in each of the two studies.
Other consequences of the interventions were not included in any of the studies, although
there could have been a broader consideration at least in the APPLE study. For example,
Taylor et al. (2007) mentioned the ripple effects of the APPLE community programme, such
as modifications to school food and physical activity policies. With this in mind, it is
therefore conceivable that such intervention-related consequences may have also had an
impact on resource use in schools and households, and thus would need to be included in an
economic evaluation undertaken from a societal perspective.
Sensitivity analysis
In general, sensitivity analysis and its impact on results were kept relatively simple in all four
studies. The number of parameters chosen to vary was very small in the two CEAs (McAuley
et al., 2010; Wang et al., 2008), but were more comprehensive in the two modelling studies
Planet Health and CATCH. Nevertheless, there was no inclusion of the possible relapse of the
intervention effect after follow-up, which was the most crucial underlying assumption of the
projection models in these two studies (Item 28).
In the evaluations of Planet Health and CATCH, sensitivity analysis was conducted with the
objective to address parameter uncertainty in a multivariate and univariate way. This was
done through Monte Carlo simulation, with the range of variation for each parameter set by
the respective 95% CI. Furthermore, the discount rate was varied. Two additional scenarios
were tested for CATCH: one was the replacement of key parameters with data that were
specific to Hispanic populations, whilst the other applied the same medical costs averted as in
Planet Health in order to facilitate better comparison. The evaluation of Planet Health was
the only study where intervention costs were varied in the sensitivity analysis.
In the evaluation of MCG FitKid, only the per capita usual after-school care costs were varied
as the unit price of the ‘no intervention’ comparator. The sensitivity analysis in the evaluation
of APPLE was also kept very brief. The effect on weight according to the range within the
95% CI of the BMI z-score was presented for all age groups to demonstrate different levels of
effectiveness.
45
Study findings
Table 4-4 provides an overview of the study results. Planet Health saved 4.1 QALY and
averted future medical expenses of US $15,887 at the cost of US $33,67711, resulting in an
incremental ratio of US $4,305 per QALY. This ratio was most sensitive to discount rate
variation. The CI of the incremental cost-effectiveness ratio (ICER) generated through
multivariate sensitivity analysis was wide, but still remained below the cost-effectiveness
threshold applicable to the study context (US $1,612 to US $9,010; mean value $4,397).12 In
the CBA, a net benefit of US $7,313 was estimated for the intervention.
CATCH saved 8.55 QALY and saved US $36,34813 future medical expenses at the cost of
US $44,039 (US $900 per QALY). This remained almost unchanged when applying Hispanic
parameters (US $903 per QALY); however, when employing the cost offsets used in Planet
Health, the ICER decreased to US $0 per QALY. The CBA resulted in a saving of US
$68,125 to society.
MCG FitKid reduced %BF by 0.76% points given a participation of at least 40% of the
sessions at total costs of US $174,070.14 Net intervention costs per student were US $317. In
the sensitivity analysis, these costs varied widely (US $98 to US $527), which depended on
the value employed for usual after-school care as the opportunity cost of the intervention.
Total costs in APPLE were NZ $357,49015 and NZ $1,281 per participant, who would weigh
two kg less at age 15. According to the sensitivity analysis, the effect on weight varied within
a range of 0.5kg and 2.4kg, depending on participant’s age and sex.
Table 4-4: Overview of results in the economic evaluations
Study
Planet Health, USA
CATCH, USA
MCG FitKid, USA
APPLE, New
Zealand
Outcome
measure in
primary study
Reduction in prevalence of
obesity defined by
composited indicator of
BMI and TSF above or at
85th percentile
Reduction in
prevalence of obesity
defined by 85th and 91st
percentile
Reduction in % BF
(for students
attending >40% of
sessions)
Reduction in BMI z-
score
Outcome
measure in
economic
evaluation
CUA: QALYs saved and
medical costs averted
based on a reduction in
overweight prevalence in
female participants at age
40-65
CBA: Averted medical
productivity losses
CUA: QALYs saved
and medical costs
averted based on
reduction in obesity
prevalence at age 40–
65
CBA: Averted medical
productivity losses
Reduction in % BF
(for students
attending >40% of
the sessions)
Reduction in weight
(kg)
11US $ in 1996
12The authors refer to an implicit cost-effectiveness threshold of US $30.000 (Wang et al. 2003).
13US $ in 2004
14US $ in 2003
15NZ $ in 2006
46
Study
Planet Health, USA
CATCH, USA
MCG FitKid, USA
APPLE, New
Zealand
Discount rate
Costs and Consequences at
3%
Costs and
Consequences at 3%
Not applicable
(Costs at 5%a)
Valuation of
effect
Yes
Yes
No
(Yesb)
Extrapolation
Yes
Yes
No
No
Results
CUA: US $4305/QALY
CBA: US $7313
($ in 1996)
CUA: US $900/QALY
CBA: US $68,125
($ in 2004)
US $317/(-0.76%
reduction in % BF)
($ in 2003)
NZ $357,490
($ in 2006) for 279
children in the
intervention
aUndiscounted estimates were presented as the main finding.
bNo significant difference was detected in the HUI-score between baseline and follow-up, hence a CEA was
conducted instead.
Comparability and transferability of results
The generalisability of the findings was discussed for Planet Health, MCG FitKid and
APPLE, with the conclusion drawn that this would be theoretically feasible with some caveats
(Item 36). In the evaluation of CATCH, no explicit statement was made in this respect;
however, since two scenarios were analysed—one with parameters for Hispanics and one for
white Americans—transferability was implicitly considered in that CUA. Authors of all four
studies made reference to alternative programmes in the discussion as far as this was
applicable, but were careful about direct comparison.
In general, the ability to compare studies and their results was limited by different evaluation
methods and outcome metrics. There appeared to be a myriad of determinants with impact on
the results across all four studies, which can be demonstrated by the following example: in the
CUA of Planet Health, a reduction in obesity prevalence was only considered to occur in
those participants with outcomes measured at follow-up, whilst the authors of CATCH
assumed the effects to apply to all participants at baseline—regardless of whether or not they
had dropped out during the intervention course. Consequently, had Brown et al. (2007a) taken
the same conservative approach as Wang et al. (2003), this would have lowered the averted
number of cases of adulthood obesity in CATCH from N= 14.93 to N= 12.29. This, in turn,
would have almost doubled the ICER of CATCH, although the authors did not further expand
on this.
4.4 Discussion
4.4.1 Economic evidence
This appraisal of economic evaluation studies about programmes aimed towards preventing
overweight and obesity in children suggests that school-based interventions may be cost-
effective when considering projected long-term consequences into adulthood. Value for
47
money may also be found in those programmes, which were evaluated based on intermediate
clinical effect outcomes, i.e. prevented weight gain in kg or reductions in %BF; however,
there was no cost-effectiveness threshold, which could have been used as a reference for these
two studies.
In general, it appeared that loss to follow-up, low attendance, and low levels of effectiveness
had a strong influence on the results in the studies; however, it was difficult to derive other
general conclusions surrounding the evidence for cost-effectiveness, as the approaches to
economic evaluation varied markedly between the studies.
The evaluations of Planet Health and CATCH represented the most sophisticated approach
through the projection of health and productivity consequences into adulthood. Besides
various gaps with respect to data availability, these studies demonstrated that it is feasible to
extrapolate prevention effects during childhood into later phases in life, when people are most
prone to developing obesity-related diseases. Nevertheless, this had to be based on strong
assumptions, whilst the overall impact of the interventions remained rather modest. For
example, only N= 5.805 cases of adulthood overweight were prevented in Planet Health out
of the N= 310 female participants in the intervention.
4.4.2 Methodological issues
The checklist appraisal of studies revealed that the four studies were generally in accordance
with the standards of conducting economic evaluation and reporting the findings. The small
number of questionnaire items with negative or unclear answers arose mainly due to a lack of
detailed descriptions or justifications provided in the articles; notably, however, these were
not due to major methodological discrepancies. Nevertheless, although the checklist proved to
be a suitable vehicle for guiding a thorough appraisal of the four studies, major issues in this
critical review evolved beyond the questionnaire items.
First of all, there appeared to be no standard in reporting primary outcomes for the evaluation
of interventions to prevent overweight and obesity; therefore, economic evaluations were
conducted on a range of different measures for effectiveness. Where these were not
transformed into quality of life, they commonly provided very little informative value to
decision makers, such as the crude change in %BF or prevented weight gain. On the other
hand, the valuation of outcomes may not be feasible for all effect measures, such as %BF
(Wang et al., 2008). Furthermore, the experience in the APPLE study suggests that direct
measurement of quality of life in children may have its own challenges; it may either be
difficult to obtain valid measurements, or alternatively, quality of life may not be strongly
affected in a short-term evaluation period. More research is clearly needed in this area, which
48
may require instruments to measure health-related quality of life that are specifically tailored
to characteristics of overweight and obesity in paediatric populations (Belfort et al., 2011).
It follows logic that, as obesity prevention in children represents a risk factor reduction, a
long-term perspective on the consequences should be the gold standard. Projecting outcomes
to adulthood made the evaluations of Planet Health and CATCH thus appear to be the
superior form of evaluation. Importantly, the authors of these two studies aimed to
approximate the true account of relevant consequences, thereby increasing the meaning of the
ICER. However, no direct projection up until age 40 could be made in these studies due to the
absence of directly linkable data, and the consequences that occurred during the earlier stages
of adulthood or even childhood were not considered. Most important to note here is that long-
term effectiveness of prevention remains unknown and could only be assumed. In total, the
sum of analytic modelling steps required bear considerable uncertainty on the overall findings
in Planet Health and CATCH, which has already been criticised elsewhere (Dalziel & Segal,
2006). Moreover, sensitivity analysis was brief in all studies. Given the number of parameters
and assumptions in the two studies that employed modelling techniques, probabilistic
sensitivity analysis would have been more appropriate. Most importantly, none of these
studies dealt with a possible remission of the intervention effect, and addressed the
uncertainty around long-term effectiveness.
Another issue that evolved was concerned with costs and the consequences of interventions,
although based on a societal perspective, were somewhat narrow in their scope. It appeared
that cost items included were limited to the more obvious resources for personnel and
materials. However, it is plausible that the interventions induced additional resource use
related to the modifications to school food and physical activity practices. Measuring and
including all relevant cost ingredients pertaining to a societal perspective may be a
challenging task (Drummond et al., 2005). Moreover, it is a common concern in the economic
evaluation of public health programmes that intervention boundaries are not clearly defined,
which makes it more difficult for the analyst to identify all relevant costs and consequences
that occur across multiple sectors (Weatherly et al., 2009). Future economic evaluations
alongside primary studies should therefore consider plausible but less obvious indirect
intervention costs, as well as ripple effects, all of which should be assessed prospectively.
Furthermore, limitations associated with the human capital approach to CBA were also
apparent in the evaluations on Planet Health and CATCH. Wages for women and Hispanics
are lower, on average, than those for white men; this probably distorts the value of the
intervention, since the monetary benefits are generally higher for the latter group. A
49
preference-based approach to CBA may have been beyond the scope of these studies, but
would have been desirable.
4.4.3 Generalisability
Due to the range of evaluation pathways taken, it was difficult to put study results side-by-
side for direct comparison. Even the methodologically most similar evaluations of Planet
Health and CATCH differed in a number of respects. This lack of comparability weakens the
ability to draw a strong general conclusion about general economic evidence for obesity
prevention and transferability of findings.
Nevertheless, there may be generalisable elements of the interventions; this could be the
intervention costs as these were reported in a disaggregated manner. It should be kept in mind
that costs of the programmes appeared to be context-specific. In addition, some resource
components that would pertain to a societal perspective were possibly being left out.
However, cost-reporting in the four studies appraised may help to construct a resource
inventory in the planned economic evaluation in Part II.
4.5 Conclusion
The dearth of economic evaluations on interventions concerned with preventing overweight
and obesity in children has been pointed out several times in the literature (Cawley, 2010;
John et al., 2010; Summerbell et al., 2005). However, the four studies identified for this
appraisal demonstrated how economic evaluation could be conducted alongside primary
studies; thus, there may be evidence for economic efficiency of small-scale prevention
programmes with a main focus on the school setting. Nevertheless, considerable
methodological challenges remain, which were predominantly centred on outcomes. This was
about the choice of outcomes measures, valuation of effects and extrapolation of long-term
consequences.
50
5 Systematic review of the effectiveness of school-based policy
interventions to promote healthy weight
Helene Luckner1, 2
John R. Moss2
1 School of Population Health and Clinical Practice, University of Adelaide, Australia
2 Department of Health Care Management, Berlin University of Technology, Germany
Prepared for peer-review
51
Abstract
Policy-based interventions are a promising strategy for combating unhealthy weight in
childhood. The aim of this systematic review was to summarise policy interventions
implemented in the school setting and to determine whether they were effective in preventing
overweight and obesity in students. A systematic literature search was conducted in five
databases up to July 2011. Five experimental studies and six studies based on a before/after
design met the inclusion criteria. The majority of these studies evaluated policies directed at
school food, but only two studies examined the provision of physical education. Overall,
study quality was weak to moderate. In many studies, it was difficult to determine policy
effectiveness due to concurrent intervention components in multi-faceted programmes or
secular trends. However, where studies reported information about implementation, this aided
the interpretation of outcomes. Findings suggest that policies that set standards for school
lunches, snacks and beverages are associated with a shift towards healthier consumption
patterns amongst students. In addition, a mandatory daily physical education class improved
weekly physical activity participation. However, there is little evidence so far for an impact
on anthropometric outcomes. Future research should address study quality, align individual
and organisational outcomes, and evaluate other types of policy on physical activity.
52
5.1 Background
Obesity prevalence in children has risen dramatically in many countries and represents a
serious threat to population health (Han et al. 2010). Policy-makers and public health
advocates are therefore currently seeking effective strategies to prevent excess weight gain
during childhood. Schools have become one major focal point, but efforts in the past
characterised as programmes for individual behaviour change have often met with little
success (Sharma 2006, Summerbell et al. 2005). Moreover, an ecological perspective on the
aetiology of obesity suggests that interventions targeted at the environment are much more
promising (Swinburn et al. 1999). Policy development is one such approach that has been
strongly advocated (Lean et al. 2006, Nestle and Jacobson 2000, Swinburn 2008, WHO 2002)
including policies directed at nutrition and physical activity in the school setting (Wechsler et
al. 2000). Policies are powerful tools in the public health inventory, especially because of
their ability to influence social norms and to foster behaviour change in individuals and
organisations (Cohen et al. 2000, Mensah et al. 2004). Furthermore, they allow wide
coverage of the target population without discrimination in access (Swinburn 2008). There are
a range of different meanings for what is a policy intervention (Oliver et al. 2010). The
present study used the same definition as Sacks et al. (Sacks et al. 2008). They referred to
formal written codes and decisions that bear legal authority i.e. legislation and regulation, but
also considered policies at the school level, as these can also be enforced so that compliance
by staff and students becomes mandatory (Sacks et al. 2008).
In recent years, jurisdictions worldwide have introduced legislative and regulatory measures
to combat obesity in the school setting (Lobstein 2009, Musingarimi 2009, Ryan et al. 2006).
However, it is not yet clear whether these policies have improved student weight status. The
only systematic review that we identified on school nutrition policies had searched the
literature up to November 2007 and found no evidence in that respect, but student dietary
intake improved (Jaime and Lock 2009). However, not all studies included in that systematic
review were about clearly enforced policies. The authors also pointed out the need for more
evaluations of large-scale policy interventions. Given this is an emerging field of research it is
worth exploring whether more policies in the school setting have been evaluated. There
should be a specific focus on a more stringent intervention definition, as enforcement may
help to understand the mechanism for policy effectiveness. In addition, no systematic review
of effectiveness has yet been conducted for other areas where school-based policies are being
implemented: such as physical activity, Body Mass Index (BMI) assessments, and school
wellness (Story et al. 2009).
53
Therefore, the aim of this systematic review was to summarise all policy interventions
implemented in the school setting to promote healthy weight and to determine whether they
were effective in preventing overweight and obesity in students.
5.2 Methods
5.2.1 Search strategy
The literature search was conducted in the databases PubMed, Scopus, CINAHL, Embase and
the Cochrane Central Register of Controlled Clinical Trials. Search strategies were developed
for each of these databases (see Appendix 7). The scope of this systematic review was the
peer-reviewed literature in English and German up to 19 July 2011.
5.2.2 Eligibility criteria
Criteria for study eligibility were developed according to Population, Intervention,
Comparator, Outcomes and Study design (PICOS).
Population: Studies were included that assessed outcomes in school students (all grades) from
industrialised western countries. They were excluded if the study population had a medical
condition.
Intervention and comparator: Eligible interventions were those where the study population
was exposed in the school setting to a policy that was aimed at the promotion of healthy
weight. As in Sacks et al., policy interventions described as non-enforced guidelines or
funding allocations to programmes were not considered eligible (Sacks et al. 2008). Studies
were also excluded if the policy was developed during the course of the intervention, due to
uncertainty about exposure time. A valid comparator was defined as no policy exposure,
which could be the current practice.
Outcomes: Studies were included if they reported outcomes at the individual level.
Anthropometric outcomes as indicators of obesity were of primary interest (e.g. BMI, BMI z-
score, prevalence of overweight and obesity, waist circumference or percentage of body fat).
However, studies limited to behavioural and cognitive outcomes were also included.
Study design: The main focus was on studies with an experimental design (randomised and
non-randomised), however a concurrent control group is often not a feasible option for the
evaluation of a policy intervention. Therefore, we also included studies that assessed
outcomes before and after the implementation. However, studies that were only based on
cross-sectional data and predictive models were excluded.
54
5.2.3 Study selection
The search records from all five databases were combined and duplicates were removed.
First, only titles and abstracts of these records were screened and those were removed where
no doubt existed that the exclusion criteria applied. Full-text articles were obtained for the
remaining records and these were assessed against the criteria for inclusion. In addition, the
bibliographies of identified systematic reviews on policy interventions were also screened for
eligible studies. The first author (HL) conducted these tasks and consulted with the second
author (JRM), where there was doubt about study inclusion.
5.2.4 Study quality
The studies included were appraised with the Quality Assessment Tool for Quantitative
Studies, which is tailored to public health interventions (Thomas). The following criteria were
rated: selection bias; study design; control of confounders; data collection methods;
withdrawals and drop-outs. In addition, the intervention integrity and appropriateness of
statistical analysis were also summarised. Blinding procedures were not included as a
criterion, as this is often not practical in obesity prevention. A study was overall rated as
‘strong’ if no criterion was ‘weak’; it was rated ‘moderate’ for one ‘weak ‘ criterion’ only and
rated ‘weak’ for two or more ‘weak’ criteria (Thomas).
5.2.5 Review process
Data were extracted on general study characteristics (population, design, evaluation length
and the policy content) as well as on student-level outcome measures (anthropometric,
behavioural and cognitive). In addition, it is recommended for systematic reviews of complex
public health interventions to include information about the study context as this is
inextricably linked to effectiveness (Waters et al. 2011b). Therefore, all information reported
about implementation of the policy was also extracted.
A narrative synthesis was used, but no quantitative pooling was conducted given the profound
differences in outcome measures, policy contents and study designs. The studies were
reviewed according to design (experimental or observational), school policy areas (nutrition,
physical activity, school wellness or BMI measurements), policy content and co-
interventions. They were also characterised by the scale (national, state/province or school
level) and the nature of the policy (legislation or regulation). In order to determine policy
effectiveness, studies were reviewed for statistically significant changes in student level
55
outcomes. Furthermore, interpretation of study findings was aided by the information
provided about implementation.
5.3 Results
5.3.1 Literature search
Eleven studies were included in the systematic review. The literature search in five databases
returned 2,416 records, and two additional records were identified by searching the reference
lists of the systematic reviews found in this search. This led in total to 2,418 records. Full-text
articles were obtained for 572 records. Finally, 15 papers reporting on eleven studies were
eligible for the evidence synthesis. A flow chart of the search is depicted in Figure 5-1.
Figure 5-1: Flow chart of the literature search
Records identified through database search
PubMed (n = 398)
Scopus (n = 976)
Embase (n = 1021)
CINAHL (n = 398)
CCTR (n = 23)
Additional records identified through other sources
(n = 2)
Records after duplicates removed
(n = 2,418)
Records screened
(n = 2,418)
Records excluded
(n = 1,846)
Full-text articles assessed for eligibility
(n = 572)
Full-text articles excluded (n = 557):
Study is not about a policy intervention (n = 258)
Study is not a policy evaluation (n = 105)
Evaluation at organizational level (n = 62)
Non-western country (n = 29)
Study design not eligible (n = 26)
Policy not related to obesity prevention (n = 24)
Systematic review (n = 19)
Policy developed during the intervention (n = 15)
No full-text could be retrieved despite maximal effort
(n = 2)
Policy outside school (n = 9)
Not peer-reviewed (n = 5)
Language other than English or German (n = 3)
Studies included in the narrative synthesis
(n = 15 full-text articles reporting on 11 studies)
56
5.3.2 General study characteristics
The eleven studies differed to some extent with respect to design, duration and the study
population. Details are provided in Table 5-1. Nine were conducted in the USA (Barroso et
al. 2009, Blum et al. 2008, Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009, Foster et
al. 2008, Jordan et al. 2008, Mendoza et al. 2010, Parcel et al. 1989, Sanchez-Vaznaugh et al.
2010, Schwartz et al. 2009, Simons-Morton et al. 1991, Woodward-Lopez et al. 2010); one
study was conducted in England (Haroun et al. 2011) and one in Canada (Mullally et al.
2010). They were all published since 2007 with the exception of one study from 1989-91
(Parcel et al. 1989, Simons-Morton et al. 1991).
There was only one randomised controlled trial (RCT) (Foster et al. 2008); four studies were
controlled trials (Blum et al. 2008, Jordan et al. 2008, Parcel et al. 1989, Schwartz et al. 2009,
Simons-Morton et al. 1991); and six studies had a before/after design based on two or more
cross-sectional surveys that did not link individual data (Barroso et al. 2009, Cullen et al.
2008, Cullen et al. 2006, Cullen et al. 2009, Haroun et al. 2011, Mendoza et al. 2010,
Mullally et al. 2010, Sanchez-Vaznaugh et al. 2010, Woodward-Lopez et al. 2010). Two
studies were of a shorter duration (≤ 1 year) (Blum et al. 2008, Jordan et al. 2008); six studies
evaluated a timeframe between 2 and 4 years (Barroso et al. 2009, Foster et al. 2008, Haroun
et al. 2011, Parcel et al. 1989, Schwartz et al. 2009, Simons-Morton et al. 1991, Woodward-
Lopez et al. 2010); and three studies had a longer duration up to 7 years (Cullen et al. 2008,
Cullen et al. 2006, Cullen et al. 2009, Mendoza et al. 2010, Mullally et al. 2010, Sanchez-
Vaznaugh et al. 2010).
Only one study measured outcomes in a sample of students representative of the general
population (Haroun et al. 2011). The others assessed students who were described as either
from ethnic minorities or from low-income backgrounds (Barroso et al. 2009, Cullen et al.
2008, Cullen et al. 2006, Cullen et al. 2009, Foster et al. 2008, Mendoza et al. 2010, Parcel et
al. 1989, Sanchez-Vaznaugh et al. 2010, Schwartz et al. 2009, Simons-Morton et al. 1991,
Woodward-Lopez et al. 2010), or who were potentially advantaged (Blum et al. 2008, Jordan
et al. 2008). In addition, one study did not provide demographic details (Mullally et al. 2010).
No specific differentiation could be made between the studies with respect to years of age, but
four studies were about children who were mostly less than 12 years old (Foster et al. 2008,
Haroun et al. 2011, Jordan et al. 2008, Parcel et al. 1989, Simons-Morton et al. 1991).
Students were mainly older in the other seven studies.
57
Table 5-1: Summary of studies on school-based policy interventions to promote healthy weight
Study details
Setting and study population
Intervention and policy description
Barroso et al. 2009
Before/after study
BL: 2004-05; FU: 2006-08
Middle schools (Texas, USA);
mainly Latinos in region with
high economic disadvantage
Texas Senate Bill 42 mandates 30 minutes of moderate-to-vigorous-physical activity for middle schools and
a minimum participation per semester in physical education class. Middle schools also have to implement a
coordinated school health programme and restore a school health advisory council.
Blum et al. 2008
Controlled trial
BL: 2004-2005; FU: 2005
High schools (Maine, USA);
convenience sample of mainly
white students
School-based policy to reduce availability of sugar-sweetened beverages and diet soda in the a la carte
programme and vending machines.
Foster et al. 2008
RCT
BL: unknown; FU: after 2 years
K-Grade 8 schools
(Philadelphia, USA); ≥ 50% of
students eligible for free or
reduced-price meals, > 60%
Black, Asian or Hispanic
A policy for vending machines and canteens restricted the choice and the size of beverages and set
mandatory standards for snacks. The policy was one component of a multi-faceted programme, which also
included nutrition education in class, social marketing in schools and family outreach.
Haroun et al. 2011
Before/after study
BL: 2005; FU: 2009
Primary schools (England);
nationally representative sample
of students
Schools have to comply by law with national Food Based Standards (FBS) and Nutrition Based Standards
(NBS) for school meals. Thirteen FBS were introduced to increase access to healthy foods and to limit the
availability of unhealthy foods. Fourteen NBS regulate energy, sugar, fat and sodium content of school food.
Jordan et al. 2008
Controlled trial
BL: 2005; FU: 2006
Elementary schools (Utah,
USA); >80% of students were
white
A state-wide scheme awards schools with designations (bronze, silver, gold or platinum) that represent
increasing levels of achievement in implementing school wellness criteria. The extensive list of criteria
includes policies for school food sources, school food environment and structured physical activity.
Mendoza et al. 2010, Cullen et al.
2009, Cullen et al. 2008, Cullen et al.
2006
Before/after study
BL: 2001-02; FU1: 2002-03; FU2:
2005-06
Middle schools (Texas, USA);
one low, one moderate and one
high SES school; high
proportions of Hispanic students
First, a local school district policy was implemented that limited choices in snack bars and removed vending
machines from cafeterias (data FU1). This was followed by the Texas Public School Nutrition Policy, which
regulates all school foods sources in middle schools (data FU2). It restricts the portion size of unhealthy food
(snacks, sugar-sweetened beverages) and limits the frequency of serving high fat vegetables.
Mullally et al. 2010
Before/after study
BL: 2001-02; FU: 2007
Elementary schools (Prince
Edward Island, Canada); no
demographic information
reported
Prince Edward Island School Nutrition Policy mandates a traffic light system for food provided by the
schools. This applies to the food sold in vending machines, food involved in fundraising and the types of
food for which advertisement is allowed in schools. In addition, the policy promotes pricing strategies and
the use of non-food items for rewarding students.
Parcel et al. 1989, Simons-Morton et
al.
1990,
Controlled trial
BL: 1985; FU: 1987
Elementary schools (Texas,
USA); approximately 35% of
students are Mexican or Black
The policy referred to food preparation in canteens and restricted the fat and salt content of school lunches.
The policy was one component of a multi-faceted programme included a health education curriculum,
training for canteen staff and revision of the physical education curriculum.
Sanchez-Vaznaugh et al. 2010
Before/after study
BL: 2001; FU: 2002-2008
Seventh and ninth graders,
(California, USA); 48% of
students Hispanic in state-wide
sample and 79% Hispanic in
subsample from Los Angeles
Los Angeles Unified School District regulates all foods, snacks and beverages offered in school. This
applies to all grades. In addition, state-wide law prohibits sale of sugary beverages and restricts competitive
fooda in elementary and middle schools (2004 California Childhood Obesity Prevention Act (Senate Bill
677) and 2007 Nutrition standards (Senate Bill 12)).
58
Study details
Setting and study population
Intervention and policy description
Schwartz et al. 2009
Controlled trial
BL: unknown; 2 years
Middle schools (Connecticut,
USA); 25% of students
Hispanic, 33% eligible for free
or reduced-price meals
A school-based policy was implemented that set guidelines for all snacks sold at the cafeteria a la carte,
from vending machines and through fundraisers.
Woodward-Lopez et al. 2010
Before/after study
BL: 2006; FU: 2008
Seventh and ninth graders
(California, USA); 65% of
students Latino, low-income
area
Californian legislation limits availability of competitive fooda in elementary and middle schools through
standards for nutrition (Senate Bill 12) and beverages (Senate Bill 965).
BL: Baseline; FBS: Food-Based Standards; FU: Follow-up; NBS: Nutrition-Based Standards SES: Socio-economic status.
a Competitive food in the US refers to food sold outside the National School Lunch Programme in cafeterias, snack bars and vending machines.
59
5.3.3 Study quality
Overall, studies were of moderate or low quality. An overview of the quality rating is
provided in Table 5-2. Only the RCT study was rated as ‘strong’ (Foster et al. 2008). There
were some issues of concern in the four non-randomised controlled trials and these were all
rated ‘weak’. This was mainly due to a lack of adjustment for confounders (Blum et al. 2008,
Jordan et al. 2008, Parcel et al. 1989, Simons-Morton et al. 1991) or insufficient reporting of
withdrawals and dropouts (Jordan et al. 2008, Parcel et al. 1989, Schwartz et al. 2009,
Simons-Morton et al. 1991).
Five studies with a before/after design were rated ‘moderate’ in quality. One study was rated
as ‘weak’ due to the absence of reporting on both the validity of data collection tools and the
adjustment for confounders (Haroun et al. 2011), which was also a weakness in two other
before/after studies (Barroso et al. 2009, Cullen et al. 2008, Cullen et al. 2006, Cullen et al.
2009, Mendoza et al. 2010). In addition, the authors in one study noted the possibility of
selection bias due to changes in school enrolment and a lower response rate in the second
survey (Mullally et al. 2010). Cullen et al. also noted possible selection bias in the data
collection process at lunch tables in their study (Cullen et al. 2008).
Furthermore, in three studies it appeared likely that other initiatives or secular trends could
have influenced the outcomes (Blum et al. 2008, Parcel et al. 1989, Sanchez-Vaznaugh et al.
2010, Simons-Morton et al. 1991). Another weakness was that the unit of analysis was not
clearly described in five studies (Barroso et al. 2009, Blum et al. 2008, Cullen et al. 2008,
Cullen et al. 2006, Cullen et al. 2009, Haroun et al. 2011, Jordan et al. 2008, Mendoza et al.
2010).
60
Table 5-2: Quality rating of studies included based on the Quality Assessment Tool for Quantitative Studies (Thomas 2003)
Study
Selection
bias
Study
design
Confounders
Data
collection
methods
Withdrawals
and drop-outs
Intervention integrity
Statistical analysis
Overall
rating
Barroso et al. 2009
B
B
C
A
B
/
- Unit of analysis not clear
B
Blum et al. 2008
C
A
C
A
A
- Possible
contamination
- Unit of analysis not clear
- No intention to treat analysis
C
Foster et al. 2008
B
A
A
A
B
- Did not measure
consistencya
/
A
Haroun et al. 2011
B
B
C
C
B
/
- Unit of analysis not clear
C
Jordan et al. 2008
B
A
C
A
C
- Did not measure
consistencya
- Unit of analysis not clear
- No intention to treat analysis
C
Mendoza et al. 2010
Cullen et al. 2009, Cullen et
2008, Cullen et al. 2006
B
B
C
A
B
- Did not measure
consistencya
- Unit of analysis not clear
B
Mullally et al. 2010
C
B
B
A
B
- Did not measure
consistencya
/
B
Parcel et al. 1989,
Simons-Morton et al. 1990
B
A
C
A
C
- Possible
contamination
- No intention to treat analysis
C
Sanchez-Vaznaugh et al.
2010
A
B
B
A
B
- Extent of exposure to
policy not clear
- Did not measure
consistency
- Possible co-
intervention
/
B
Schwartz et al. 2009
B
A
A
C
C
- Did not measure
consistencya
/
C
Woodward-Lopez et al. 2010
C
B
B
A
B
/
/
B
A: Strong; B: Moderate; C: Weak;
aConsistency means that all participants received the intervention as designed.
61
5.3.4 Interventions
Most policies were concerned with regulating nutrition in schools. All studies with an
experimental design referred to policies at the school level only (Blum et al. 2008, Foster et
al. 2008, Parcel et al. 1989, Schwartz et al. 2009, Simons-Morton et al. 1991). Two of these
studies evaluated multi-faceted programmes where a policy was implemented as one amongst
several intervention components (Foster et al. 2008, Parcel et al. 1989, Simons-Morton et al.
1991). In addition, Jordan et al. evaluated an award scheme that set incentives for schools to
implement a list of school health criteria, which included policies (Jordan et al. 2008).
Conversely, all studies that evaluated policies implemented on a large scale were based on a
before/after design. Four studies from the USA (Barroso et al. 2009, Cullen et al. 2008,
Cullen et al. 2006, Cullen et al. 2009, Mendoza et al. 2010, Sanchez-Vaznaugh et al. 2010,
Woodward-Lopez et al. 2010) and the Canadian study (Mullally et al. 2010) evaluated
policies at the state or the province level respectively; the English study evaluated a national
policy (Haroun et al. 2011). Three of these studies were about legislation (Barroso et al. 2009,
Haroun et al. 2011, Woodward-Lopez et al. 2010). The school food policies in the Canadian
province Prince Edward Island and in Texas were regulatory (Cullen et al. 2008, Cullen et al.
2006, Cullen et al. 2009, Mendoza et al. 2010, Mullally et al. 2010). In addition, one study
evaluated both Californian legislation as well as regulation in the Los Angeles Unified School
District (Sanchez-Vaznaugh et al. 2010).
The majority of study authors described the policy content in detail. Additional information
from the Internet was obtained for two studies (Jordan et al. 2008, Mullally et al. 2010,
PEI Healthy Eating Alliance. , Utah Department of Health.). Ten out of the eleven included
studies evaluated nutrition-related policies. These were mainly policies about either school
meals (Parcel et al. 1989, Simons-Morton et al. 1991), snack food (Foster et al. 2008,
Schwartz et al. 2009, Woodward-Lopez et al. 2010) or both (Cullen et al. 2008, Cullen et al.
2006, Cullen et al. 2009, Haroun et al. 2011, Mendoza et al. 2010, Mullally et al. 2010). This
also included regulation of beverages in five of these studies (Cullen et al. 2008, Cullen et al.
2006, Cullen et al. 2009, Foster et al. 2008, Mendoza et al. 2010, Mullally et al. 2010,
Sanchez-Vaznaugh et al. 2010, Woodward-Lopez et al. 2010). In addition, one study
evaluated an intervention that only limited the availability of beverages (Blum et al. 2008).
Three studies on policies in the USA supplemented federal law by restricting the so-called
competitive food, which is sold outside the National School Lunch Programme in cafeterias,
snack bars and vending machines (Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009,
Mendoza et al. 2010, Sanchez-Vaznaugh et al. 2010, Woodward-Lopez et al. 2010). In
62
addition, two studies were found with policies referring to the wider school food environment
such as bans on advertising and not using food as a reward in class (Jordan et al. 2008,
Mullally et al. 2010). Only two studies covered a policy for the provision and frequency of
structured physical activity classes (Barroso et al. 2009, Jordan et al. 2008). One criterion in
the Gold Medal Programme in the US was also a policy about specialist requirements for staff
delivering physical education classes (Jordan et al. 2008).
Furthermore, two studies covered policies on school wellness. One was that a set of options
for staff wellness had to be in place in the Gold Medal Programme (Jordan et al. 2008) and
one was the mandatory restoration of a school health advisory council in Texas (Barroso et al.
2009).
5.3.5 Behavioural outcomes
Behaviour change was measured in all but one study (Sanchez-Vaznaugh et al. 2010), and
three studies also assessed cognitive outcomes (Jordan et al. 2008, Parcel et al. 1989,
Schwartz et al. 2009, Simons-Morton et al. 1991). Table 5-3 summarises the study outcomes.
Behavioural outcomes were not equivocal in the five studies with experimental design. The
nutrition policy in the only RCT study did not improve dietary intake (Foster et al. 2008). The
only significant effect for behavioural outcomes in that study was a reduction in TV-viewing.
On the other hand, Schwartz et al. reported that a snack policy in schools was effective in
reducing the consumption of most targeted foods and beverages in the intervention group
(Schwartz et al. 2009). There was also no evidence for a compensatory effect at home.
Conversely, Blum et al. found a shift towards healthier drinking patterns in the intervention
group following the introduction of a beverage policy, but there was no change in
consumption of sugar-sweetened drinks compared to controls (Blum et al. 2008). The authors
speculated that students might have consumed more of these drinks at home instead. The
Gold Medals Award programme in Jordan et al. found no effect on behavioural outcomes
(Jordan et al. 2008). A parent survey showed statistically significant improvements for soft
drink consumption and participation in active transport, although this was compromised by
low participation rates. Parcel et al. found for the comprehensive health education and food
service modification policy that intervention students had a statistically significant reduction
in sodium intake (Parcel et al. 1989, Simons-Morton et al. 1991).
Behavioural outcomes generally improved where this was assessed in before/after studies.
Woodward-Lopez et al. found the introduction with Californian school food legislation to be
associated with a statistically significant reduction in the consumption of soda drinks at
school and there was no evidence for compensation at home (Woodward-Lopez et al. 2010).
63
Water intake also improved, but vegetable consumption decreased (p<0.01). The Texas
School Nutrition Policy was also associated with healthier lunch patterns with respect to most
food and beverage items assessed (Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009,
Mendoza et al. 2010). Nevertheless, the percentage of kilojoules from saturated fat was not
significantly reduced and the percentage of kilojoules from fat and was still above the
recommended levels. Improvement in lunch energy density was most pronounced in students
with a relatively higher socio-economic status. This could be due to these students having
been more likely to be able to afford competitive foods before policy implementation
(Mendoza et al. 2010). Furthermore, nutrition policies in England and in Prince Edward
Island in Canada were associated with statistically significant improvements in school lunch
dietary intake (Haroun et al. 2011, Mullally et al. 2010). In addition, Barroso et al. reported
that days with structured physical activity significantly increased for a sample of middle
school students in disadvantaged areas of Texas (Barroso et al. 2009).
5.3.6 Anthropometric outcomes
Student weight status improved in three out of four studies with anthropometric outcomes
(Table 5-3) (Barroso et al. 2009, Foster et al. 2008, Jordan et al. 2008, Sanchez-Vaznaugh et
al. 2010). The RCT study reported statistically significant reductions of overweight incidence
in the intervention group, as well as a higher remission for overweight and obesity combined
(Foster et al. 2008). The non-randomised controlled trial on The Gold Medals Award
programme resulted in a significant increase in BMI z-score for the control group only
(Jordan et al. 2008).
Furthermore, the before/after study of Sanchez-Vaznaugh et al. found that overweight was no
longer significantly increasing in fifth graders in Los Angeles and in seventh graders in the
rest of California (Sanchez-Vaznaugh et al. 2010). However, there was no notable difference
in overweight trends between their district and the state-wide samples despite the regulations
in Los Angeles tending to be more stringent. Conversely, there was no change in
anthropometric outcomes resulting from the physical education policy in Texas, however
little time had elapsed between law implementation and follow-up (Barroso et al. 2009).
5.3.7 Implementation
The right-hand column in Table 5-3 contains the information in studies about policy
implementation, although this was often not thoroughly reported. Only five studies
specifically assessed policy compliance in schools (Barroso et al. 2009, Blum et al. 2008,
Haroun et al. 2011, Parcel et al. 1989, Simons-Morton et al. 1991, Woodward-Lopez et al.
64
2010), whilst three other studies provided a brief description (Cullen et al. 2008, Cullen et al.
2006, Cullen et al. 2009, Jordan et al. 2008, Mendoza et al. 2010, Sanchez-Vaznaugh et al.
2010). Three studies did not assess implementation nor report relevant information (Foster et
al. 2008, Mullally et al. 2010, Schwartz et al. 2009).
Schools generally adhered to policies in the five studies where this was assessed. However,
Simons-Morton et al. noted that their intervention was unevenly implemented across schools
and it remained unclear whether the policy component was also affected (Parcel et al. 1989,
Simons-Morton et al. 1991). Haroun et al. also found that the energy content of lunches still
exceeded the standard in most schools (Haroun et al. 2011). In Woodward-Lopez et al.,
policy compliance increased over time for competitive food items in the middle schools, but
did not reach the desired level (Woodward-Lopez et al. 2010). In contrast, little was known
about the state-wide compliance with the policies in other Californian schools and Sanchez-
Vaznaugh et al. noted that there may have been a large variation (Sanchez-Vaznaugh et al.
2010). Cullen et al. did not specifically evaluate policy implementation, but noted that
compliance was high in the schools participating in that study (Cullen et al. 2008). Jordan et
al. did not provide any information about policy implementation, but this could be implicitly
assumed as sufficient given that schools received the Gold Medal distinction after study
completion (Jordan et al. 2008). Nevertheless, adherence to these policies was not described.
65
Table 5-3: Outcomes of studies on school-based policy interventions to promote healthy weight
Study
Anthropometric outcomes
Behavioural and cognitive outcomes
Policy implementation
Barroso et al.
2009
Assessed: 1) mean BMI, 2) percentage of
overweight and obese
Found: No change in either outcome
Assessed: 1) self-reported physical education class days per week, 2)
daily TV watching/playing video games
Found: 1) statistically significant increase, 2) reduction not statistically
significant
The majority of key informants (state-wide and
border regions) were aware of Senate Bill 42
and most school districts had implemented a
school health advisory council. Frequency and
length of physical education exceeded the
required amount and quality of class met the
standards that it was compared to.
Blum et al.
2008
/
Assessed: Consumption of 1) Diet soda drinks 2) juice 3) milk 3) sugar-
sweetened beverages
Found: 1) Statistically significant reduction for girls only, 2) statistically
significant reduction for girls only, 3) statistically significant increase for
boys only, 3) no statistically significant change
Offerings of sugar-sweetened beverages and
diet drinks were virtually eliminated in a la
carte and vending programmes, but were only
reduced slightly in control schools.
Foster et al.
2008
Assessed: 1) Incidence (overweight,
obesity), 2) prevalence (overweight,
obesity), 3) Remission (overweight,
obesity), 4) mean BMI and mean BMI z-
score
Found: 1) Statistically significant
reduction for overweight but not for
obesity, 2) no statistically significant
reduction, 3) statistically significant for
overweight and obesity combined 4)
reduction not statistically significant
Assessed: 1) hours of inactivity, 2) weekday television watching, 3)
hours of total activity, 4) dietary intake, 5) adverse events
Found: 1) statistically significant reduction, 2) statistically significant
reduction, 3) no statistically significant change, 4) no statistically
significant change, 5) no evidence for potential adverse events
/
Haroun et al.
2011
/
Assessed: Percentage of students having 1) vegetables and salad, 2) fruit,
3) starchy foods not cooked in fat, 4) sandwiches, 5) dairy products, 6)
water, 7) fruit juice 8) fruit-based desserts, 9) other desserts, 10)
condiments, 11) non-permitted foods, 12) drinks, 13) starchy foods
cooked in fat, 14) baked beans, 15) main dishes
Found: 1)-8) statistically significant increase, 9)-13) statistically
significant reduction, 14) no change, 15) no change
School compliance with the new standards was
generally satisfying except for the provisions
for fruit and vegetables as well as for some
nutrient standards. The energy density remained
above the standards in most schools. Students
consumed healthier lunches in schools with
good compliance.
66
Study
Anthropometric outcomes
Behavioural and cognitive outcomes
Policy implementation
Jordan et al.
2008
Assessed: BMI z-score
Found: Statistically significant increase in
the control group only.
Assessed (student survey): 1) Dietary habits, 2) physical activity, 3)
sedentary activity, 4) self-efficacy (dietary and exercise)
Found (student survey): No statistically significant change in outcomes
Assessed (parent survey): 1) Soft drink consumption, 2) days per week
walking or biking to school, 3) other eating habits (not specified), 4)
physical activity, 5) sedentary activity, 6) parent perceptions (not
specified)
Found (parent survey): 1) Statistically significant reduction, 2)
statistically significant increase, 3)-6) no change in outcomes
Implementation was not specifically assessed. It
was stated that both schools achieved Gold
Medal status after study completion.
Mendoza et al.
2010
Cullen et al.
2009, Cullen et
2008, Cullen et
al. 2006
/
Assessed in Mendoza et al. (33): 1) Total energy density of lunch intake
(with and without beverages), 2) percentage of energy from food groups
a) mixed entrees, b) desserts, c) vegetables, d) fruits, e) fat/oil, f) candy,
g) snack/chips, h) grains
Found in Mendoza et al. (32): 1) statistically significant reduction 2)a)-d)
statistically significant increase, e)-g) statistically significant reduction,
h) no change
Assessed in Cullen et al. (29): Consumption of 1) energy, 2) protein,
fibre, 3) vitamin A, 4) vitamin C, 5) calcium 6) sodium 7) vegetables, 8)
milk, 9) % of kilojoules from fat, 10) sweetened beverages, 11) soft
drinks, 12) dessert foods, 13) snack chips, 14) fruit or fruit juice, 15)
high-fat vegetables, 16) % of kilojoules from saturated fat, 17) Iron, 18)
high-fat vegetables, 19) candy
Found in Cullen et al. (28): 1)-8) Statistically significant increase 9)-13)
statistically significant reduction, 14)-19) changes were not statistically
significant
Implementation was not specifically assessed. It
was stated that schools were very compliant
with the state policy and that snack machine
inventory adhered to the standards.
Mullally et al.
2010
/
Assessed: Intake of 1) milk and alternative beverages, 2) vegetables and
fruit, 3) low-nutrient density foods 4) relative odds for meeting the
recommended levels of intake
Found: 1) Statistically significant increase, 2) no statistically significant
change, 3) statistically significant reduction, 4) statistically significant for
fruits and vegetables as well as for low-nutrient density food.
/
Parcel et al.
1989,
Simons-
Morton et al.
1990
/
Assessed: 1) Salt use, 2) intake of fruit and vegetables as a percentage of
total foods, 3) aerobic activity, 4) cognitive dietary and exercise
outcomes (behavioural capability, self-efficacy, behavioural
expectations)
Found: 1) Statistically significant reduction, 2) no statistically significant
change, 3) statistically significant increase when student was the unit of
analysis, 4) statistically significant improvements in diet behavioural
capability, diet self-efficacy (student level analysis only) and diet
behavioural expectations
Nutrient analysis of lunches showed that
intervention schools reduced sodium content
and saturated fat. Authors noted that
implementation might have varied across the
schools.
67
Study
Anthropometric outcomes
Behavioural and cognitive outcomes
Policy implementation
Sanchez-
Vaznaugh et al.
2010
Assessed: Rate of increase in the
combined prevalence of overweight and
obesity
Found: Statistically significant decline for
fifth graders in Los Angeles and for fifth
grade boys and seventh graders in the rest
of California.
/
Implementation was not directly assessed, but
referred to from other sources as being
inadequate in Los Angeles. Compliance with
the policies in the rest of California remained
unknown.
Schwartz et al.
2009
/
Assessed: 1) Consumption of items meeting nutritional standards
(beverages, salty snacks, sweet snacks), 2) consumption of items not
meeting nutritional standards, 3) compensatory consumption at home, 4)
Adverse events (dieting, weight concern)
Found: 1) Statistically significant increase in all items, 2) Statistically
significant reduction for beverages and salty snacks, no evidence for 2)
and 3)
Authors note that they did not monitor
implementation in schools.
Woodward-
Lopez et al.
2010
/
Assessed: 1) Soda drinks (home, school), 2) water (home, school), 3)
vegetables (home, school), 4) milk, 5) fruit, 6) sport drinks, 7) candy, 8)
chips
Found: 1) Statistically significant decrease (at school, but no change at
home) 2) statistically significant increase (school and home), 3)
statistically significant decrease (school only), 4)-8) no statistically
significant change in outcomes
Compliance with standards remained constant
in middle schools, but the product mix offered
changed. However, the total number of types of
competitive foods did not decrease notably.
Compliance was generally higher for beverages
than for food items.a
aThe information about implementation described in this table is limited to the set of schools (HEAC schools) where student level outcomes were obtained.
68
5.4 Discussion
The purpose of this systematic review was to summarise policy interventions in the school
setting and to assess whether these are effective in preventing overweight and obesity. We
appraised study quality, outcomes and contextual information in order to determine evidence
for effectiveness. Nutrition was the most frequently targeted policy area, but only two studies
included a policy for the provision of structured physical activity. In general, the studies had a
moderate to weak quality score and six out of eleven studies were based on an observational
design.
Policy effectiveness was uneven. Ten studies reported a statistically significant change in a
main outcome, but only three of these were in anthropometric measures. Furthermore, in four
studies it was unclear whether changes in outcomes could be attributed to the policy, since
either concurrent intervention components in multi-faceted programmes (Foster et al. 2008,
Jordan et al. 2008) or secular trends in before/after studies (Mullally et al. 2010, Sanchez-
Vaznaugh et al. 2010) may have had a stronger influence. Conversely, policy effectiveness
became more plausible in studies where implementation was sufficiently described.
Therefore, there were promising findings, though mainly based on studies with an
observational design and limited to behavioural outcomes. Policies for school lunches were
associated with shifts towards healthier consumption patterns (Cullen et al. 2008, Cullen et al.
2006, Cullen et al. 2009, Haroun et al. 2011, Mendoza et al. 2010, Parcel et al. 1989, Simons-
Morton et al. 1991, Woodward-Lopez et al. 2010). This was also observed for policies
directed at snack food and beverage availability in vending machines and cafeterias (Cullen et
al. 2008, Cullen et al. 2006, Cullen et al. 2009, Mendoza et al. 2010, Schwartz et al. 2009,
Woodward-Lopez et al. 2010). In addition, one study reported how mandatory daily physical
education was associated with increased self-reported weekly physical activity (Barroso et al.
2009). Furthermore, the quality of large-scale policies may still leave scope for improvement.
Despite comprehensive nutrition policies being in place in England, California and Texas,
studies still reported insufficient school compliance or dietary intake exceeding recommended
levels for some foods and nutrients (Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009,
Haroun et al. 2011, Mendoza et al. 2010, Woodward-Lopez et al. 2010).
A long time can elapse before an enacted policy translates into health outcomes and
additional allowance must be made for appropriate evaluation to undergo peer-review
(Macdiarmid et al. 2011). As this is still an early stage in the evaluation of school-based
policies, the number of studies included in this systematic review was still small. There were
also notably fewer studies evaluating policies about physical activity. Other reviews on
69
wellness policies in US also found less attention to the provision for such policies compared
the nutritional domain (Metos and Murtaugh 2011, Moag-Stahlberg et al. 2008, Nanney et al.
2010). In addition, policies referring to other areas of school health such as mandatory health
advisory councils or staff wellness were only included in two studies, but their impact was not
directly assessed (Barroso et al. 2009, Jordan et al. 2008). No study assessed BMI
measurement programmes in schools. Furthermore, there was no evaluation of policy options
to influence the wider school food environment e.g. through a closed campus or regulating
food brought from home.
It was encouraging to identify a number of natural experiments being evaluated (Barroso et
al. 2009, Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009, Haroun et al. 2011,
Mendoza et al. 2010, Mullally et al. 2010, Sanchez-Vaznaugh et al. 2010, Woodward-Lopez
et al. 2010). Although findings from such studies can never be fully conclusive, they can still
provide indicative evidence that a policy is working (Petticrew et al. 2005). The availability
of baseline data is crucial for harnessing natural experiments (Ramanathan et al. 2008).
Sometimes this requires creative solutions as seen in one study that followed up the control
group of a previously conducted trial before the state-wide school nutrition policy was
introduced (Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009, Mendoza et al. 2010).
This also emphasizes the importance of appropriate surveillance systems being in place in
order to track changes over time at a population level (Story et al. 2009).
The policies reviewed were intended to induce an environmental change that is in turn
expected to impact on individual behaviour (Mensah et al. 2004), but there are complex
mechanisms at play in how this change is translate into prevented weight gain in students.
Therefore, understanding whether a policy is working requires anthropometric and
behavioural outcomes in individuals as well as processes at the organizational level. Where
such information is available, this can help to tease out the effectiveness of a specific policy
as part of multi-component initiative (McGraw et al. 2000). Barroso et al. and Woodward-
Lopez et al. demonstrated how aligning outcomes at both individual and organisational level
can add strength to the ability to interpret study findings (Barroso et al. 2009, Woodward-
Lopez et al. 2010).
Many studies pointed out difficulties related to the measurement of behavioural outcomes,
notably where these were self-reported. This concerns the time span assessed for dietary
intake (Blum et al. 2008, Cullen et al. 2008, Cullen et al. 2006, Cullen et al. 2009, Mendoza
et al. 2010), the importance of capturing all food sources (Blum et al. 2008), and the validity
of responses to questionnaires (Foster et al. 2008, Jordan et al. 2008, Parcel et al. 1989,
Schwartz et al. 2009, Simons-Morton et al. 1991). An implication for the assessment of
70
nutritional policies is that data collection should comprise a representative time span like a
24-hour recall including an indication about the location of consumption and the food
sources. This is so important, because any observation of behaviour change resulting from a
policy needs more information about shifts in patterns throughout the day, not only the dietary
intake at lunch-time. Only two studies also assessed dietary intake at home, but found no
evidence for compensation (Schwartz et al. 2009, Woodward-Lopez et al. 2010).
Nevertheless, this did still leave out other locations. Furthermore, despite it being insightful
where studies include a range of outcome indicators, they should not refrain from nominating
the ones of primary interest (Craig et al. 2008).
A limitation of this systematic review is that only peer-reviewed papers were included and
therefore publication bias cannot be ruled out. Furthermore, studies were included regardless
of the type of policy involved. Nevertheless, there may be profound differences whether the
action is legislative, regulatory or an enforced guideline as well as the level and scale it refers
to. This is not to mention the variation in stringency and the dependence on unique contextual
circumstances. Facilitating comparison between policies is thus inherently difficult, and the
generalisability of findings may be low.
In conclusion, this systematic review found limited evidence for school-based policies to
reduce overweight and obesity in children. However, it is plausible that policies targeted at
school food sources improved student’s dietary consumption patterns. The emerging number
of evaluations in this field is encouraging, however study quality needs to improve. Future
evaluations should include more areas for school policies, particularly for physical activity.
71
PART II: ECONOMIC EVALUATION OF A HYPOTHETICAL
SET OF INTERVENTIONS TO PREVENT OVERWEIGHT AND
OBESITY IN SOUTH AUSTRALIA
Objective in Part II
In order to adequately inform policy-makers in South Australia and elsewhere about optimal
investment towards obesity prevention strategies for children, economic evaluation should be
conducted in order to prioritise between the most promising strategies (Drummond et al.
2005, Ganz 2003). This ought to be based on the best evidence available for effects as well as
a sound understanding on the resource use involved taken from the experience gained
elsewhere, so that a hypothetical intervention can be modelled for the local context. This then
yields a list of prevention options ranked from the one with the most favourable ICER down
the ones with little or no value for money. However, in the case of obesity prevention in
children the following points need to be made based on conclusions derived in the previous
four chapters.
The first issue was that comparing the ICER of single interventions might yield little
informative value. The meta-analysis in chapter 2 found that there was evidence for multi-
component interventions to be effective in preventing obesity in children, which is coherent
with the current literature on theory and practice of obesity prevention (Kumanyika et al.
2008, Swinburn and Egger 2002). Others have also criticised how common economic
evaluation ignores the likely synergies between single interventions in terms of both costs and
effects (Hutubessy et al. 2003). Moreover, it could be seen in chapter 3 that it is inherently
difficult to disentangle the effectiveness of single components within a multi-faceted
programme. Therefore, conducting health economic evaluation on single interventions that
are part of a whole bundle was considered to be less meaningful. Conversely, the implication
of this first notion was that one whole package of interventions should be subject to economic
evaluation. This was also specifically relevant to the decision-making context, since the
implementation of a multi-faceted community-based intervention (CBI) was being discussed
in South Australia (Daniel et al. 2009). A typical feature of these interventions is that they
consist of coordinated multiple intervention components at a several levels (King et al. 2011).
Furthermore, they are implemented in multiple settings within a community, and a wider
range of stakeholders is engaged in the planning and the operation of the intervention.
Secondly, considerable uncertainty could be expected around the estimates of both effects and
costs when modelling the cost-effectiveness of a CBI in South Australia. In terms of effects,
this was due to the fact that the current evidence base only yields modest body composition
72
outcomes for children. In addition, evidence was only derived from short-term evaluations,
but there is a gap of knowledge on the maintenance of this effect in the long-run, since a
relapse is possible. Therefore, no evidence-based judgement about the health consequences in
adulthood could be derived. Furthermore, modelling a multi-faceted intervention in the
community setting requires a thorough understanding of the resources involved in every
single component as well as the processes linking all intervention elements. A CBI engages
with multiple stakeholders, who inevitably either dedicate resources or receive benefits
through the programme or both (Johansson et al. 2009). Identifying these inter-sectoral
consequences may prove to be salient information, when considering barriers and facilitators
to an intervention (Johansson et al. 2009). This comprehensive understanding required about
the resource use contrasts with the data available on intervention costs in published studies as
outlined in chapter 4. Unfortunately, economic evaluations in the field of obesity prevention
are still scant in general and, to date, only one exists for a CBI study - from New Zealand
(McAuley et al. 2010). This meant that a number of additional assumptions for transferability
as well as for missing cost elements would be required in the economic evaluation of the CBI.
Hence, it appeared prudent to restrict the analysis to the present time frame in terms of
effectiveness, but to explore the assumptions needed in order to estimate the programme costs
from a societal perspective.
These two implications presented here prompted the original purpose of this chapter,
assessing the cost-effectiveness of single strategies, to be reframed. The amended aim was to
examine how the cost and consequences of one complex CBI to prevent overweight and
obesity in children could be estimated for the South Australian context under the assumption
of large-scale implementation.
This was organised by two main steps. The first one presents the background and the
methodology of the economic evaluation in the format of a paper submitted for peer-review.
The second part contains additional details of the calculation and presents the results, as well
as a discussion of the findings.
73
6 Economic issues in the planning of community-based obesity
prevention
Helene Luckner1, 2
John R. Moss2
1 School of Population Health and Clinical Practice, University of Adelaide, Australia
2 Department of Health Care Management, Berlin University of Technology, Germany
Prepared for peer-review
74
ABSTRACT
Multi-faceted interventions in the community are currently considered to be a promising
approach in the combat to prevent childhood obesity. However, implementing interventions
requires detailed planning in advance, which ought to be based on evidence for effectiveness
as well as for economic efficiency. Nevertheless, determining the cost-effectiveness is more
difficult for interventions with multiple components commonly found in the field of public
health. Therefore, this study presents the underlying conceptual issues that arose when we
modelled the costs and consequences of a hypothetical community-based intervention in an
Australian State. A foremost challenge was the highly context-dependent intervention content
that makes it difficult to determine an adequate resource inventory. In addition, there may be
economies of scale due to state-wide implementation and possible synergies between the
intervention components. The study explored relevant assumptions that would reflect these
influences on the resource use in an indicative set of intervention strategies. In terms of
consequences, projected impact on health and survival during adulthood bear considerable
uncertainty, whereas intermediate anthropometric outcomes expressed as mean reductions in
body mass index may lack decision-making relevance. Overall, there appeared to be little
guidance on how to contextualise a multi-faceted intervention in a model to assess cost-
effectiveness. A comprehensive framework is required in order to enhance priority setting on
public health interventions with multiple components.
75
6.1 Introduction
Childhood obesity is a public health concern in Australia and globally (Han et al. 2010,
Lobstein et al. 2004). It has been widely acknowledged that, due to a complex aetiology of
excess weight gain in populations, a multi-faceted preventative response is needed
(Kumanyika et al. 2008, Swinburn et al. 1999). Promising evidence has emerged from
interventions implemented at the community level, consisting of multiple prevention
components across different settings (Economos et al. 2007, Sanigorski et al. 2008, Taylor et
al. 2007).
Since past efforts at prevention have often been based on single short-term components with
limited or no evidence for effectiveness, multi-faceted programmes in the community thus
may appeal to decision makers. Nonetheless they require detailed planning in advance. A vital
decision aid in view of the ever-present scarcity of public resources is information on their
cost-effectiveness. This is commonly assessed through health economic evaluations that
model local circumstances based on interventions conducted elsewhere. This requires a sound
understanding of both costs and consequences in the original setting as well as about the need
for local adjustments. However, studies reporting on the cost-effectiveness of more
comprehensive programmes like community-based interventions (CBI) are scant with the
exception of one evaluation of a small-scale programme (McAuley et al. 2010). This gap
contrasts with the increasing recognition these CBI receive by the research community and
policy-makers (Swinburn and de Silva-Sanigorski 2010). Economic evaluation of trials to
prevent overweight and obesity has so far been mainly confined to school-based interventions
with distinct components (Brown et al. 2007a, Kesztyus et al. 2011, Wang et al. 2008, Wang
et al. 2003). A reason for this mismatch may be the challenges posed by the varied nature of
public health interventions. The usual health economic evaluation frameworks appear to be
more conducive to clinical settings and pharmaceuticals, where the boundaries of the
intervention are clearly defined (Schwappach et al. 2007).
Given the multiple facets of public health interventions, it appears prudent to consider a
disaggregated cost-consequences framework (Brousselle and Lessard 2011, Coast 2004). This
allows different dimensions of outcomes to be incorporated, including those with insufficient
understanding about long-term valuation, leaving it to the decision makers to apply their own
judgement (Mauskopf et al. 1998). Based on the data available on costs and effects, we
modelled the costs and consequences of a hypothetical CBI to prevent overweight and obesity
in children in an Australian state. The aim of the present study is to provide insight into
conceptual issues faced in the conduct of this analysis, as well as to present how challenges in
modelling a multi-faceted intervention could be addressed.
76
6.2 Model overview
The study sought to simulate the costs and consequences of an intervention to promote
healthy weight in children in an Australian state from a societal perspective to inform local
decision makers. This was to be based on the best evidence available for both costs and
effectiveness. The model was conceptualised under the assumption of a state-wide
implementation to represent a population-based prevention approach with the comparator
being no intervention.
Studies on effectiveness were identified through a systematic literature search and were
pooled according to their intervention components to obtain estimates measured as reductions
in mean BMI. Findings from this meta-analysis suggested evidence for interventions that
consist of multiple components that involve education, physical activity and nutrition mainly
in the school setting and also for interventions that aimed to reduce television viewing (TV)
(Luckner et al. 2011). In addition, the only economic evaluation studies identified alongside
this literature search were predominantly about school-based educational interventions with
little outreach to other parts of the community (Brown et al. 2007a, McAuley et al. 2010,
Wang et al. 2008, Wang et al. 2003). Nevertheless, these studies were used as a backbone in
order to base the resource inventory of the present work on a credible source. However, the
simulated programme was therefore a multi-faceted CBI with a main focus on the primary
school setting. Furthermore, the analysis was restricted to the present time horizon without
extrapolation to long-term consequences.
The specific tasks involved in this analysis were about defining this intervention content to
identify the resource quantities and to estimate the costs of a large-scale implementation with
local prices. Furthermore, it involved expressing intervention consequences in terms of
outcomes meaningful to the decision-making context. These tasks and the issues that evolved
around them are described in the following sections.
6.3 Estimating costs
6.3.1 Intervention content and resource inventory
In order to estimate the programme costs, a definition of what constituted the intervention was
needed. This required a precise understanding of the structure of the intervention as well as
the explicit activities involved so that an adequate resource inventory could be derived.
However, it quickly became clear that this was challenged by the nature of the intervention.
CBI to prevent obesity typically comprise a variety of strategies, with the content and
77
subsequently the resource use likely to be unique to the specific setting (Hawe et al. 2004).
Consequently, no generic programme could serve as a standard template to identify resource
quantities and to value these with local unit costs. Alternatively, if the average cost per
participant generated from a programme conducted elsewhere were to be used as a proxy, this
would offer little insight into what determines variation in the local setting, especially with
respect to different population groups. Therefore, an intervention was modelled that could at
least be understood as an indicative CBI. This involved first gaining a thorough understanding
about common characteristics of a CBI to derive a general structure for the intervention.
Secondly, a synthesis of the literature was conducted to add specific activities that could be
understood as plausible examples to allow detailed costing.
As a source to inform the design of this indicative CBI, peer-reviewed studies that evaluated
the effectiveness of CBI were compared for commonalities (de Silva-Sanigorski et al. 2010,
Economos et al. 2007, Sanigorski et al. 2008, Taylor et al. 2007). In addition, a list of generic
principles on CBI for obesity prevention was used as guidance with respect to relevance for
costing (King et al. 2011). In order to select indicative activities and specific strategies in the
school setting, a synthesis was made based on the effectiveness studies, the cost-effectiveness
analyses identified in the literature search and cost information provided by a recent
comprehensive economic modelling project (Carter et al. 2009, Haby et al. 2006, Moodie et
al. 2009a, Moodie et al. 2009b). The synthesis was organized in the following manner. Any
strategy selected had to be of general applicability to the context of the present study; that is it
was plausible that the processes involved could be implemented. In addition, sufficient
information had to be available for economic evaluation, defined as a clear understanding of
the resource inputs required for the strategy. If the strategy appeared complex, inclusion
depended on the availability of a peer-reviewed publication, but in the case of simple
strategies the resource inputs could be inferred from other sources. Cost-effectiveness
analyses that were used to inform the modelling of single strategies where not equally
consistent with scope of cost items considered, for example some had included catering and
venue hire where training for teachers or volunteers was needed, but others had not.
Therefore, we generally followed the source that provided the most comprehensive list of cost
items (Moodie et al. 2009a, Moodie et al. 2009b).
As a result, the following structure was assumed. The CBI would operate at state, community
and school level for three years, preceded by an implementation phase of six months. It was
furthermore assumed that some training and volunteer recruitment activities would need to be
repeated during the course of the routine operation. Activities at the first two levels were
mainly concerned with programme coordination and engagement of community members. At
78
the school level, components of the intervention were aligned to match effectiveness data
available from the meta-analysis and hence covered physical activity, education, nutrition and
reducing television viewing. An overview of the indicative strategies in these components is
provided in Table 6-1. Furthermore, Table 6-2 summarises the resources required for all
activities according to stakeholder and implementation level.
Table 6-1: Content of the hypothetical programme by intervention phase and by
implementation level
Intervention
phase
Level
Indicative Strategy/Activity
Ongoing
(3.5 years)
State
Steering committee: Four staff members located at the State Department of Health
manage the programme and are responsible for the social marketing campaign.
Community
Local coordination: In each local government council, a coordinator is employed (full-
time) who has the task to initiate, coordinate and oversee the activities in the
community. This includes meetings with programme stakeholders, support for
community capacity building and contributions to the parent forums.
Set-up
phase (6
months)
State
Social Marketing campaign: Development and production of resources of a state-wide
campaign.
Community
Community engagement: Programme planning and promotion meetings are held with
programme stakeholders in the community.
School
Canteen food improvement: Canteen staff members and volunteering parents receive
training (on site).
Water dispenser program: Installation of water dispensers in classrooms.
Physical activity events: Recruitment and training of volunteers.
Active transport: Recruitment and training of parents/volunteers.
Routine
operation
(3 years)
State
Social marketing campaign: The campaign promotes healthy life-style on TV, radio
and with posters in public spots. It also includes the promotion of an annual event to
reduce TV-viewing ("switch off" week).
Community
Community engagement: Continuous meeting with community stakeholders, but at
25% intensity compared to the set-up phase.
School
Physical activity coordinators: The coordinator spends 1.5 hours each weekday to
encourage children to be active during breaks and after school. A monthly blanket
amount covers ongoing expenses for materials and equipment needed.
Physical activity events: Regular one-day events are held on the school premises that
promote physical activity including competitions. Recruited volunteers who received
police checks and a liaising teacher are responsible for operating the event. Volunteer
recruitment and training continues at 25% intensity compared to the set-up phase.
Active transport: Children walk to school in a bus formation of eight students and two
adults. Recruited parents act as volunteers who receive police checks and training. A
liaising teacher helps with the coordination. The routes are assessed for safety.
Participating children receive a resource kit and the bus theme is also embedded in the
curriculum. Volunteer recruitment and training continues at 25% intensity compared to
the set-up phase.
Training for teachers: All schoolteachers receive training to enhance their curriculum
with health and life-style themes. Relief teachers compensate for the absence from
class.
Parent forums: Four times per year a two-hour forum is organised by a liaising teacher.
Speakers could be the local community coordinator, a general practitioner, a dietician
or the physical activity coordinator.
Social marketing campaign: Health promotion resources are distributed in schools.
Fruit program: Every student receives a free piece of fruit on every school day. The
fruit theme is also incorporated into the curriculum. A liaising teacher coordinates the
scheme.
Dietician support: A dietician dedicates one hour per school week to consult with
canteen staff and with students in class.
Education to reduce TV-viewing: A designated number of teachers receive additional
training to incorporate reduction of TV-viewing as a theme into the curriculum.
Routine monitoring: Liaising teachers conduct routine evaluation of activities every
school term.
79
Table 6-2: Cost items in the hypothetical programme by implementation level and stakeholder
Level
Indicative strategy
Parents and other volunteers
School and community
State and programme funds
School
Physical activity coordinators
Physical activity coordinator (7.5 hours per week incl. preparation
time)
Programme supplies (blanket amount)
Physical activity events
Time in recruiting session (20 volunteers; 2
hours + 0.5 hours travel time)
Time spent in event (5 volunteers; 8 hours +
0.5 hours for travel)
Travel-related expenses (charge per
kilometre)
Teacher’s time in events (1 teacher; 8 hours)
Budget for event expenses (blanket amount)
Police checks for volunteers (20 volunteers)
Active transport
Time spent in recruiting session (15
volunteers; 2 hours + 0.5 hours travel time)
Time spent in training session (8 volunteers;
2 hours + 0.5 hours travel time)
Time spent in route assessment (4
volunteers; 1 hour)
Time spent in routine operation (4
volunteers per week; 0.4 hours per return
route)
Travel-related expenses (charge per
kilometre)
Liaising teacher’s time (1 hour per week)
Trainer for volunteers (2 hours + 1 hour preparation)
Catering during training (9 portions)
Venue for training (venue hire)
Training materials (8 photocopies)
Road safety officer for route assessment (2 hours)
Resources for participating children (16 resource kits)
Curriculum materials (7 manuals)
Supplies for programme promotion (blanket amount)
Police checks for volunteers (8 volunteers)
Training for teachers
Trainer for teachers (8 hours + 4 hours preparation)
Venue for training (venue hire)
Catering (15 portions)
Training materials (14 manuals)
Funds for relief teaching (8 hours)
Parent forums
Time spent at forum (64 participants; 2
hours)
Travel-related expenses (charge per
kilometre)
Liaising teacher’s time (1 hour per week)
Reimbursement for forum speaker (2 hours + 0.5 hours for travel)
Venue for forum (venue hire)
Social marketing
Supply with marketing resources (blanket amount)
Canteen food improvement
Time spent in training sessions (15
volunteers; 48 hours)
Canteen staff time in training sessions (1 staff
member; 48 hours)
Trainer for canteen staff and volunteers (48 hours + 8 hours
preparation)
Catering (17 portions per training day)
Dietitian support
Reimbursement for school-dedicated dietitian (1 hour per week)
Fruit and water program
Liaising teacher’s time (1 hour per week)
Water dispenser instalment (blanket amount)
Daily supply with fruits (1 piece per day)
Teaching materials (212 booklets)
80
Level
Indicative strategy
Parents and other volunteers
School and community
State and programme funds
Education to reduce TV-
viewing
Trainer for teachers (3 hours + 1.5 hours preparation)
Venue for training (venue hire)
Catering (4 portions)
Training materials (3 curriculums)
Funds for relief teaching (3 hours)
Routine monitoring
Time dedicated by teachers (0.5 hour per
term)
Community
Local coordination
Community coordinator (1 staff member; full-time equivalent)
Engagement with community
Time of local professionals in meetings (70
participants; 18 consultations lasting 2 hours)
Time of other community stakeholders in
meetings (44 participants; 18 consultations
lasting 2 hours)
Venue for meetings (venue hire)
State
Steering committee
Steering committee (4 staff members; full-time equivalent)
Social marketing campaign (funds for production and media buy)
81
6.3.2 Large scale implementation
The programme costs of the indicative intervention could be estimated for a single
representative community, but it would be questionable to assume that these could be simply
added up to reflect state-wide implementation costs. The notion of non-constant returns to
scale implies that the relationship between resource input and output varies disproportionately
as the volume (i.e. level of population coverage) increases (Elbasha and Messonnier 2004,
Varian 1992). This is particularly relevant in a cost model, where the estimates of resource
utilization are obtained from economic evaluation of trials conducted in a small-scale setting,
whereas the model is concerned with assessing efficiency on a large scale.
Reasons for decreasing average costs could be the specialisation of labour that promotes an
ease of a large scale programme operation (Elbasha and Messonnier 2004), the spread of
fixed and semi-fixed costs through higher population coverage (Brooker et al. 2008, Mansley
et al. 2002) as well as increased bargaining power allowing cheaper material procurement
(Stevens et al. 2005). Conversely, a rise in average costs is also possible through an increased
workload on programme management, short-term shortages in resources and outreach to
remote areas (Johns and Torres 2005).
The possibility of non-linear costs is commonly neglected in economic evaluations (Elbasha
and Messonnier 2004, Stillwaggon 2009). However, incorporating non-constant returns
requires considerable knowledge about a program’s marginal and average cost curves beyond
the scope of a modelling study (Elbasha and Messonnier 2004, Over 1986). Rather than being
able to precisely quantify the impact of scale variations, the present study therefore explored
which plausible non-constant returns could be incorporated in the model as well as which
assumptions would be needed. In a first step, the costs of all strategies in an average school
and the involvement of members in an average community were estimated. This formed the
reference unit of the analysis. In a second step, these costs were multiplied by the state-wide
number of schools and communities without taking into account non-linearity. However, this
was further explored through four separate scenarios (Table 6-3).
The first scenario explored the possibility of economies of scale. The sharing of semi-fixed
resources between schools and the advantages of bulk procurement of materials were found
reasonable and feasible options to incorporate. This involved screening the single cost items
in each strategy to make plausible adjustments. The second scenario considered diseconomies
of scale due to an increase in project management load using a range of values. The rationale
for such assumptions was always to obtain an empirical source as a reference to enhance
validity.
82
Large-scale determinants were also considered likely to impact on the total intervention costs.
For the sake of this study, these were restricted to geographic and socio-economic variation.
The third scenario estimated differences in intervention content and price levels in rural and
remote areas. The last scenario was concerned with modifications to the costs in schools
located in areas of high socio-economic disadvantage.
Table 6-3: Overview of adjustments in each scenario for large-scale implementation
Scale assumption A
Scale assumption B
Geographic Variation
Socio-economic variation
1) Reduction in programme
costs due to shared resources
between schools and
communities (semi-fixed costs)
a) Assume that two schools
share training sessions for
teachers and parents/volunteers
b) Assume two schools share
one physical activity
coordinator
c) Assume two schools share
one parent forum
d) Assume two communities
share one local coordinator
2) Reduction in programme
costs due to a bulk
procurement advantage
a) Assume a 10% reduction for
all materials and equipment
Higher work load due
to large-scale
operation leads to an
increase in
programme
management costs
(e.g. increase in
communication and
travel costs)
a) 5%
b) 10%
c) 25%
Assume adjusted resource use
for schools in outer regional
and remote areas:
a) No active transport operates
in these areas
b) Higher travel cost for
parents due to longer
distances (x17)
c) Teacher coordinates sport
instruction instead of physical
activity coordinator
d) Higher cost for teaching
and promotion material due to
transport surcharge (10%)
e) Higher cost for fruit supply
due to transport surcharge
(20%)
f) Higher travel costs for
school dedicated dietitian due
to longer distances (x2)
g) Lower coordination needs
(half-time equivalent only) in
the community council in
remote areas due to lower
population density
Assume for schools located
in areas of high socio-
economic disadvantage:
a) Higher time load for
teachers, dietitian and
physical activity
coordinators (x2)
b) Fewer volunteers
recruited for physical
activity events (replaced by
teachers)
c) Lower participation in
parent forums
d) Higher need for social
marketing resources (x2)
6.3.3 Synergies
The intervention was modelled to comprise several major components, which were further
broken down into strategies. However, the content and the cost data sources were derived
from economic evaluations of strategies conducted in isolation. The concept of economies of
scope suggests that the costs of producing a number of different services through the same
infrastructure could be lower than the total costs of producing each individually (Panzar and
Willig 1981). The design of the modelled intervention accounted for shared management-
related costs, but additional resource synergies between the different strategies in the
intervention at the school level could be explored. As with the approach for scale variation,
this consideration also employed a separate scenario. Consequently, single cost items were
screened for possible overlaps between strategies. It was found that training for teachers
across different strategies as well as their time spent for routine monitoring could be further
83
merged in this scenario. For example, this reduced training required from eleven to eight
hours saving resources for trainer remuneration, relief teaching, venue hire and catering.
6.3.4 Resource mobilization
Capturing the broader perspective of public health interventions in an economic evaluation
also involves outlining to whom costs and consequences of an intervention accrue
(Drummond et al. 2008). Since obesity prevention concerns different groups in society, who
may be required to contribute to preventative solutions, the cost model further distinguished
between intervention stakeholders, which are defined here as those groups involved in the
intervention who contributed tangible resources. It has been suggested that the division
between costs covered by the project funds and those imposed on intervention stakeholders
and on participants would represent the resources mobilised by the intervention (Johansson et
al. 2009). This distinction also helps in gaining an understanding of potential barriers to a
successful implementation from a policy maker’s perspective and may indicate a possible
need for incentives to promote contributions within the community (Weatherly et al. 2009).
Stakeholder groups identified in this case study were the State, the community, the school, the
parents and other volunteers. Further distinction could also be made about what type of
resources stakeholders contribute, for example whether those are direct financial contributions
or indirect costs such as time of parents and volunteers (Johansson et al. 2009). However, the
divide in this model between project funds and stakeholders in terms of costs was not always
obvious, because this may in some cases depend on actual programme execution.
Nevertheless, the resources mobilized represent an easily obtained indicator that added more
general understanding to the intervention about the involvement of societal groups.
6.4 Estimating consequences
6.4.1 Intermediate outcomes
Health economic evaluations commonly express outcomes as an impact on a generic utility
score in QALYs (or less often DALYs), which is comparable across interventions and
conditions. However, the impact of obesity on health-related quality of life at child age is not
sufficiently clarified yet (John et al. 2010). In addition, in the case of obesity prevention,
major health consequences are expected to occur during later stages in life. However, this
requires projection of the prevented weight gain during childhood to estimate changes in
morbidity, quality of life and survival during adulthood. Therefore, long-term outcomes of
prevention during childhood need to be based on strong assumptions, but there is contention
84
as to what degree these are justified (Dalziel and Segal 2006, Segal and Dalziel 2007). Given
the modest effect sizes so far achieved in obesity prevention a conservative approach is to
restrict an economic evaluation to the present time horizon based on intermediate
anthropometric outcomes. Despite it may not be the most valid obesity indicator, BMI is
known as an acceptable proxy on a population level (Dietz and Robinson 1998).
In the present study, effectiveness data were obtained from the meta-analysis as a reduction in
mean BMI representing an intermediate continuous outcome estimate, but which may by
itself be lacking decision-making relevance. Hence, the effect on the prevalence of
overweight and obesity was approximated by imposing this reduction in mean BMI on an
Australian BMI distribution obtained from a representative survey. However, it should be
noted that the cut-off points are based on an arbitrary definition for weight categories and
moreover, the impact on the BMI distribution through prevention is possibly not represented
adequately. It is indeed more likely that the BMI distribution becomes less skewed rather than
shifting to the left as suggested in the classic work by Geoffrey Rose (Rose 1992, Walls et al.
2009, Walls et al. 2010). Nevertheless, how the distribution’s shape might change remains
unknown based on current data.
6.4.2 Other outcomes
Despite the model being confined to body composition indicators we recognise the
appropriateness of other of outcomes that are relevant consequences These are spill-over
effects to community members who were not targeted by the intervention, but who are
indirectly also affected by it. This could be an improved physical activity level for parents
who participate in the active transport arrangements. In addition, a CBI generally aims to
provoke a change affecting the whole setting rather than individual behaviour alone.
Therefore, further outcomes to consider would be changes in practices and policies that
indicate the impact on values and norms in the community (Trickett et al. 2011). As of yet,
outcomes that emerge beyond the individual level perspective are not commonly found in
economic evaluations (Shiell et al. 2008). It is even less clear, which other outcomes are
relevant to be included in a prospective modelling study.
6.5 Discussion
The present study has illustrated how costs and consequences were estimated for a
hypothetical CBI to promote healthy weight in primary school children in an Australian state.
Issues were explored on how to derive a model applicable to the context under the constraint
of availability for evidence on effectiveness and information on economic efficiency. Hence,
85
the strategies in the model served as examples that could be replaced once more relevant
information becomes available.
Economic modelling of future interventions is becoming more and more important in health
policy planning, because it provides a consistent framework to synthesize the information
relevant in setting priorities (Leidl 2010). However, there appears to be little guidance on how
to contextualise a multi-faceted intervention in a model. How an intervention needs to be
implemented in order to work is generally not of major concern in economic evaluations
when the intervention can be sufficiently standardised (Shiell et al. 2008, Walker et al. 2010).
Conversely, CBI are probably best understood as events in complex systems with dynamic
interactions (Hawe et al. 2009). Where the interaction between content and context of an
intervention is likely, but not sufficiently understood, this becomes an impediment to
estimating the resource input. In order to overcome that, the present study instead followed a
common reductionist approach by breaking down the intervention into distinct components
and strategies, but it was also driven to explore why a standard set of strategies would differ
between communities. These different scenarios served as a lens to understand the multi-fold
variations likely to occur in a complex intervention. Nevertheless, this was explorative in
nature and more work is needed to develop conventions for variations as well as empirical
data on context-specific cost determinants i.e. from multi-site CBI studies.
As more economic appraisals of CBI need to be conducted to enrich the evidence base about
efficiency, there should also be more consistency in the cost categories included, especially
with respect to indirect costs involving all stakeholders like their time dedicated to the
intervention. Generally, the more agencies involved in the intervention the more difficult it
gets to identify all cost-related processes (de Salazar et al. 2007). Nevertheless, there is a need
to make underlying processes involving resources more transparent, especially about
partnerships between organizations and engagement with community members. A resource
mobilisation analysis presenting all contributions by group of stakeholders could help to
match the increasing interest in inter-sectoral consequences of public health interventions
(Johansson et al. 2009).
There is a need to incorporate a range of outcomes into the consequences side of any
economic evaluation of CBI as there are likely to be impacts on multiple levels of the system
beyond the individual as well as different degrees on statistical confidence that can be placed
around them. Extrapolation of effects into adulthood offers valuable insights into the overall
objectives of prevention, but with an unknown likelihood of that becoming reality. Therefore,
meaningful intermediate endpoints for health economic evaluations on obesity prevention in
children are needed. This will require a better understanding of the impact on the shape of the
86
BMI distribution and on health-related quality of life as well as pertinent healthcare cost
offsets at child age (John et al. 2010). Furthermore, activities such as parent participation in
active transport could be viewed as a long-term consequence rather than as a programme cost
and ideally this life-style modification would continue beyond the intervention time frame of
three years. One question that arises in an evaluation that projects future outcomes beyond
intervention ending is whether there are opportunity costs in these ongoing activities that
would still need to be considered. This would not be the case if parents changed their attitude
throughout the intervention, so that they receive utility from active transport participation that
exceeds their loss of time (Krauth et al. 2011). However, these notions would benefit from
more empirical investigation.
The concept of synergetic effects is considered to be at the core of a multi-faceted health
promotion intervention, but as multiple components are increasingly being used within one
intervention, researchers have thus found it more difficult to identify the ‘active ingredients’
(Brownson et al. 2009). This may obscure whether the intervention comprises some
components with very little value for money. The contribution of single components and
strategies need to be better understood as well as synergies with respect to both costs and
consequences through adequate process evaluation.
The bottom-up approach to costing offered the necessary flexibility to explore variation,
because it allowed reflection over single resource items as to how they could be affected by
the scenario and clarified where more precise information would be needed to inform a highly
contextualized model. Despite the available cost-effectiveness analyses providing good
guidance for building a resource inventory, a limitation was still that quantities of resources
require a more detailed understanding of local practice and capacity. In addition, notions
related to complexity science became apparent in this study, but this should be explored more
thoroughly for economic evaluations of CBI.
6.6 Conclusion
Setting up this model helped to enrich the understanding what is currently known about the
costs and consequences of multi-faceted obesity prevention. Modelled cost estimates are
challenged by little information available on what determines resource use in different
settings and in addition deriving a standardised cost package necessary to project large-scale
implementation appears to contradict the ‘organic’ nature of a CBI. Long-run consequences
of prevention during childhood are surrounded by much uncertainty, but intermediate
outcomes need to be expressed with more applicability for the decision-making context. More
focus should be given to outcomes measured beyond the children targeted. The literature has
87
already pointed out the absence of an overall framework to capture economic value of public
health interventions (Owen et al. 2011, Shiell 2007, Weatherly et al. 2009). Substantially
more guidance is required on both the conduct of economic evaluations of primary trials as
well as for modelling studies to inform priority setting for obesity prevention.
88
7 Modelled Costs and Consequences of Multi-faceted Obesity
Prevention in South Australia
7.1 Supplementary details on the calculation
7.1.1 Overview
Chapter 6 thoroughly described the concept of the economic evaluation. The task of this
chapter was to present and discuss the results of the cost-consequence analysis. Prior to this,
supplementary information specific to the calculation is presented in an attempt to illustrate
the way in which results were derived.
In the cost-consequence analysis, societal costs and effects of a hypothetical programme were
assessed and compared to a complete lack of intervention. This comparator was chosen since
there were no relevant concerted prevention efforts at that time in South Australia. The
evaluation timeframe was restricted to a three-year intervention period, plus a half-year set-up
phase during 2010, which was defined as the base year. The discount rate applied to both
costs and effects was 5%, which is the common recommendation for economic evaluation in
Australia (DoHA 2008a).
The research tasks in the present study involved the following steps to obtain base case
estimates:
• Intervention content: synthesis of the literature to define the applicable programme
content
• Effectiveness: approximation of the population impact of a reduction in mean BMI
after intervention ending
• Cost model: the transfer of resource inventories from published sources; assigning
local prices to the identified quantities and estimation of intervention costs for a
large-scale implementation.
The base case estimates were derived based on the information available in the published
literature, and assumed costs to be the same in every community and every school across the
State. However, in order to understand the way in which estimates could be influenced by the
context of the intervention, a number of separate scenarios were used to explore the options
for plausible modifications, as well as their impacts on the results.
89
7.1.2 Intervention content
A detailed description of the intervention content and how it was derived is provided in
Chapter 6, particularly in Table 6-1. For illustration purposes, Figure 7-1 depicts the
simplified structure of the intervention.
It was assumed that central administration would be the responsibility of a committee of
senior staff members at the Department of Health in Adelaide, and that local coordinative
support would be provided at the community level. For the purpose of this study, a
community in South Australia was defined as one of the 68 local government areas (LGAs),
although, in reality, not all LGAs necessarily exhibit typical characteristics of a single
coherent community.
Given the policy-making perspective of this analysis, it was assumed that all 489 government-
run primary or combined primary and secondary schools in the state would adopt the
programme, which would imply a population coverage of approximately N= 100,000 students
annually (ABS 2010c, Productivity Commission, 2011). In South Australia, there are eight
primary school grades (Reception and years 1–7), which includes children aged 5–13 years.
Figure 7-1: Structure of the indicative community-based intervention
STATE
- Central
coordination
- State-wide social
marketing campaign
COMMUNITY
- Local coordination
- Programm
promotion
- Engagement with
stakeholders
SCHOOL
PHYSICAL
ACTIVITY
- Activity
coordinators
- Events
- Active transport
SCHOOL
EDUCATION
- Training for
teachers
- Parent forums
- Social marketing
SCHOOL
NUTRITION
- Canteen
improvement
- Fruit and water
program
- School-dedicated
dietitian
SCHOOL
TV
- Training for
teachers
90
7.1.3 Effectiveness
A main finding in the meta-analysis in Chapter 2 was that interventions that combined
physical activity with health education and modification to nutrition, mainly within the school
setting, resulted in a mean BMI reduction by -0.1 kg/m2 (95% CI [-0.17, -0.04]; I2 = 5%; N=
10,257). This outcome estimate was used in the present study under the assumption that such
a multi-component intervention could serve as proxy for a CBI in South Australia.
Furthermore, since a reduction in mean BMI as an outcome may lack interpretability in a
decision-making context, its potential population impact was approximated. This was defined
as the number of additional children no longer above the normal weight range as a result of
the intervention. The reduction in mean BMI had to be translated into a corresponding
reduction in prevalence of overweight and obesity, as can be seen in Figure 7-2. Accordingly,
the reduction of -0.1 kg/m2 was imposed on BMI data from a recent Australian cross-sectional
survey in children that included anthropometric measurements obtained through computer-
assisted personal interviews (The 2007 National Children’s Nutrition and Physical Activity
Survey). Permission to access the survey for this purpose was kindly granted by the
Australian Social Sciences Archive.
The BMI measurements were available for N= 4,787 children between two and 16 years of
age; however, merely the BMI data of children between seven and 13 years were used here
(N= 1,599) since only these would have had full three-year intervention exposure. This
referred to the average intervention length of the meta-analysed primary studies
(approximately 2.6 years).
The prevalence of overweight and obesity was determined based on the IOTF age- and
gender-adjusted cut-off points using the LMS Growth Microsoft Excel module (Cole et al.,
2000, 2007). The prevalence was obtained for both the modified and unmodified data set. The
difference in prevalence was then applied to children with three years of exposure to the
intervention in an attempt to obtain the number of cases that were no longer above the healthy
weight range.
91
Figure 7-2: Shift in the BMI distribution as a result of the intervention
7.1.4 Cost model
Societal costs were estimated according to published sources for each intervention
component, as described in Chapter 6. An ingredients approach to costing was used, which
first involved the identification of all resource items in the intervention, and secondly valued
them with unit costs from South Australia (Tan-Torres et al., 2003). The resource quantities
in each strategy were either adopted as the exact amounts described in the original study, or
were otherwise modified where there was reason to believe that this was more applicable. All
data sources and assumptions for transfer are documented in Appendix 8.
Unit costs in Australian Dollars for the year 2010 were collected from South Australian
sources where this was possible; otherwise, national prices were used and, in single instances
where no local estimate could be obtained, overseas prices from original studies were taken
and converted to AU$ (CCEMG – EPPI-Centre Cost Converter (v.1.2)).
In many cases, the costing approach of the ACE-Obesity project was followed, with such
patterns also applied to strategies where relatively little was known about the resource use
(Carter et al., 2009). For instance, the societal costs related to involving volunteers were
estimated in the same manner as described in Moodie et al.: volunteers would need to be
recruited, trained, and receive police checks (Moodie et al., 2009a). Their training would
involve the time of a trainer, the venue hire, catering, and the provision of training materials.
Furthermore, time- and travel-related costs would accrue for every intervention-related
activity of the volunteer. The examples of published economic evaluations from ACE-Obesity
were also followed in regard to factors for salary related on-costs and office use (Moodie et
al., 2009a; Moodie et al., 2009b).
BMI
% of
Population
Overweight or obese
Normal weight
Underweight
Pre-Intervention
Post-Intervention
92
First, costs were estimated for one average primary school and community in South Australia.
The costs of these reference units were then multiplied by the total number of schools (N=
489) and communities (N= 68) respectively, and state-level costs were added. The average
primary school was approximated to comprise 212 students and to employ 13 teachers, with
90% of schools running a canteen with one full-time employee. The average community was
assumed to employ one coordinator and involve, on average, 70 members to be included in
promotion and planning meetings. Such units of reference are approximations only, given the
variation between the sites in reality. Detailed information on the sources for this calculation
is provided in Appendix 8.
Efficient use of all resources was assumed, even though this was not always clear from the
original studies. This meant, for example, that staff would act on the best practice principles
without learning curves, except where training to improve skills was explicitly part of the
intervention. Furthermore, resources related to research beyond routine monitoring were
excluded because they were irrelevant to the decision-making context. Time costs were
included for adults, but not for children.
Costs were categorised regarding their time horizon (one-off, ongoing, or recurrent), their
variability with scale (variable, fixed, semi-fixed), and if annuitiszing was applicable (capital
investments during set-up phase). In the absence of information relating to recurring events,
such as training refreshment sessions and the annual recruitment of volunteers, it was
assumed that this would occur at a reduced intensity of 25% compared with the start-up
phase. Capital items were annuitized based on a 5% discount rate and a useful life-span of
four years. Finally, equivalent annual costs were estimated, the results of which were
expressed in the intervention as costs per child per year.
7.1.5 Resources mobilised
CBI is characterised by the multiple stakeholders who either participate in the operation or at
least have an indirect involvement with the programme. It has been suggested that the
division between costs covered by the project funds and those imposed on intervention
collaborators and on participants would represent the resources mobilised by the intervention
(Johansson et al., 2009). In this context, stakeholders were defined as those groups involved
in the intervention that contributed resources other than those covered by the specific project
93
funds of the CBI. These were the state16, the community, the school, the parents, and other
volunteers.
7.1.6 Scenarios
The scenarios dealing with economies of scale and scope were described in detail in Chapter
6. Additional information is therefore presented here concerning scenarios on geographic and
socio-economic variations. Detailed information on the sources is also provided in Appendix
9.
The base case calculation assumed that there was no difference between schools and
communities; however, despite being the country’s third largest state, South Australia has a
resident population of only 1.6 million, 71% of whom live in the Adelaide metropolitan area.
The rest of the state is sparsely populated (ABS, 2010b); therefore, geographic location is a
plausible determinant for the variation of costs. An established classification scheme was used
in order to determine the number of schools in each geographic category (metropolitan, inner
provincial, outer provincial, remote, and very remote), and the respective average schools
sizes (MCEECDYA, 2009).
In addition, it was recognised that the socio-economic environment of the local setting could
also have an impact costs and effects; hence, the number of schools with high socio-economic
disadvantage in South Australia was obtained from a recent classification
(South Australia Smarter Schools National Partnerships). Furthermore, the impact of lower
levels of intervention uptake was also explored in children from low-income families, as
suggested in a recent modelling study in New Zealand (Mernagh et al., 2010); therefore, in
separate scenarios, the ways in which effectiveness outcomes would change were tested, i.e.
whether the uptake would be reduced by either 5%, 10% or 25%. This referred to
approximately 23% of all primary school students in South Australia (PHIDU, 2008).
7.1.7 Parameter uncertainty
Throughout the analysis, there was careful documentation of uncertainties surrounding the
data collected and the assumptions made subsequently. Originally, a multivariate sensitivity
analysis was being planned; however, the parameters used to construct the resource inventory
were mostly proxies without known upper and lower limits; hence, in a considerable number
of cases, no credible judgement could be derived concerning the distributions surrounding
16Resources contributed by the state as opposed to project funds were those that were not specific to the intervention (e.g. police
checks for volunteers).
94
such parameters (Claxton, 2008). Therefore, in addition to the scenarios described above,
one-way sensitivity analysis was conducted.
In order to test the sensitivity of intervention costs, all pertinent parameters with unknown
ranges were varied by +/-50% unless specific values for upper and lower limits were
available. A tornado diagram was used to display the impact on annual intervention costs per
child.
Effectiveness variability was tested by replacing the outcome estimate for multi-component
interventions with the effect obtained from studies that promote a reduction in TV-viewing (-
0.27 kg/m2 (95% CI [-0.4; -0.13]; I2= 20%; N= 3,962) (see chapters 2 and 3).
7.2 Results
7.2.1 Intervention consequences
Effectiveness in the base case:
• The intervention would reach approximately N= 100,000 children every year in
public primary schools.
• After three years, approximately N= 75,000 children between seven and 13 years of
age would have had full intervention exposure.
• The unmodified prevalence of overweight and obesity combined would be 24.7%.
• The reduction in the combined prevalence of overweight and obesity was
approximated to be -0.75% points. This remained the same when discounting the
effect estimate on mean BMI.
• This resulted in N= 563 additional children no longer being above the healthy weight
range as a consequence of the intervention.
7.2.2 Intervention costs
Intervention costs in the base case:
• Total societal costs of the intervention were estimated to be AU $166,947,510 in the
base case and AU $146,256,995 when discounted.
• Expressed in 2010 price levels, the equivalent annual costs per child in the
intervention would be AU $415 or AU $84,863 per school.
The majority of the total costs (approximately 85%) would be incurred by the strategies at the
school level (see Table 7-1). The fruit programme would be the single most expensive
strategy, and notably would represent approximately one-third of the total costs. This would
95
be followed by employing physical activity coordinators, and by training in canteens for staff
and volunteers. Local coordination in the community would represent the highest cost share
outside of schools.
Table 7-1: Estimated costs of the indicative intervention by single strategies
Intervention strategy
Costs in AU $ in 2010
(discounted)
% of total intervention costs
Physical activity coordinator in schools
$17,831,140
12%
Physical activity events
$14,806,410
10%
Active transport
$5,793,009
4%
Training for teachers
$6,978,739
5%
Parent forums
$12,500,745
9%
Social marketing in schools
$126,826
0%
Canteen food improvement & Dietitian support
$16,689,978
11%
Fruit and water programme
$47,623,109
33%
Education to reduce TV-viewing
$1,342,786
1%
Routine monitoring
$300,577
0%
Local Coordination
$18,014,413
12%
Engagement with community
$1,069,768
1%
Steering committee
$1,453,033
1%
State-wide campaign
$1,726,463
1%
Total costs
$146,256,995
-
7.2.3 Resources mobilised
The division between project funds and stakeholders groups was found to be as following:
• Approximately 74% of the intervention costs would be project-funded.
• Parents and other volunteers would contribute 16.4% of resources.
• The schools would contribute 8.8% of resources.
• Community and state level contributions would be less than 1% each.
Approximately 77% of all the costs would be financial, whereas the remaining 23% refer to
other resources, such as time and travel costs incurred by parents, teachers and volunteers that
are additional to their every-day routine.
7.2.4 Scenarios
Table 7-2 contains the scenario results. Economies of scale and adjustments made for schools
with socio-economic disadvantages generated the highest impact on the annual cost per child.
In addition, assuming lower levels of intervention uptake would notably decrease the
additional number of children in the healthy weight range.
96
Table 7-2: Impact on costs and consequences under scenario assumptions
Scenario
Equivalent
annual cost per
child in AU $ in
2010
(discounted)
Change to base
case (%)
Number of additional
children no longer
above the healthy
weight range
LARGE SCALE IMPLEMENTATION
Bulk procurement & shared resource use
Increase in management load 10%
Increase in management load 25%
Increase in management load 50%
$366
$420
$429
$442
-11.7%
+1.3%
+3.3%
+6.6%
SCOPE EFFICIENCY GAINS
$406
-2.1%
GEOGRAPHIC VARIATION
Adjusted school size
Adjusted components, resource quantities
and prices
$410
$406
-1.2%
-2.2%
SOCIO-ECONOMIC VARIATION
Higher staff load, lower parent participation
rate
5% lower intervention uptake
10% lower intervention uptake
25% lower intervention uptake
$459
+10.7%
553
543
513
7.2.5 Parameter uncertainty
The parameter variations with the strongest impact on the annual costs per child are displayed
in the Tornado diagram (Table 7-3); all were part of the most resource-intense strategies (i.e.
fruit programme, employment of PA coordinators, and canteen staff training). When
eliminating the most expensive strategy in each component of the programme, the estimated
annual costs per child were reduced to AU $195.
In addition, the lower and upper bound values of the 95% CI around the BMI effect estimate
were also tested. The reductions in the prevalence of overweight and obesity were estimated
as –0.9% and –0.3% points, respectively. The corresponding number of children not above
the healthy weight range would be N= 704 and N= 188.
Furthermore, when applying a reduction in mean BMI obtained from pooling studies aimed at
reducing TV-viewing, the prevalence of overweight and obesity was lowered by -1.6% points,
thereby resulting in N= 1,173 additional children no longer being above the healthy weight
range.
97
Figure 7-3: Tornado diagram of selected parameter variations with notable impact on
annual intervention cost per child (AU $ in 2010)
$300 $400 $500
No. of fruit serves per week
[1; 5; 10]
Value of volunteer hour
[$0; $11; $16]
No. sport sessions per week
[1; 5; 10]
No. coordinators per community
[0.5; 1; 2]
Canteen staff training time (min)
[1440; 2880; 4320]
No. activity coordinators per
school
[0.5; 1; 2]
Price of fruit [$0.62; $0.77;
$0.92]
No. canteen staff trained
[0.5; 1; 2]
Intensity training and recruitment
(ongoing) [10%; 25%; 75%]
Parameter
[minimum; base case; maximum]
98
7.3 Discussion
The research in this chapter had the purpose to estimate the costs and consequences of a
hypothetical multi-faceted obesity prevention programme in primary schools and
communities in South Australia. Such a programme would have the potential to reach a large
share of the target population, i.e. primary school children; however, the findings indicate that
this would also come at a considerable cost, whilst the expected consequences on the
prevalence of obesity could be rather modest. Considering context-specific determinants and
assessing their impact enhanced the understanding gained of our findings. Importantly, it
appeared that intervention costs were relatively robust with respect to modifications and
parameter variation, whereas the number of additional children no longer above the healthy
weight range was very sensitive to the magnitude of the effects on BMI. The costs were
particularly influenced by resource-intense strategies, such as the fruit programme and the
employment of physical activity coordinators in schools. It needs to be taken into
consideration that a CBI may have spill-over effects towards societal change to adopt
healthier lifestyles in the whole population. Nevertheless, even with this additional objective
taken into consideration, the costs per child estimated in this study remain a relatively high
recurring expenditure, and contention may arise concerning whether this provides value for
money.
If these findings were to be expressed in an ICER, this would yield AU $73,749 equivalent
annual costs per additional child no longer above the healthy weight range. Unfortunately,
little is known about the meaning of this intermediate outcome in terms of QALYs for
children, and thus, the findings could not be expressed in such a comparable metric (Tsiros et
al., 2009). The scope through which the findings can be compared with other studies was also
limited due to the small number of published economic evaluations on obesity prevention,
which so far mostly focused on single-component interventions. Only the APPLE programme
incorporated several facets and the involvement of the community (McAuley et al., 2010).
Annual intervention costs per child in the study were AU $740 (adjusted for year and
currency), thus indicating that the present study’s finding, AU $415, may be within a
plausible cost range for a multi-faceted CBI. However, it should be noted that the APPLE
intervention comprised fewer components, and was implemented on a considerably smaller
population scale (N= 279 children).
Sorting the costs according to the type of stakeholder indicated that a considerable fraction of
the resources dedicated would be borne by individuals, families, and schools. However, this
estimated quarter of the total intervention costs might only be the lower bound, since this was
99
still a rather broad categorisation. It is conceivable that, in reality, more of these costs would
be shifted onto other stakeholders, depending on the execution of the actual programme.
In general, conducting resource mobilisation analysis requires precise cost data to be available
for each stakeholder group (Johansson & Tillgren, 2011). Understanding the imbalances of
contribution and benefits could provide policy makers with an indication of where incentives
are needed (Weatherly et al., 2009); thus, future economic evaluation studies should be
explicit concerning the resource impacts of an intervention on different groups within society.
This research was based on a large synthesis of studies, reports, and other relevant sources. A
main discovery made throughout the process of collating this information was the striking
paucity of materials readily available to inform economic modelling. The numerous obesity
prevention studies so far published contain little explicit reporting on cost ingredients; others
have also noted that the comprehensiveness of CBI challenged adequate costing in on-going
trials (Haby et al., 2012). Moreover, complex interventions, such as this modelled CBI,
involved a variety of processes markedly difficult to capture and standardise, such as
fostering of networks and partnerships. Clearly, those activities do incur opportunity costs.
The present study was only able to embody this notion by assuming regular meetings with
community members to be held with a lower intensity over the intervention time-course.
Nevertheless, this may only represent the tip of the iceberg, and so the incorporation of such
information in future programme evaluations is warranted.
The study also sheds light on the challenge of estimating the costs of a complex intervention
applied to a complex setting that is not one particular school, but a state-wide implementation.
Generalising the features of settings across the large scale is inevitable in such an analysis;
however, it should be recognised that the needs and the capacity of schools and communities
vary considerably in reality. The comparability of future research may benefit from a more
systematic approach to deriving plausible assumptions for this purpose.
The review of the methods literature in preparation for this study revealed how
underdeveloped this important area appears to be. For instance, from an economic
perspective, it is relevant to consider the notion of scale economies for a state-wide
implementation, although little guidance currently exists concerning the ways in which its
actual incorporation into the analysis should be conducted. Subsequently, the approach to the
scenarios was based mainly on intuition, and thus must be understood as an exploration of the
options available for relevant assumptions. Accordingly, future research should be directed
towards garnering an understanding of the influential contextual factors studied in the
scenarios; this includes the development of conventions on how these may be incorporated
100
within economic models, as well as obtaining empirical estimates, e.g. for non-constant
returns to scale in public health interventions.
The foremost limitation in this study was the use of proxies for the intervention content, cost
components, and effects. The intervention strategies chosen may not represent current state of
the art of CBI research, as this is increasingly shifting towards capacity building and
strengthening partnerships rather than running resource-intense activities (King et al., 2011).
However, this study was constrained to design the intervention content based on available
information for economic evaluation, which was a small number of published CEAs and
CUAs. In addition, strong assumptions regarding transferability were needed when
generalising the resource use from these studies.
Another limitation was the crude approach to modelling a reduction in prevalence of
overweight and obesity. Imposing the reduction in mean BMI on the whole distribution
equally ignores how the weight categories of children in particular may respond differently to
the intervention. Rather than the whole BMI distribution shifting to the left, it is conceivable
that there may instead be a reduction of its skewness to the right. However, the extent to
which the shape of the distribution changes is unknown based on current effectiveness data.
The estimated 0.75% point reduction in the prevalence of overweight and obesity may be
reasonable on a population level: for instance, a hypothetical 1% point reduction in the
prevalence of overweight and obesity was assumed in two American studies, both of which
estimated the potential long-term healthcare savings for children and adolescents (Trasande,
2010; Wang et al., 2010).
Furthermore, the approach taken in this study inevitably increased the number of children in
the underweight category. Given that the current evidence base for obesity prevention in
children does not suggest major adverse events, this aspect was not further considered in this
analysis (Waters et al., 2011a).
A further limitation when deriving the change in prevalence through a reduction in mean BMI
were the survey data used, which had only a 41% response rate, and which may have been
biased towards households with higher socio-economic status, as well as in regard to cultural
variables (DoHA, 2008b). However, the survey was the only recent source of population-
based BMI assessments that included children from South Australia.
7.4 Conclusion
A modelled intervention with multiple facets to prevent obesity in children was found to be
highly demanding in resources, also with regard to the contributions for parents, schools and
101
volunteers. As the corresponding level of effectiveness was estimated as modest, these
findings represent a challenge for decision makers.
A prior upfront understanding concerning an intervention will become increasingly important
when striving to enhance the accountability of decisions over the public dollar. The role of
health economics in this process is to identify all the costs and consequences, including by
whom they are borne, as well as to enhance the understanding surrounding influential factors
and uncertainty. This also applies to a CBI to prevent overweight and obesity, although their
complexity inevitably makes this a challenging task.
This study aimed to identify those cost determinants and stakeholders involved based on the
information currently available in peer-reviewed journals. As this was still fairly limited, it is
recognised that findings are probably only indicative of what can truly be expected as cost
and consequences.
102
8 Discussion
Summary of findings
This thesis pursued the aim of identifying promising interventions to prevent overweight and
obesity in children in order to inform the decision-making process in South Australia. This
involved the quantitative and narrative synthesis of evidence, and an economic evaluation of a
modelled set of interventions.
In terms of the best evidence available for effectiveness, the main findings obtained through
comprehensive meta-analyses based on 68 studies were statistically significant reductions in
mean BMI for two sets of mainly education-based intervention types, although effect
magnitudes were generally modest. These two types were multi-faceted programmes in the
school setting, and interventions that promoted a reduction in TV-viewing alongside other
strategies. However, when further investigating the role of the TV-reduction component, its
contribution to intervention effectiveness could not fully be understood.
In addition, a small number of economic evaluations alongside primary studies could be
identified in the systematic literature search for the meta-analyses. Although programmes
evaluated were generally considered to be cost-effective, particular challenges with respect to
expressing adequate outcomes became apparent as a result of the critical appraisal.
The systematic review in Chapter 5 provided an outlook towards the emerging evaluation
research on environmental approaches for obesity prevention in children. There was
promising evidence for policies that set standards for nutrition in the school setting to
improve eating behaviour; however, understanding the effectiveness of policies in the school
setting was problematic due to the predominantly observational study designs. Furthermore,
evidence for an impact on anthropometric outcomes was still limited
When estimating the societal costs and the consequences associated with a hypothetical multi-
faceted programme for South Australian communities, it was found that this would be highly
resource-demanding in return for low-population impact. This analysis also particularly
highlighted the challenge of conceptualising the economic evaluation of complex
interventions with multiple components.
The implications arising from these findings concerned three areas: the foremost one was
centred on recommendations derived for South Australian decision makers; in addition, the
findings also provided insights relevant to future of obesity prevention trials; finally, the
methods used to synthesise evidence on public health interventions was also noteworthy.
103
Obesity prevention in South Australia
The implications for South Australia ought to be considered in the context of current local
prevention efforts. In the year 2010, a multi-faceted CBI was being planned in a number of
LGAs throughout the state. By now, it is in its final stages of implementation. Therefore,
recommending this intervention type or arguing against the implementation has become
obsolete, but important points could to be made here.
The findings in Chapter 6 emphasised the knowledge gap arising from limited understanding
of resource use in CBIs. Consequently, a clear recommendation derived here was that the
evaluation of this CBI in South Australia should closely monitor the multi-fold processes
involved. It should also specifically identify the contribution of all intervention stakeholders
and, where possible, consider the benefits received in return, which would allow light to be
shed on potential barriers and facilitators in the intervention. Given the multiple LGAs
involved, there is also a real opportunity to assess the setting-specific determinants of
resource use for this programme.
However, eventually, decision makers in South Australia should be prepared that the impact
of the CBI could remain modest with respect to body composition. It is difficult to predict
whether the societal costs of this recently implemented CBI will be equally as high as the
indicative estimates derived in the present study. Essentially, this will also depend on whether
individual strategies within the CBI are particularly resource-intense. Nevertheless, what the
present study has shown is that there will be many stakeholders involved in a countless
number of activities. It may well be that beyond the financial costs of the CBI; there will be a
considerable effort by community members in being part of the intervention. The respective
opportunity costs—in terms of time and other intangible resources—should not be neglected.
The share societal costs of the intervention should be communicated when presenting the
results of the intervention.
The second main implication derived for South Australia was in relation to policies in the
school setting. The state may consider expanding the current nutrition policy that applies to
government-run schools only, but all non-government schools are exempt from it (DECD,
2008); however, the effectiveness of this policy is not being formerly evaluated. Adherence to
a similar canteen policy implemented in the Australian state of Victoria was found to be low
(de Silva-Sanigorski et al., 2011). The authors of that study suggested that this may be due to
insufficient funding for implementation support. Hence, an implication for the South
Australian context is that the introduction of new policies or the expansion of existing ones
should be aligned with on-going evaluations at both individual and organisational levels.
Furthermore, the current nutrition policy in South Australian public schools refers to school
104
meals and vending machines, but does not include food and beverages brought from home.
As suggested in the systematic review in Chapter 5, students thus might find ways to
circumvent the policy, and continue to consume unhealthy food in schools. Policy-makers in
South Australia may thus consider designing food policies that concern the whole school
environment. However, it would be helpful if more evidence became available in order to
support this notion.
National Australian legislation currently mandates two hours of physical activity per week in
primary schools (DoHA, 2005). Conversely, the physical activity policy from Texas required
students to participate in daily sessions of at least 30 minutes (Barroso et al., 2009). Since the
effectiveness of this policy is yet to be fully evaluated, such as in terms of body composition
outcomes, it is difficult to derive a direct implication here for South Australia; however, it
may well be questioned whether or not the current national mandate for physical activity does
suffice as an effective strategy to prevent obesity in children.
Another implication that could be derived for South Australia was concerned with
surveillance systems for obesity prevention. Currently, the prevalence of overweight and
obesity amongst school children is not being assessed systematically, neither at the state nor
national level (Haby et al., 2012). However, good monitoring is required in order to better
determine the effectiveness population-based interventions, and also to be able to capture
natural experiments, such as school nutrition policies (Ramanathan et al., 2008; Stubbs &
Achat, 2009). Since low participation rates have been an issue in recent surveys, an expert
report suggests that such an Australian surveillance system should track the BMI in school
children based on properly informed passive consent, which allows parents to ‘opt-out’, if
desired (Kremer et al., 2010).
Research on obesity prevention in children
The thesis findings should also be reflected with respect to obesity prevention research in
general. Multi-faceted interventions, strategies to promote a reduction in TV-viewing and the
potentials offered by school-based policies are all lending support to the belief that current
interventions may be effective in preventing overweight and obesity. However, the impact
garnered thus far is modest, which is coherent with the findings from other studies that
conducted comprehensive synthesis (Flynn et al., 2006; Safron et al., 2011; Waters et al.,
2011a).
To date, no country has implemented comprehensive obesity prevention measures, nor is
there evidence to suggest success on a large scale that the tide obesity has substantially been
reverted. Even though a recent plateauing of the childhood obesity prevalence was observed
105
in many countries, researchers doubt this could be attributable to prevention efforts during the
last decade (Olds et al., 2011; Rokholm et al., 2010).
Hence, the search for effective interventions continues to be a global task. A likely obstacle to
the development of clear-cut evidence-based recommendations for prevention may be the
complexity of obesity, which refers to both the causes as well as the possible solutions. In
fact, obesity has been described as a ‘wicked problem’ (Gortmaker et al., 2011), meaning that
obesity is an issue that is difficult to define, continuously evolves in different contexts, and
lacks a single solutions to suit all circumstances (Rittel & Webber, 1973). Nevertheless,
although it is difficult to generalise what works to prevent obesity, it is also little helpful to
simply deem it as a problem too complex to deal with (Müller, 2010). Instead, two main ways
forward are being proposed (Swinburn & de Silva-Sanigorski, 2010). One is direct action
through the implementation of those interventions where evidence for effectiveness and cost-
effectiveness is promising. However, in addition, there is the need for a response that exceeds
the realm of evidence-based public health interventions. This line of thought refers to the
notion of a ‘systems approach’ to obesity prevention, and thus implies support through cross-
cutting, multi-level actions involving governments, international agencies, the private sector,
institutions in the civil society, health professionals, and also individuals (Gortmaker et al.,
2011; Huang et al., 2009). It is hoped that these two streams, when brought together, will
strengthen the knowledge base—not only in terms of what works, but also the ways in which
it should be implemented (Finegood et al., 2010).
Therefore, continuously building a sound evidence base will require enhancements in the
evaluation alongside the implementation of future interventions. The research conducted in
this thesis has highlighted particular areas for improvement related to outcomes. Evaluations
of obesity-prevention need to enhance the reporting of outcome data so that imputation for
meta-analyses can be avoided. The reductions in mean BMI and %BF appear modest, but
may nevertheless conceal stronger effects in specific groups; therefore, more understanding is
needed concerning the way in which the shape of the whole BMI distribution is affected by an
intervention. There is also the specific need to enhance outcomes for the assessment of
behavioural change, especially in relation to TV-viewing. Furthermore, extended follow-up is
required to determine long-term effectiveness. Finally, it remains to be seen whether or not an
impact on children’s wellbeing can be assessed as a result of preventive interventions. A
recent Australian study found overweight and obese adolescents to report lower health-related
quality of life scores than their peers with normal weight; however, this was based on cross-
sectional data only so that causality could not be established (Keating et al., 2011).
106
Methods to synthesise evidence for public health decision making
The recurrent theme across the different studies conducted in this thesis was the role of single
components within multi-faceted interventions. This also makes evidence synthesis on multi-
faceted interventions more difficult. In Chapter 3, it was outlined that it is difficult to
understand how individual components contribute to overall effectiveness. Consequently, in
the light of the apparent shift towards more multi-faceted interventions for the prevention of
obesity, it becomes an increasing challenge to distinguish between the efficient and non-
efficient parts of a programme or a set of different policies. The recently updated Cochrane
Review on obesity prevention in children also highlighted this issue (Waters et al., 2011a).
One way of addressing this problem is through focusing on outcomes beyond the individual
level, which evaluate the process of the intervention (Schneider et al., 2009; Story et al.,
2000). It was observed in the systematic review of policy-based interventions in Chapter 5
that such information adds explanatory power to study findings, particularly where
interventions comprise multiple components. Therefore, systematic reviews on the
effectiveness of multi-component public health interventions may generally benefit from
aligning individual level findings with process evaluation outcomes.
In addition, it was useful to combine quantitative and narrative synthesis in this thesis. Meta-
analysis, in itself, is an important tool for the assessment of intervention effectiveness,
especially when statistical power in individual trials is low. However, where interventions
increasingly comprise multiple components, the mechanisms to produce effectiveness become
more and more obscured; therefore, strong narrative synthesis is needed in order to enlighten
the processes involved, as seen through the review on interventions aimed at a reduction in
TV-viewing. It also provides a good opportunity to link individual-level outcomes with
processes evaluation; however, it is inherent to narrative synthesis that there remains some
discretion to the analyst concerning how such results are derived. Nevertheless, the
experience garnered throughout this thesis was that, when mapping the intervention logic
right at the beginning of the reviewing process, the synthesis structure automatically evolved
in a transparent manner. Following the narrative synthesis framework proposed by Popay et
al. (2006) may therefore be seen as an effective way of reducing the risk associated with
deriving subjective conclusions.
Findings that evolved from the synthesis in this thesis include the evidence for the
effectiveness of interventions classifiable as multi-faceted education programmes; however,
this was more difficult to determine for policy-based environmental modifications. However,
it is the evidence on this latter intervention type that decision makers in public policy
predominantly seek as this is most relevant to population-based prevention of overweight and
107
obesity. This divide between the needs and availability of information concerning effective
interventions can also be described as the ‘inverse evidence law’ (Nutbeam, 2003).
Interventions most likely to influence whole populations tend to be evaluated with study
designs of the lowest quality, and thus good evidence to support them is limited. This makes
it inherently difficult to inform decision makers according to their needs.
Economic evaluations of complex obesity prevention interventions are scant, and there is an
urgent need to extend the evidence base; however, although a larger number of such studies
would be desirable, a hurdle to informed decision-making based on previous experience
gained elsewhere is the limited direct generalisability of the findings that can be derived from
most economic evaluations (Goeree et al., 2007; Sculpher & Drummond, 2006). Simply
transferring an ICER between settings is unlikely to give a true account of the economic
efficiency of an intervention, nor will the mere adjustment of resource quantities and unit
costs to the target setting (Johns & Torres, 2005). This is because the intervention content and
the determinants of the resource use are highly dependent on the specific setting and the
population scale (Welte et al., 2004). On the other hand, it was found through the course of
this thesis that, when transferring the economic evaluation to the local decision making
context, this introduces additional uncertainties owing to the number of assumptions needing
to be made. Another challenge is posed by large-scale public health interventions, such as
CBI. The standardised information concerning programme content generated from studies
conducted elsewhere contends with the essence of these types of interventions, where
activities evolve organically and depend on the context. It appears that common economic
evaluation has paid little attention to such context-dependency, and the dynamics of an
intervention—especially when this is implemented on a large scale. Despite these challenges,
however, decisions still have to be made concerning the best possible use of the information
available at the time. Therefore, the means to help decision makers to prioritise on public
health interventions need to be improved.
The future evaluation of obesity-prevention trials should focus keenly on a transparent
reporting of all underlying processes, and establishing what determines differences in costs
and effects across different settings, when the capacities of fixed cost items exhaust in large
scale implementations, and whether or not synergies can be observed between the
intervention components. It is considered that such an approach will help to enrich the
understanding of all types and patterns of resource use, with such information further helping
to guide the conduct of future economic evaluations of complex programmes.
In addition, health economic modelling techniques that capture the full value of obesity
prevention need to be further developed. The economic model used in this thesis omitted
108
long-term costs and consequences because such projections bear considerable uncertainty for
the findings. For example, a myriad of assumptions was needed in the ACE-project, which
was conducted by an Australian research team in order to inform the decision-making context
in the state of Victoria (Carter et al., 2009; Haby et al., 2006). However, although the study
provided a useful framework to compare cost-effectiveness for several interventions, the
structural validity of the economic model may be questioned, as others have already
highlighted (John et al., 2012; Segal & Dalziel, 2007). However, including long-term costs
and consequences should remain the gold standard when striving to assess obesity prevention
(Lehnert et al., 2012); therefore, it is salient to develop good standards for economic
modelling practices (Levy et al., 2011).
The work presented in this thesis has sought to strike a balance between the best evidence
available (Chapter 2 and Chapter 3) with the information required to feed a health economic
evaluation model (Chapter 4, Chapter 6 and Chapter 7), but also aimed to specifically
consider context-relevant policy interventions (Chapter 5). However, a limiting factor of the
findings in this thesis is that they yielded little direct implications for the prioritisation of
large-scale policies to prevent obesity, since published evidence for such undertakings has
been lacking.
109
References
ABS (2008) National Health Survey 2007-08. cat. no. 4364.0. Canberra: Australian Bureau of
Statistics.
http://www.abs.gov.au/AUSSTATS/[email protected]/Lookup/4364.0Main+Features12007-
2008 (Reissue)?OpenDocument last accessed 05 April 2012.
ABS (2010a) Employee earnings and hours cat. no. 6306.0. Canberra: Australian Bureau of
Statistics.
http://www.abs.gov.au/AUSSTATS/[email protected]/ProductsbyCatalogue/7F76D15354BB
25D5CA2575BC001D5866?OpenDocument last accessed 06 April 2012.
ABS (2010b) Regional Population Growth, Australia: Estimated Resident Population,
Remoteness Areas, Australia. cat. no. 3218.0. Canberra: Australian Bureau of
Statistics. http://www.abs.gov.au/AUSSTATS/[email protected]/DetailsPage/3218.02009-
10?OpenDocument last accessed 13 April 2011.
ABS (2010c) Schools Australia cat. no. 4221.0. Canberra: Australian Bureau of Statistics.
http://www.abs.gov.au/ausstats/[email protected]/Latestproducts/4221.0Main
Features202011?opendocument&tabname=Summary&prodno=4221.0&issue=2011&
num=&view= last accessed 05 April 2012.
Amaro S, Viggiano A, Di Costanzo A, Madeo I, Baccari ME, et al. (2006) Kaledo, a new
educational board-game, gives nutritional rudiments and encourages healthy eating in
children: a pilot cluster randomized trial. European Journal of Pediatrics 165: 630-
635.
Arai L, Britten N, Popay J, Roberts H, Petticrew M, et al. (2007) Testing methodological
developments in the conduct of narrative synthesis: a demonstration review of
research on the implementation of smoke alarm interventions. Evidence & Policy: A
Journal of Research, Debate and Practice 3: 361-383.
Bandura A (1986) Social Foundations of Thought and Action. Englewood Cliffs, New Jersey:
Prentice Hall Int Inc.
Baranowski T, Cerin E, Baranowski J (2009) Steps in the design, development and formative
evaluation of obesity prevention-related behaviour change trials. The International
Journal of Behavioural Nutrition and Physical Activity 6: 6.
Barroso CS, Kelder SH, Springer AE, Smith CL, Ranjit N, et al. (2009) Senate Bill 42:
Implementation and Impact on Physical Activity in Middle Schools. Journal of
Adolescent Health 45: S82-S90.
Bassett MT, Perl S (2004) Obesity: The Public Health Challenge of Our Time. American
Journal of Public Health 94: 1477-1477.
Bayne-Smith M, Fardy PS, Azzollini A, Magel J, Schmitz KH, et al. (2004) Improvements in
heart health behaviours and reduction in coronary artery disease risk factors in urban
teenaged girls through a school-based intervention: the PATH programme. American
Journal of Public Health 94: 1538-1543.
Beets MW, Beighle A, Erwin HE, Huberty JL (2009) After-school programme impact on
physical activity and fitness: a meta-analysis. American Journal of Preventive
Medicine 36: 527-537.
Belfort MB, Zupancic JA, Riera KM, Turner JH, Prosser LA (2011) Health state preferences
associated with weight status in children and adolescents. BMC Pediatrics 11: 12.
Bere E, Hilsen M, Klepp KI (2010) Effect of the nationwide free school fruit scheme in
Norway. The British Journal of Nutrition 104: 589-594.
110
Biddle SJ, O'Connell S, Braithwaite RE (2011) Sedentary behaviour interventions in young
people: a meta-analysis. British Journal of Sports Medicine 45: 937-942.
Blum JE, Davee AM, Beaudoin CM, Jenkins PL, Kaley LA, et al. (2008) Reduced
availability of sugar-sweetened beverages and diet soda has a limited impact on
beverage consumption patterns in Maine high school youth. Journal of Nutrition
Education and Behaviour 40: 341-347.
Brooker S, Kabatereine NB, Fleming F, Devlin N (2008) Cost and cost-effectiveness of
nationwide school-based helminth control in Uganda: intra-country variation and
effects of scaling-up. Health Policy and Planning 23: 24-35.
Brousselle A, Lessard C (2011) Economic evaluation to inform healthcare decision-making:
promise, pitfalls and a proposal for an alternative path. Social Science & Medicine
72: 832-839.
Brown HS, 3rd, Perez A, Li YP, Hoelscher DM, Kelder SH, et al. (2007a) The cost-
effectiveness of a school-based overweight programme. International Journal of
Behavioural Nutrition and Physical Activity 4: 47.
Brown T, Kelly S, Summerbell C (2007b) Prevention of obesity: a review of interventions.
Obesity Reviews 8 Suppl 1: 127-130.
Brown T, Summerbell C (2009) Systematic review of school-based interventions that focus
on changing dietary intake and physical activity levels to prevent childhood obesity:
an update to the obesity guidance produced by the National Institute for Health and
Clinical Excellence. Obesity Reviews 10: 110-141.
Brownell KD (2010) The humbling experience of treating obesity: Should we persist or
desist? Behaviour Research and Therapy 48: 717-719.
Brownson RC, Fielding JE, Maylahn CM (2009) Evidence-based public health: a
fundamental concept for public health practice. Annual Review of Public Health 30:
175-201.
Burke V, Milligan RA, Thompson C, Taggart AC, Dunbar DL, et al. (1998) A controlled trial
of health promotion programmes in 11-year-olds using physical activity "enrichment"
for higher risk children. The Journal of Pediatrics 132: 840-848.
Caballero B (2007) The global epidemic of obesity: an overview. Epidemiologic Reviews 29:
1-5.
Caldwell DM, Ades AE, Higgins JP (2005) Simultaneous comparison of multiple treatments:
combining direct and indirect evidence. British Medical Journal 331: 897-900.
Carter R, Moodie M, Markwick A, Magnus A, Vos T, et al. (2009) Assessing cost-
effectiveness in obesity (ACE-obesity): an overview of the ACE approach, economic
methods and cost results. BMC Public Health 9: 419.
Cawley J (2010) The economics of childhood obesity. Health Affairs 29: 364-371.
Cawley J, Rizzo JA, Haas K (2007) Occupation-specific absenteeism costs associated with
obesity and morbid obesity. Journal of Occupational and Environmental Medicine 49:
1317-1324.
CCEMG – EPPI-Centre Cost Converter (v.1.2).
http://eppi.ioe.ac.uk/costconversion/default.aspx Last accessed on 03 August 2011.
CGC (2008) Transport of rural school children. Canberra: Commonwealth Grants
Commission.
http://www.cgc.gov.au/__data/assets/pdf_file/0005/10769/10_U2008_Transport_of_
Rural_School_Children.pdf last accessed 06 April 2012.
111
Chavarro JE, Peterson KE, Sobol AM, Wiecha JL, Gortmaker SL (2005) Effects of a school-
based obesity-prevention intervention on menarche (United States). Cancer Causes &
Control : CCC 16: 1245-1252.
Claxton K (2008) Exploring uncertainty in cost-effectiveness analysis. Pharmacoeconomics
26: 781-798.
Coast J (2004) Is economic evaluation in touch with society's health values? British Medical
Journal 329: 1233-1236.
Cochrane EPOC Group (2002) Data collection checklist. Ottawa (ON):
Cochrane Effective Practice and Organisation of Care Group.
http://epoc.cochrane.org/epoc-resources-review-authors. Last accessed on 18 June
2010.
Cohen DA, Scribner RA, Farley TA (2000) A structural model of health behaviour: a
pragmatic approach to explain and influence health behaviours at the population
level. Preventive Medicine 30: 146-154.
Colagiuri S, Lee CM, Colagiuri R, Magliano D, Shaw JE, et al. (2010) The cost of overweight
and obesity in Australia. The Medical Journal of Australia 192: 260-264.
Cole TJ, Bellizzi MC, Flegal KM, Dietz WH (2000) Establishing a standard definition for
child overweight and obesity worldwide: international survey. British Medical
Journal 320: 1240-1243.
Cole TJ, Faith MS, Pietrobelli A, Heo M (2005) What is the best measure of adiposity change
in growing children: BMI, BMI %, BMI z-score or BMI centile? European Journal of
Clinical Nutrition 59: 419-425.
Cole TJ, Flegal KM, Nicholls D, Jackson AA (2007) Body mass index cut offs to define
thinness in children and adolescents: international survey. British Medical Journal
335: 194.
Cole TJ, Freeman JV, Preece MA (1995) Body mass index reference curves for the UK, 1990.
Archives of Disease in Childhood 73: 25-29.
Connelly JB, Duaso MJ, Butler G (2007) A systematic review of controlled trials of
interventions to prevent childhood obesity and overweight: a realistic synthesis of the
evidence. Public Health 121: 510-517.
Craig P, Dieppe P, Macintyre S, Michie S, Nazareth I, et al. (2008) Developing and
evaluating complex interventions: the new Medical Research Council guidance.
British Medical Journal 337: a1655.
CRD (2008) Systematic Reviews: CRD's guidance for undertaking reviews in healthcare.
York: Centre for Reviews and Dissemination, University of York.
http://www.york.ac.uk/inst/crd/SysRev/!SSL!/WebHelp/SysRev3.htm last accessed
05 April 2012.
Cullen KW, Watson K, Zakeri I (2008) Improvements in middle school student dietary intake
after implementation of the Texas Public School Nutrition Policy. American Journal
of Public Health 98: 111-117.
Cullen KW, Watson K, Zakeri I, Ralston K (2006) Exploring changes in middle-school
student lunch consumption after local school food service policy modifications.
Public Health Nutrition 9: 814-820.
Cullen KW, Watson KB, Fithian AR (2009) The impact of school socioeconomic status on
student lunch consumption after implementation of the Texas Public School Nutrition
Policy. Journal of School Health 79: 525-531.
112
Dalziel K, Segal L (2006) Point: uncertainty in the economic analysis of school-based obesity
prevention programmes: urgent need for quality evaluation. Obesity 14: 1481-1482.
Daniel M, Paquet C, McDermott R (2009) Obesity prevention in France. Yes, but how and
why? Public Health Nutrition 12: 1305-1306; author reply 1306-1307.
de Salazar L, Jackson S, Shiell A (2007) Guide to Economic Evaluation in Health Promotion.
Washington DC: Pan American Health Organisation.
de Silva-Sanigorski A, Breheny T, Jones L, Lacy K, Kremer P, et al. (2011) Government food
service policies and guidelines do not create healthy school canteens. Australian and
New Zealand Journal of Public Health 35: 117-121.
de Silva-Sanigorski A, Prosser L, Carpenter L, Honisett S, Gibbs L, et al. (2010) Evaluation
of the childhood obesity prevention programme Kids--'Go for your life'. BMC Public
Health 10: 288.
DECD (2008) Right Bite Policy. Adelaide:
Department for Education and Child Development.
http://www.decs.sa.gov.au/eatwellsa/pages/eatwell/rightbite/?reFlag=1 last accessed
05 April 2012.
Dennison BA, Russo TJ, Burdick PA, Jenkins PL (2004) An intervention to reduce television
viewing by preschool children. Archives of Pediatrics & Adolescent Medicine 158:
170-176.
Dietz WH (1998) Health consequences of obesity in youth: childhood predictors of adult
disease. Pediatrics 101: 518-525.
Dietz WH, Jr., Gortmaker SL (1985) Do we fatten our children at the television set? Obesity
and television viewing in children and adolescents. Pediatrics 75: 807-812.
Dietz WH, Robinson TN (1998) Use of the body mass index (BMI) as a measure of
overweight in children and adolescents. The Journal of Pediatrics 132: 191-193.
Dixon JB (2010) The effect of obesity on health outcomes. Molecular and cellular
endocrinology 316: 104-108.
Dixon-Woods M, Agarwal S, Jones D, Young B, Sutton A (2005) Synthesising qualitative
and quantitative evidence: a review of possible methods. Journal of Health Service
Research and Policy 10: 45-53.
DoHA (2005) Active School Curriculum. Department of Health and Aging.
http://www.health.gov.au/internet/healthyactive/publishing.nsf/Content/active_school
_curriculum.pdf/$File/active_school_curriculum.pdf. Last accessed on 06 April 2012.
DoHA (2008a) Guidelines for preparing submissions to the Pharmaceutical Benefits Advisory
Committee (Version 4.3). Canberra: Australian Government, Department of Health
and Aging.
http://www.health.gov.au/internet/main/publishing.nsf/content/AECB791C29482920
CA25724400188EDB/$File/PBAC4.3.2.pdf last accessed 05 April 2012.
DoHA (2008b)
The 2007 Australian National Children’s Nutrition and Physical Activity Survey-
Main findings. Canberra.
http://www.health.gov.au/internet/main/publishing.nsf/content/66596E8FC68FD1A3
CA2574D50027DB86/$File/childrens-nut-phys-survey.pdf last accessed 05 April
2012.
Drewnowski A (2004) Obesity and the food environment: dietary energy density and diet
costs. American Journal of Preventive Medicine 27: 154-162.
113
Drummond M, Sculpher M, Torrance G, O’Brien B, Stoddart G (2005) Methods for
economic health evaluation of healthcare programmes. 3rd, ed. Oxford: Oxford
University Press.
Drummond M, Weatherly H, Ferguson B (2008) Economic evaluation of health interventions.
British Medical Journal 337: a1204.
Drummond MF, Jefferson TO (1996) Guidelines for authors and peer reviewers of economic
submissions to the BMJ. British Medical Journal 313: 275-283.
Economos CD, Hyatt RR, Goldberg JP, Must A, Naumova EN, et al. (2007) A community
intervention reduces BMI z-score in children: Shape Up Somerville first year results.
Obesity 15: 1325-1336.
Egger G, Swinburn B (1997) An "ecological" approach to the obesity pandemic. British
Medical Journal 315: 477-480.
Elbasha EH, Messonnier ML (2004) Cost-effectiveness analysis and healthcare resource
allocation: decision rules under variable returns to scale. Health Economics 13: 21-
35.
Eliakim A, Nemet D, Balakirski Y, Epstein Y (2007) The effects of nutritional-physical
activity school-based intervention on fatness and fitness in preschool children.
Journal of Pediatric Endocrinology & Metabolism 20: 711-718.
Finegood DT, Merth TD, Rutter H (2010) Implications of the foresight obesity system map
for solutions to childhood obesity. Obesity 18 Suppl 1: S13-16.
Fitzgibbon ML, Stolley MR, Dyer AR, VanHorn L, KauferChristoffel K (2002) A
community-based obesity prevention programme for minority children: rationale and
study design for Hip-Hop to Health Jr. Preventive Medicine 34: 289-297.
Fitzgibbon ML, Stolley MR, Schiffer L, Van Horn L, KauferChristoffel K, et al. (2005) Two-
year follow-up results for Hip-Hop to Health Jr.: a randomized controlled trial for
overweight prevention in preschool minority children. The Journal of Pediatrics 146:
618-625.
Flynn MAT, McNeil DA, Maloff B, Mutasingwa D, Wu M, et al. (2006) Reducing obesity
and related chronic disease risk in children and youth: a synthesis of evidence with
‘best practice’ recommendations. Obesity Reviews 7: 7-66.
Foster GD, Sherman S, Borradaile KE, Grundy KM, Vander Veur SS, et al. (2008) A policy-
based school intervention to prevent overweight and obesity. Pediatrics 121: e794-
e802.
Ganz ML (2003) The economic evaluation of obesity interventions: its time has come.
Obesity Research 11: 1275-1277.
Gill TP, Baur LA, Bauman AE, Steinbeck KS, Storlien LH, et al. (2009) Childhood obesity in
Australia remains a widespread health concern that warrants population-wide
prevention programmes. The Medical Journal of Australia 190: 146-148.
Glanz K, Bishop DB (2010) The role of behavioural science theory in development and
implementation of public health interventions. Annual Review of Public Health 31:
399-418.
Goeree R, Burke N, O'Reilly D, Manca A, Blackhouse G, et al. (2007) Transferability of
economic evaluations: approaches and factors to consider when using results from
one geographic area for another. Current Medical Research Opinion 23: 671-682.
Gold M, Siegel J, Russell L, Weinstein M (1996) Cost-Effectiveness in Health and Medicine.
New YorK, NY: Oxford University Press.
114
Gonzalez-Suarez C, Worley A, Grimmer-Somers K, Dones V (2009) School-based
interventions on childhood obesity: a meta-analysis. American Journal of Preventive
Medicine 37: 418-427.
Gortmaker SL, Peterson K, Wiecha J, Sobol AM, Dixit S, et al. (1999) Reducing obesity via a
school-based interdisciplinary intervention among youth: Planet Health. Archives of
Pediatrics & Adolescent Medicine 153: 409-418.
Gortmaker SL, Swinburn BA, Levy D, Carter R, Mabry PL, et al. (2011) Changing the future
of obesity: science, policy, and action. The Lancet 378: 838-847.
Haby M, Vos T, Carter R, Moodie M, Markwick A, et al. (2006) A new approach to assessing
the health benefit from obesity interventions in children and adolescents: the
assessing cost-effectiveness in obesity project. International Journal of Obesity 30:
1463-1475.
Haby MM, Doherty R, Welch N, Mason V (2012) Community-based interventions for obesity
prevention: lessons learned by Australian policy-makers. BMC Research Notes 5: 20.
Han JC, Lawlor DA, Kimm SYS (2010) Childhood obesity. The Lancet 375: 1737-1748.
Haroun D, Harper C, Wood L, Nelson M (2011) The impact of the food-based and nutrient-
based standards on lunchtime food and drink provision and consumption in primary
schools in England. Public Health Nutrition 14: 209-218.
Harrell JS, McMurray RG, Bangdiwala SI, Frauman AC, Gansky SA, et al. (1996) Effects of
a school-based intervention to reduce cardiovascular disease risk factors in
elementary-school children: the Cardiovascular Health in Children (CHIC) study. The
Journal of Pediatrics 128: 797-805.
Harris KC, Kuramoto LK, Schulzer M, Retallack JE (2009) Effect of school-based physical
activity interventions on body mass index in children: a meta-analysis. Canadian
Medical Association Journal 180: 719-726.
Harrison M, Burns CF, McGuinness M, Heslin J, Murphy NM (2006) Influence of a health
education intervention on physical activity and screen time in primary school
children: 'Switch Off--Get Active'. Journal of Science and Medicine in Sport 9: 388-
394.
Harten N, Olds T (2004) Patterns of active transport in 11-12 year old Australian children.
Australian and New Zealand Journal of Public Health 28: 167-172.
Hawe P, Shiell A, Riley T (2004) Complex interventions: how "out of control" can a
randomised controlled trial be? British Medical Journal 328: 1561-1563.
Hawe P, Shiell A, Riley T (2009) Theorising interventions as events in systems. American
Journal of Community Psychology 43: 267-276.
Hill JO, Melanson EL (1999) Overview of the determinants of overweight and obesity:
current evidence and research issues. Medicine and Science in Sports and Exercise
31: S515-521.
Howard BV, Manson JE, Stefanick ML, Beresford SA, Frank G, et al. (2006) Low-fat dietary
pattern and weight change over 7 years: the Women's Health Initiative Dietary
Modification Trial. Journal of the American Medical Association 295: 39-49.
Huang TT, Drewnosksi A, Kumanyika S, Glass TA (2009) A systems-oriented multilevel
framework for addressing obesity in the 21st century. Preventing Chronic Disease 6:
A82.
115
Hutubessy R, Chisholm D, Edejer TT (2003) Generalised cost-effectiveness analysis for
national-level priority-setting in the health sector. Cost Effectiveness and Resource
Allocation : C/E 1: 8.
IASO (2012) Global obesity, ranking by country 2012.
International Association for the Study of Obesity.
http://www.iaso.org/site_media/uploads/Global_prevalence_of_adult_obesity_Rankin
g_by_country_2012.pdf. Last accessed on 03 April 2012.
Jadad AR, Moore RA, Carroll D, Jenkinson C, Reynolds DJ, et al. (1996) Assessing the
quality of reports of randomized clinical trials: is blinding necessary? Control Clinical
Trials 17: 1-12.
Jaime PC, Lock K (2009) Do school based food and nutrition policies improve diet and
reduce obesity? Preventive Medicine 48: 45-53.
Jain A (2005) Treating obesity in individuals and populations. British Medical Journal 331:
1387-1390.
Johansson P, Tillgren P (2011) Financing intersectoral health promotion programmes: some
reasons why collaborators are collaborating as indicated by cost-effectiveness
analyses. Scandinavian Journal of Public Health 39: 26-32.
Johansson PM, Eriksson LS, Sadigh S, Rehnberg C, Tillgren PE (2009) Participation,
resource mobilization and financial incentives in community-based health promotion:
an economic evaluation perspective from Sweden. Health Promotion International 24:
177-184.
John J, Wenig CM, Wolfenstetter SB (2010) Recent economic findings on childhood obesity:
cost-of-illness and cost-effectiveness of interventions. Current Opinion on Clinical
Nutrion and Metabolic Care 13: 305-313.
John J, Wolfenstetter SB, Wenig CM (2012) An economic perspective on childhood obesity:
Recent findings on cost of illness and cost effectiveness of interventions. Nutrition
published online 26 March 2012: 10.1016/j.nut.2011.11.016.
Johns B, Torres TT (2005) Costs of scaling up health interventions: a systematic review.
Health Policy and Planning 20: 1-13.
Jordan KC, Erickson ED, Cox R, Carlson EC, Heap E, et al. (2008) Evaluation of the Gold
Medal Schools programme. Journal of the American Dietetic Association 108: 1916-
1920.
Kamath CC, Vickers KS, Ehrlich A, McGovern L, Johnson J, et al. (2008) Clinical review:
behavioural interventions to prevent childhood obesity: a systematic review and
metaanalyses of randomized trials. Journal of Clinical Endocrinology and
Metabolism 93: 4606-4615.
Katz DL, O'Connell M, Njike VY, Yeh MC, Nawaz H (2008) Strategies for the prevention
and control of obesity in the school setting: systematic review and meta-analysis.
International Journal of Obesity 32: 1780-1789.
Keating CL, Moodie ML, Swinburn BA (2011) The health-related quality of life of
overweight and obese adolescents - a study measuring body mass index and
adolescent-reported perceptions. International Journal of Pediatric Obesity 6: 434-
441.
Kesztyus D, Schreiber A, Wirt T, Wiedom M, Dreyhaupt J, et al. (2011) Economic evaluation
of URMEL-ICE, a school-based overweight prevention programme comprising
metabolism, exercise and lifestyle intervention in children.
116
The European Journal of Health Economics published online October 11, 2011:
10.1007/s10198-011-0358-3.
King L, Gill T, Allender S, Swinburn B (2011) Best practice principles for community-based
obesity prevention: development, content and application. Obesity Reviews 12: 329-
338.
Kipping RR, Payne C, Lawlor DA (2008) Randomised controlled trial adapting US school
obesity prevention to England. Archives of Disease in Childhood 93: 469-473.
Kolotkin RL, Meter K, Williams GR (2001) Quality of life and obesity. Obesity Reviews 2:
219-229.
Konig HH, Lehnert T, Riedel-Heller S, Konnopka A (2011) [Economic aspects of prevention
and treatment of overweight and obesity in children and adolescents].
Bundesgesundheitsblatt, Gesundheitsforschung, Gesundheitsschutz 54: 611-620.
Krauth C, John J, Suhrcke M (2011) Gesundheitsökonomische Methoden in der Prävention
[Health economic evaluation in prevention]. Prävention und Gesundheitsförderung 6:
85-93.
Kremer P, de Silva-Sanigorski A, Leslie E, Lacy K, Jones L (2010) Analysis of critical
features for establishing a monitoring system for childhood obesity: Monitoring
system design. Geelong: Deakin University. http://www.co-
ops.net.au/Pages/Public/2011_Monitoring_Reports.aspx last accessed 05 April 2012.
Kropski JA, Keckley PH, Jensen GL (2008) School-based obesity prevention programmes: an
evidence-based review. Obesity 16: 1009-1018.
Kuczmarski RJ, Ogden CL, Guo SS, Grummer-Strawn LM, Flegal KM, et al. (2002) 2000
CDC Growth Charts for the United States: methods and development. Vital and
health statistics Series 11, Data from the national health survey: 1-190.
Kumanyika S, Obarzanek E, Stettler N, Bell R, Field A, et al. (2008) Population-based
prevention of obesity: the need for comprehensive promotion of healthful eating,
physical activity, and energy balance: a scientific statement from American Heart
Association Council on Epidemiology and Prevention, Interdisciplinary Committee
for Prevention (formerly the expert panel on population and prevention science).
Circulation 118: 428-464.
Lean M, Gruer L, Alberti G, Sattar N (2006) ABC of obesity. Obesity--can we turn the tide?
British Medical Journal 333: 1261-1264.
Lehnert T, Sonntag D, Konnopka A, Riedel-Heller S, Konig HH (2012) The long-term cost-
effectiveness of obesity prevention interventions: systematic literature review.
Obesity Reviews.
Leidl R (2008) Promoting economic value in public health. European Journal of Public
Health 18: 216.
Leidl R (2010) Economic modelling in public health: a tool of growing relevance. European
Journal of Public Health 20: 365-366.
Levine MD, Klem ML, Kalarchian MA, Wing RR, Weissfeld L, et al. (2007) Weight gain
prevention among women. Obesity 15: 1267-1277.
Levy DT, Mabry PL, Wang YC, Gortmaker S, Huang TT, et al. (2011) Simulation models of
obesity: a review of the literature and implications for research and policy. Obesity
Reviews 12: 378-394.
Lobstein T (2009) International moves to strengthen food policies. The Proceedings of the
Nutrition Society 68: 221-224.
117
Lobstein T, Baur L, Uauy R (2004) Obesity in children and young people: a crisis in public
health. Obesity Reviews 5 Suppl 1: 4-104.
Luckner H, Moss JR, Gericke CA (2011) Effectiveness of interventions to promote healthy
weight in general populations of children and adults: a meta-analysis.
European Journal of Public Health published online October 12, 2011: ckr141 [pii]
10.1093/eurpub/ckr141.
Macdiarmid JI, Loe J, Douglas F, Ludbrook A, Comerford C, et al. (2011) Developing a
timeline for evaluating public health nutrition policy interventions. What are the
outcomes and when should we expect to see them? Public Health Nutrition 14: 729-
739.
Maniccia DM, Davison KK, Marshall SJ, Manganello JA, Dennison BA (2011) A meta-
analysis of interventions that target children's screen time for reduction. Pediatrics
128: e193-210.
Mansley EC, Dunet DO, May DS, Chattopadhyay SK, McKenna MT (2002) Variation in
average costs among federally sponsored state-organized cancer detection
programmes: economies of scale? Medical Decision Making 22: S67-79.
Mauskopf JA, Paul JE, Grant DM, Stergachis A (1998) The role of cost-consequence analysis
in healthcare decision-making. Pharmacoeconomics 13: 277-288.
McAuley KA, Taylor RW, Farmer VL, Hansen P, Williams SM, et al. (2010) Economic
Evaluation of a Community-based Obesity Prevention Programme in Children: The
APPLE Project. Obesity 18: 131-136.
MCEECDYA (2009) SCHOOLS GEOGRAPHIC LOCATION DATABASE. Carlton South
Vic: Ministerial Council for Education, Early Childhood Development and Youth
Affairs. http://www.mceecdya.edu.au/mceecdya/ last accessed 05 April 2012.
McGraw SA, Sellers D, Stone E, Resnicow KA, Kuester S, et al. (2000) Measuring
implementation of school programmes and policies to promote healthy eating and
physical activity among youth. Preventive Medicine 31: S86-S97.
Mendoza JA, Watson K, Cullen KW (2010) Change in dietary energy density after
implementation of the Texas Public School Nutrition Policy. Journal of the American
Dietetic Association 110: 434-440.
Mensah GA, Goodman RA, Zaza S, Moulton AD, Kocher PL, et al. (2004) Law as a tool for
preventing chronic diseases: expanding the spectrum of effective public health
strategies. Preventing Chronic Disease 1: A11.
Mernagh P, Peach D, Weston A (2010) Cost effectiveness report of public health
interventions to prevent obesity. Report prepared for the health research council of
New Zealand. Sydney: Health Technology Analysts Pty Ltd.
http://www.victoria.ac.nz/hsrc/downloads/CE Obesity Prevention Full Report
publish.pdf last accessed 05 April 2012.
Metos J, Murtaugh M (2011) Words or reality: Are school district wellness policies
implemented? A systematic review of the literature. Childhood Obesity 7: 90-100.
Moag-Stahlberg A, Howley N, Luscri L (2008) A national snapshot of local school wellness
policies. Journal of School Health 78: 562-568.
Moher D, Liberati A, Tetzlaff J, Altman DG (2009) Preferred reporting items for systematic
reviews and meta-analyses: the PRISMA statement. PLoS Medicine 6: e1000097.
Moodie M, Haby M, Galvin L, Swinburn B, Carter R (2009a) Cost-effectiveness of active
transport for primary school children - Walking School Bus programme. International
Journal of Behavioural Nutrition and Physical Activity 6: 63.
118
Moodie ML, Carter RC, Swinburn BA, Haby MM (2009b) The cost-effectiveness of
Australia's Active After-School Communities programme. Obesity 18: 1585-1592.
Mullally ML, Taylor JP, Kuhle S, Bryanton J, Hernandez KJ, et al. (2010) A province-wide
school nutrition policy and food consumption in elementary school children in Prince
Edward Island. Canadian Journal of Public Health 101: 40-43.
Müller MJ (2010) How are we going to turn the obesity prevention experience? Obesity
Reviews 11: 101-104.
Musingarimi P (2009) Obesity in the UK: A review and comparative analysis of policies
within the devolved administrations. Health Policy 91: 10-16.
Nanney MS, Nelson T, Wall M, Haddad T, Kubik M, et al. (2010) State School Nutrition and
Physical Activity Policy Environments and Youth Obesity. American Journal of
Preventive Medicine 38: 9-16.
Nestle M, Jacobson M (2000) Halting the obesity epidemic: a public health policy approach.
Public Health Reports 115: 12-24.
Neumark-Sztainer D, Story M, Hannan PJ, Rex J (2003) New Moves: a school-based obesity
prevention programme for adolescent girls. Preventive Medicine 37: 41-51.
NHMRC (1999) A guide to the development, implementation and evaluation of clinical
practice guidelines. Canberra: National Health and Medical Research Council.
http://www.nhmrc.gov.au/_files_nhmrc/publications/attachments/cp30.pdf?q=publica
tions/synopses/_files/cp30.pdf last accessed 05 April 2012.
Nutbeam D (2003) How does evidence influence public health policy? Tackling health
inequalities in England? Health Promotion Journal of Australia 14: 154–158.
Olds T, Maher C, Zumin S, Péneau S, Lioret S, et al. (2011) Evidence that the prevalence
of childhood overweight is plateauing: data from nine countries. International Journal
of Pediatric Obesity 6: 342-360.
Oliver S, Bagnall AM, Thomas J, Shepherd J, Sowden A, et al. (2010) Randomised controlled
trials for policy interventions: a review of reviews and meta-regression. Health
Technology Assessment 14: 1-165, iii.
Over M (1986) The effect of scale on cost projections for a primary healthcare programme in
a developing country. Social Science & Medicine 22: 351-360.
Owen L, Morgan A, Fischer A, Ellis S, Hoy A, et al. (2011) The cost-effectiveness of public
health interventions. Journal of Public Health published online September 20, 2011:
fdr075 [pii] 10.1093/pubmed/fdr075.
Panzar JC, Willig RD (1981) Economies of Scope. American Economic Review 71: 268-272.
Parcel GS, Simons-Morton B, O'Hara NM, Baranowski T, Wilson B (1989) School
promotion of healthful diet and physical activity: impact on learning outcomes and
self-reported behaviour. Health Education Quarterly 16: 181-199.
PEI Healthy Eating Alliance. School Nutrition Policies. http://www.healthyeatingpei.ca/pei-
school-nutrition-policy.php. Last accessed on 06 April 2012.
Petticrew M, Cummins S, Ferrell C, Findlay A, Higgins C, et al. (2005) Natural experiments:
an underused tool for public health? Public Health 119: 751-757.
Pettman T, McAllister M, Verity F, Magarey A, Dollman J, et al. (2010) eat well be active
Community Programmes – Final report. Adelaide: South Australian Department of
Health. http://www.health.sa.gov.au/pehs/branches/health-
promotion/ewba/ewbaCPFinalReport2010Electronic.pdf last accessed 06 April 2012.
119
PHIDU (2008) Monitoring Inequality in Australia: South Australia. Adelaide:
Public Health Information Development Unit.
http://www.publichealth.gov.au/interactive_graphics/australia_2010/sa/families.html
last accessed 03 August 2011.
Popay J, Roberts H, Sowden A, Petticrew M, Arai L, et al. (2006) Guidance on the conduct of
narrative synthesis in systematic reviews: a product from the ESRC Methods
Programme. Lancaster: Institute for Health Research, Lancaster University.
http://www.lancs.ac.uk/shm/research/nssr/research/dissemination/publications/NS_Sy
nthesis_Guidance_v1.pdf last accessed 05 April 2012.
Productivity Commission (2011) REPORT ON GOVERNMENT SERVICES 2011 - School
Education. http://www.pc.gov.au/gsp/rogs/2011 last accessed 05 April 2012.
Proper KI, Hildebrandt VH, Van der Beek AJ, Twisk JW, Van Mechelen W (2003) Effect of
individual counseling on physical activity fitness and health: a randomized controlled
trial in a workplace setting. American Journal of Preventive Medicine 24: 218-226.
Ramanathan S, Allison KR, Faulkner G, Dwyer JJ (2008) Challenges in assessing the
implementation and effectiveness of physical activity and nutrition policy
interventions as natural experiments. Health Promotion International 23: 290-297.
Reilly JJ, Kelly J (2011) Long-term impact of overweight and obesity in childhood and
adolescence on morbidity and premature mortality in adulthood: systematic review.
International Journal of Obesity 35: 891-898.
Reilly JJ, Methven E, McDowell ZC, Hacking B, Alexander D, et al. (2003) Health
consequences of obesity. Archives of Disease in Childhood 88: 748-752.
Ricci JA, Chee E (2005) Lost productive time associated with excess weight in the U.S.
workforce. Journal of Occupational and Environmental Medicine 47: 1227-1234.
Rittel HWJ, Webber MM (1973) Dilemmas in a general theory of planning. Policy Sciences
4: 155-169.
Robinson TN (1999) Reducing children's television viewing to prevent obesity: a randomized
controlled trial. Journal of the American Medical Association 282: 1561-1567.
Robinson TN, Killen JD, Kraemer HC, Wilson DM, Matheson DM, et al. (2003) Dance and
reducing television viewing to prevent weight gain in African-American girls: the
Stanford GEMS pilot study. Ethnicity & Disease 13: S65-77.
Rodgers M, Sowden A, Petticrew M, Arai L, Roberts H, et al. (2009) Testing Methodological
Guidance on the Conduct of Narrative Synthesis in Systematic Reviews. Evaluation
15: 49-73.
Rokholm B, Baker JL, Sørensen TIA (2010) The levelling off of the obesity epidemic since
the year 1999 – a review of evidence and perspectives. Obesity Reviews 11: 835-846.
Ronckers ET, Groot W, Steenbakkers M, Ruland E, Ament A (2006) Costs of the 'Hartslag
Limburg' community heart health intervention. BMC Public Health 6: 51.
Rose G (1992) The strategy of preventive medicine. Oxford: Oxford University Press.
Ryan KW, Card-Higginson P, McCarthy SG, Justus MB, Thompson JW (2006) Arkansas
fights fat: Translating research into policy to combat childhood and adolescent
obesity. Health Affairs 25: 992-1004.
Rychetnik L, Frommer M, Hawe P, Shiell A (2002) Criteria for evaluating evidence on public
health interventions. Journal of Epidemiology and Community Health 56: 119-127.
Rychetnik L, Hawe P, Waters E, Barratt A, Frommer M (2004) A glossary for evidence based
public health. Journal of Epidemiology and Community Health 58: 538-545.
120
SA Departement of Health (2006) Healthy Weight Strategy for South Australia 2006-2010.
Adelaide: Government of South Australia, Department of Health.
http://www.health.sa.gov.au/pehs/branches/health-promotion/hp-healthy-weight.htm
last accessed 05 April 2012.
Sacks G, Swinburn BA, Lawrence MA (2008) A systematic policy approach to changing the
food system and physical activity environments to prevent obesity. Australian and
New Zealand Health Policy 5: 13.
Sadowsky H, Sawdon J, Sheiner M, Sticklin A (1999) Eight week moderate intensity exercise
intervention elicits body composition change in adolescents. Cardiopulmonary
Physical Therapy Journal 10: 38-44.
Safron M, Cislak A, Gaspar T, Luszczynska A (2011) Effects of School-based Interventions
Targeting Obesity-Related Behaviours and Body Weight Change: A Systematic
Umbrella Review. Behavioural Medicine 37: 15-25.
Sanchez-Vaznaugh EV, Sánchez BN, Baek J, Crawford PB (2010) 'Competitive' food and
beverage policies: Are they influencing childhood overweight trends? Health Affairs
29: 436-446.
Sanigorski AM, Bell AC, Kremer PJ, Cuttler R, Swinburn BA (2008) Reducing unhealthy
weight gain in children through community capacity-building: results of a quasi-
experimental intervention programme, Be Active Eat Well. International Journal of
Obesity 32: 1060-1067.
Schmidt ME, Haines J, O'Brien A, McDonald J, Price S, et al. (2012) Systematic Review of
Effective Strategies for Reducing Screen Time Among Young Children. Obesity.
Schneider M, Hall WJ, Hernandez AE, Hindes K, Montez G, et al. (2009) Rationale, design
and methods for process evaluation in the HEALTHY study. International Journal of
Obesity 33 Suppl 4: S60-67.
Schwappach DL, Boluarte TA, Suhrcke M (2007) The economics of primary prevention of
cardiovascular disease - a systematic review of economic evaluations. Cost
Effectiveness and Resource Allocation : C/E 5: 5.
Schwartz MB, Novak SA, Fiore SS (2009) The impact of removing snacks of low nutritional
value from middle schools. Health Education and Behaviour 36: 999-1011.
Sculpher MJ, Drummond MF (2006) Analysis sans frontieres: can we ever make economic
evaluations generalisable across jurisdictions? Pharmacoeconomics 24: 1087-1099.
Segal L, Dalziel K (2007) Economic evaluation of obesity interventions in children and
adults. International Journal of Obesity 31: 1183-1184; author reply 1185.
Seo DC, Sa J (2010) A meta-analysis of obesity interventions among U.S. minority children.
Journal of Adolescent Health 46: 309-323.
Sharma M (2006) School-based interventions for childhood and adolescent obesity. Obesity
Reviews 7: 261-269.
Shepperd S, Lewin S, Straus S, Clarke M, Eccles MP, et al. (2009) Can We Systematically
Review Studies That Evaluate Complex Interventions? PLoS Medicine 6: e1000086.
Shiell A (2007) In search of social value. International Journal of Public Health 52: 333-334.
Shiell A, Hawe P, Gold L (2008) Complex interventions or complex systems? Implications
for health economic evaluation. British Medical Journal 336: 1281-1283.
Simons-Morton BG, Parcel GS, Baranowski T, Forthofer R, O'Hara NM (1991) Promoting
physical activity and a healthful diet among children: results of a school-based
intervention study. American Journal of Public Health 81: 986-991.
121
South Australia Smarter Schools National Partnerships (2009).
http://smarterschools.gov.au/ParticipatingSchools/Pages/SAOverview.aspx. Last
accessed on 03 August 2011.
Stevens W, Wiseman V, Ortiz J, Chavasse D (2005) The costs and effects of a nationwide
insecticide-treated net programme: the case of Malawi. Malar Journal 4: 22.
Stice E, Shaw H, Marti CN (2006) A meta-analytic review of obesity prevention programmes
for children and adolescents: the skinny on interventions that work. Psychological
Bulletin 132: 667-691.
Stillwaggon E (2009) Complexity, cofactors, and the failure of AIDS policy in Africa. Journal
of the International AIDS Society 12: 12.
Stokols D (1992) Establishing and maintaining healthy environments. Toward a social
ecology of health promotion. The American Psychologist 47: 6-22.
Stolley MR, Fitzgibbon ML, Dyer A, Van Horn L, KauferChristoffel K, et al. (2003) Hip-
Hop to Health Jr., an obesity prevention programme for minority preschool children:
baseline characteristics of participants. Preventive Medicine 36: 320-329.
Story M, Mays RW, Bishop DB, Perry CL, Taylor G, et al. (2000) 5-a-day Power Plus:
process evaluation of a multicomponent elementary school programme to increase
fruit and vegetable consumption. Health Education and Behaviour 27: 187-200.
Story M, Nanney MS, Schwartz MB (2009) Schools and obesity prevention: Creating school
environments and policies to promote healthy eating and physical activity. Milbank
Quarterly 87: 71-100.
Stubbs JM, Achat HM (2009) Individual rights over public good? The future of
anthropometric monitoring of school children in the fight against obesity. The
Medical Journal of Australia 190: 140-142.
Summerbell CD, Waters E, Edmunds LD, Kelly S, Brown T, et al. (2005) Interventions for
preventing obesity in children. Cochrane Database of Systematic Reviews:
CD001871.
Swinburn B, Egger G (2002) Preventive strategies against weight gain and obesity. Obesity
Reviews 3: 289-301.
Swinburn B, Egger G, Raza F (1999) Dissecting obesogenic environments: the development
and application of a framework for identifying and prioritizing environmental
interventions for obesity. Preventive Medicine 29: 563-570.
Swinburn B, Gill T, Kumanyika S (2005) Obesity prevention: a proposed framework for
translating evidence into action. Obesity Reviews 6: 23-33.
Swinburn BA (2008) Obesity prevention: the role of policies, laws and regulations. Australia
and New Zealand Health Policy 5: 12.
Swinburn BA, de Silva-Sanigorski AM (2010) Where to from here for preventing childhood
obesity: an international perspective. Obesity 18 Suppl 1: S4-7.
Tan-Torres T, Baltussen R, Adam T, Hutubessy R, Acharya A, et al. (2003) Making choices
in health : WHO guide to cost-effectiveness analysis. Geneva: World Health
Organization.
Taylor RW, McAuley KA, Barbezat W, Strong A, Williams SM, et al. (2007) APPLE
Project: 2-y findings of a community-based obesity prevention programme in primary
school age children. American Journal of Clinical Nutrition 86: 735-742.
122
The 2007 National Children's Nutrition and Physical Activity Survey:Available from the
Australian Social Science Data Archive. http://www.assda.edu.au/ Last accessed on
06 April 2012.
The Cochrane Collaboration (2008) Review Manager (RevMan). 5.0 ed. Copenhagen: The
Nordic Cochrane Centre, The Cochrane Collaboration.
Thomas H (2003) Quality Assessment Tool for Quantitative Studies. Effective Public Health
Practice Project. Toronto: McMaster University.
http://www.myhamilton.ca/myhamilton/CityandGovernment/HealthandSocialService
s/Research/EPHPP/ last accessed 06 April 2012.
Thomas H (2006) Obesity prevention programmes for children and youth: why are their
results so modest? Health Education Research 21: 783-795.
Trasande L (2010) How Much Should We Invest In Preventing Childhood Obesity? Health
Affairs 29: 372-378.
Trickett EJ, Beehler S, Deutsch C, Green LW, Hawe P, et al. (2011) Advancing the science of
community-level interventions. American Journal of Public Health 101: 1410-1419.
Tsiros MD, Olds T, Buckley JD, Grimshaw P, Brennan L, et al. (2009) Health-related quality
of life in obese children and adolescents. International Journal of Obesity 33: 387-
400.
Utah Department of Health. Gold Medal Schools Criteria.
http://health.utah.gov/obesity/gms/criteria.html. Last accessed on 06 April 2012.
Varian HR (1992) Microeconomic analysis. 3rd, ed. New York: Norton.
Vaska V, Volkmer R (2004) Increasing prevalence of obesity in South Australian 4-year-olds:
1995-2002. Journal of Paediatrics and Child Health 40: 353-355.
Walker DG, Teerawattananon Y, Anderson R, Richardson G (2010) Generalisability,
Transferability, Complexity and Relevance. In: Shemilt I, Mugford M, Vale L, Marsh
K, Donaldson C, editors. Evidence-Based Decisions and Economics: Healthcare,
Social Welfare, Education and Criminal Justice. 2nd edition ed. Oxford: Wiley-
Blackwell. pp. 56-66.
Walls HL, McNeil JJ, Peeters A (2009) Population Versus High-risk Interventions for
Obesity. Epidemiology 20: 929-930.
Walls HL, Wolfe R, Haby MM, Magliano DJ, de Courten M, et al. (2010) Trends in BMI of
urban Australian adults, 1980–2000. Public Health Nutrition 13: 631-638.
Wang L, Gutin B, Barbeau P, Moore J, Hanes J, et al. (2008) Cost-effectiveness of a school-
based obesity prevention programme. Journal of School Health 78: 619-624.
Wang L, Yang Q, Lowry R, Wechsler H (2003) Economic analysis of a school-based obesity
prevention programme. Obesity Research 11: 1313-1324.
Wang LY, Denniston M, Lee S, Galuska D, Lowry R (2010) Long-term health and economic
impact of preventing and reducing overweight and obesity in adolescence. Journal of
Adolescent Health 46: 467-473.
Waters E (2009) Evidence for public health decision-making: towards reliable synthesis.
Bulletin of the World Health Organisation 87: 164.
Waters E, de Silva-Sanigorski A, Hall Belinda J, Brown T, Campbell Karen J, et al. (2011a)
Interventions for preventing obesity in children. Cochrane Database of Systematic
Reviews Issue 12 Art No: CD001871 DOI: 101002/14651858CD001871pub3.
123
Waters E, Hall BJ, Armstrong R, Doyle J, Pettman TL, et al. (2011b) Essential components of
public health evidence reviews: capturing intervention complexity, implementation,
economics and equity. Journal of Public Health 33: 462-465.
Weatherly H, Drummond M, Claxton K, Cookson R, Ferguson B, et al. (2009) Methods for
assessing the cost-effectiveness of public health interventions: key challenges and
recommendations. Health Policy 93: 85-92.
Wechsler H, Devereaux RS, Davis M, Collins J (2000) Using the School Environment to
Promote Physical Activity and Healthy Eating. Preventive Medicine 31: S121-S137.
Welte R, Feenstra T, Jager H, Leidl R (2004) A decision chart for assessing and improving
the transferability of economic evaluation results between countries.
Pharmacoeconomics 22: 857-876.
Whitaker RC, Wright JA, Pepe MS, Seidel KD, Dietz WH (1997) Predicting obesity in young
adulthood from childhood and parental obesity. The New England Journal of
Medicine 337: 869-873.
WHO (2000) Obesity: Preventing and Managing the Global Epidemic. Report of a WHO
consultation. Geneva.
http://apps.who.int/bookorders/anglais/detart1.jsp?sesslan=2&codlan=1&codcol=10
&codcch=894 last accessed 05 April 2012.
WHO (2002) Reducing Risks, Promoting Healthy Life. Geneva: World Health Organization.
WHO (2006) Fact Sheet Number 311, Obesity and overweight: What Are Overweight and
Obesity? Geneva: World Health Organization.
http://www.who.int/mediacentre/factsheets/fs311/en/index.html last accessed 06
April 2012.
WHO (2011) Obesity and Overweight. Geneva: World Health Organization. Fact Sheet No.
311. http://www.who.int/mediacentre/factsheets/fs311/en/index.html last accessed
April 2012.
WHO Collaborating Centre for Obesity Prevention (2008) Report 6: Lessons learned from
Colac’s Be Active Eat Well project. Geelong: Deakin University.
http://www.goforyourlife.vic.gov.au/hav/admin.nsf/Images/ssop6_report_6_baew_fin
al.pdf/$File/ssop6_report_6_baew_final.pdf last accessed 05 April 2012.
Woodward-Lopez G, Gosliner W, Samuels SE, Craypo L, Kao J, et al. (2010) Lessons
learned from evaluations of California's statewide school nutrition standards.
American Journal of Public Health 100: 2137-2145.
Yin Z, Gutin B, Johnson MH, Hanes J, Jr., Moore JB, et al. (2005) An environmental
approach to obesity prevention in children: Medical College of Georgia FitKid
Project year 1 results. Obesity Research 13: 2153-2161.
124
Appendix 1
Search Strategies for PubMed, Scopus, EMBASE, CINAHL and Cochrane Central
Register of Controlled Trials in chapter 2
PubMed: (Overweight[mh] OR Overweight[tiab] OR Obes*[tiab]) AND (Prevention and
control[sh] OR Preventive health service[mh] OR Prevent*[tiab] OR Health promotion[mh]
OR Health promotion[tiab] OR Primary prevention[mh:noexp] OR Diet[mh] OR
Exercise[mh]) AND ("control group"[tiab] OR controls[tiab] OR comparator[tiab] OR
“control school”[tiab] OR "comparison group"[tiab] OR Intervention studies[mh] OR
Programme evaluation[mh:noexp]) AND (“Body fat”[tiab] OR Bmi[tiab] OR Body mass
index[mh] OR “Body mass index”[tiab] OR Anthropometry[mh] OR Body composition[mh]
OR Treatment outcome[mh] OR Outcome assessment[mh:noexp] OR Outcome*[tiab] OR
Follow up studies[mh]) Limited to Humans, English, French, German, Spanish
Scopus: (TITLE-ABS-KEY(overweight) OR TITLE-ABS-KEY(obes*)) AND (TITLE-ABS-
KEY("prevention and control") OR KEY("Preventive health service") OR TITLE(prevent*)
OR ABS(prevent*) OR TITLE-ABS-KEY("Health promotion") OR KEY("primary
prevention") OR KEY(diet) KEY(exercise) OR KEY(“health education”)) AND (TITLE-
ABS-KEY("Control group") OR TITLE-ABS-KEY("Comparison group") OR TITLE-ABS-
KEY(controls) OR KEY("intervention studies") OR KEY("Programme evaluation")) AND
(ABS(“body fat”) OR ABS(bmi) OR TITLE-ABS-KEY(“body mass index”) OR
KEY(anthropometry) OR KEY(“body composition”) OR KEY("Treatment outcome") OR
KEY("Outcome assessment") OR ABS(outcome*) OR TITLE-ABS-KEY("Follow up
studies")) AND (LIMIT-TO(LANGUAGE, "English") OR LIMIT-TO(LANGUAGE,
"German") OR LIMIT-TO(LANGUAGE, "Spanish") OR LIMIT-TO(LANGUAGE,
"French"))
EMBASE: ('obesity'/exp OR obes*:ta OR obes*:ab OR overweight:ta OR overweight:ab)
AND ('prevention and control'/mj OR 'prevention'/mj OR 'primary prevention'/mj OR
prevent*:ta OR prevent*:ab OR ‘preventive health service’/mj OR 'health promotion'/mj OR
'health promotion':ta OR 'health promotion':ab OR 'nutrition'/exp/mj OR 'physical
activity'/exp) AND ('control group'/mj OR controls:ab OR controls:ta OR comparator:ta OR
comparator:ab OR ‘control school’:ab OR ‘control school’:ta OR 'comparison group':ta OR
'comparison group':ab OR 'intervention study'/mj OR 'evaluation'/mj OR 'evaluation and
follow up'/mj) AND ('body fat':ta OR 'body fat':ab OR 'body mass'/mj OR bmi:ta OR bmi:ab
OR 'anthropometry'/exp OR 'body composition'/mj OR 'treatment outcome'/mj OR 'outcome
125
assessment'/mj OR 'follow up'/mj OR outcome*:ta OR outcome*:ab) AND ([english]/lim OR
[french]/lim OR [german]/lim OR [spanish]/lim) AND [humans]/lim
CINAHL: ((MH “Obesity+/PC”) OR (TI “Overweight”) OR (AB “Overweight”) OR (TI
“Obes*”) OR (AB “obes*”)) AND ((MM “Preventive trials”) OR (TI “Prevent*”) OR (AB
“Prevent*”) OR (MM “Health promotion”) OR (AB “Health promotion”) OR (MH “Diet”)
OR (MH "Physical Activity") OR (MH "Exercise")) AND ((AB "control group") OR (AB
”controls”) OR (AB "comparison group") OR (MH “Intervention trials”) OR (MH
"Programme Evaluation")) AND ((TI ”Body fat”) OR (AB “Body fat”) OR (AB “Bmi”) OR
(MH “Body mass index”) OR (TI “Body mass index”) OR (AB “Body mass index”) OR (MH
“Anthropometry”) OR (MH “body composition”) OR (MH “Treatment outcomes”) OR (MH
“Outcome assessment”) OR (AB “Outcome*”) OR (MH “Prospective Studies”)))
Cochrane Central Register of Controlled Trials:
# 1 MeSH descriptor Overweight explode all trees
#2 MeSH descriptor Primary Prevention explode all trees
# 3 MeSH descriptor Anthropometry explode trees 2 and 3
# 4 (#1 AND ( #2 OR prevent* ) AND #3)
126
Appendix 2
Formulae used in the meta-analysis
Source: Higgins JPT, Green S (editors). Cochrane Handbook for Systematic Reviews of Interventions Version
5.0.2 [updated September 2009]. The Cochrane Collaboration, 2009. Available from www.cochrane-
handbook.org.
Imputing standard deviations for changes from baseline
Formula 1a: Correlation coefficient (CorrI,C) from a similar study that reports the standard deviations of the mean
at baseline (SDbaseline), the final mean (SDFinal) and of the mean change (SDChange) for both intervention (I) and
control group (C)
!"##
!=!
!"!!(!"#$%&'$)
!!!"!!(!"#$%)
!!!"!(!!!"#$)
!
!∗!"!!(!"#$%&'$)∗!"!!(!"#$%)
Analogue for Corrc
Formula 1b: Imputation of standard deviation of change from baseline
!"!!!!"#$ !=!!"!!!"#$%&'$
!+!!"!!(!"#$%)
!−!(2∗!"##
!∗!"!!!"#$%&'$ ∗!"!!!"#$% )
Analogue for SDC (change)
Combining groups
Formula 2a: Number of participants (Ncombined) for combining groups
Ncombined = N1 + N2
Formula 2b: Mean change from baseline (Mcombined) for combined groups
!!"#$%&'( =!
!!!!+!!!!
!!+!!
!
Formula 2c: Standard deviation of mean change from baseline (SDcombined) for combining groups
!"!"#$%&'( =!
!!−1!"!
!+!!−1!"!
!+!!!!
!!+!!(!!
!+!!
!−2!!!!)
!!+!!−1
Adjusting for cluster design
Formula 3a: Design effect (DE) for the additional adjustment of cluster design
DE = 1 + [(C-1) * ICC]
(C = mean cluster size; ICC = intracluster correlation coefficient)
Formula 3b: Adjusted standard error (SEadjusted) of the outcome estimate
SE(adjusted) = SE(unadjusted)* √DE
127
Appendix 3
Characteristics of 23 systematic reviews included
Reference
Identified in
Database
Number of studies
obtained from
systematic review
Targeted age
group
Special focus
Study designs
included
Bautista-Castano I, Doreste J, Serra-Majem L. Effectiveness of interventions
in the prevention of childhood obesity. Eur J Epidemiol 2004;19:617-22.
DARE
14
Children and
adolescents
RCT and non-
randomized trials
Berry D, Sheehan R, Heschel R, Knafl K, Melkus G, Grey M. Family-based
interventions for childhood obesity: a review. J Fam Nurs 2004:429-49.
DARE
13
Children
Family
RCT
Brown T, Summerbell C. Systematic review of school-based interventions
that focus on changing dietary intake and physical activity levels to prevent
childhood obesity: an update to the obesity guidance produced by the
National Institute for Health and Clinical Excellence. Obes Rev 2009;10:110-
41.
PubMed
38
Children
School
RCT and non-
randomized trials
Campbell K, Waters E, O'Meara S, Summerbell C. Interventions for
preventing obesity in childhood. A systematic review. Obes Rev 2001;2:149-
57.
DARE
7
Children and
adolescents
RCT and non-
randomized trials
Campbell KJ, Hesketh KD. Strategies which aim to positively impact on
weight, physical activity, diet and sedentary behaviours in children from zero
to five years. A systematic review of the literature. Obes Rev 2007;8:327-38.
PubMed
9
Young children
Wide range
including
uncontrolled
studies
Connelly JB, Duaso MJ, Butler G. A systematic review of controlled trials of
interventions to prevent childhood obesity and overweight: a realistic
synthesis of the evidence. Public Health 2007;121:510-7.
DARE
28
Children and
adolescents
RCT and non-
randomized trials
DeMattia L, Lemont L, Meurer L. Do interventions to limit sedentary
behaviours change behaviour and reduce childhood obesity? A critical review
of the literature. Obes Rev 2007;8:69-81.
DARE
12
Children and
adolescents
Sedentary
behaviour
RCT and non-
randomized trials
128
Doak CM, Visscher TL, Renders CM, Seidell JC. The prevention of
overweight and obesity in children and adolescents: a review of interventions
and programmes. Obes Rev 2006;7:111-36.
DARE
25
Children and
adolescents
School
RCT and non-
randomized trials
Douketis JD, Feightner JW, Attia J, Feldman WF. Periodic health
examination, 1999 update: 1. Detection, prevention and treatment of obesity.
Canadian Task Force on Preventive Healthcare. CMAJ 1999;160:513-25.
DARE
3 (for prevention)
Adults
Prospective cohort
study or RCT
Flodmark CE, Marcus C, Britton M. Interventions to prevent obesity in
children and adolescents: a systematic literature review. Int J Obes
2006;30:579-589.
DARE
10
Children and
adolescents
RCT and non-
randomized trials
Flynn MA, McNeil DA, Maloff B, Mutasingwa D, Wu M, Ford C, et al.
Reducing obesity and related chronic disease risk in children and youth: a
synthesis of evidence with 'best practice' recommendations. Obes Rev 2006;7
Suppl 1:7-66.
PubMed
101 (excluded those
ranked 'low
methodological rigour')
Children and
adolescents
RCT and non-
randomized trials
Fogelholm M, Lahti-Koski M. Community health-promotion interventions
with physical activity: does this approach prevent obesity? Scand J Nutr
2002:173-7.
DARE
5
Overall
Population
Community
interventions,
Physical
activity
Cohorts and
cross-sectional
surveys, with or
without control
Hardeman W, Griffin S, Johnston M, Kinmonth AL, Wareham NJ.
Interventions to prevent weight gain: a systematic review of psychological
models and behaviour change methods. Int J Obes Relat Metab Disord
2000;24:131-43.
DARE
9
Overall
Population
RCT and non-
randomized trials
Jerum A, Melnyk BM. Effectiveness of interventions to prevent obesity and
obesity-related complications in children and adolescents. Pediatr Nurs
2001;27:606-10.
DARE
3
Children and
adolescents
RCT
Kropski JA, Keckley PH, Jensen GL. School-based obesity prevention
programmes: an evidence-based review. Obesity 2008;16:1009-18.
DARE
14
Children and
adolescents
School
RCT and non-
randomized trials
Lemmens VE, Oenema A, Klepp KI, Henriksen HB, Brug J. A systematic
review of the evidence regarding efficacy of obesity prevention interventions
among adults. Obes Rev 2008;9:446-55.
PubMed
9
Adults
RCT and non-
randomized trials
Lissau I. Prevention of overweight in the school arena. Acta Paediatr 2007;96
Suppl:12-8.
DARE
14
Children and
adolescents
School
RCT and non-
randomized trials
Reilly JJ, McDowell ZC. Physical activity interventions in the prevention and
treatment of paediatric obesity: systematic review and critical appraisal. Proc
Nutr Soc 2003;62:611-9.
PubMed
2 (only prevention
studies included)
Children and
adolescents
RCT
129
Saunders KL. Preventing obesity in pre-school children: a literature review.
J Public Health. 2007;29:368-75.
PubMed
6
Pre-school
children
Wide variety of
study designs
including
observational
studies
Small L, Anderson D, Melnyk BM. Prevention and early treatment of
overweight and obesity in young children: a critical review and appraisal of
the evidence. Pediatr Nurs. 2007;33:149-52, 55-61, 27.
DARE
6 (only prevention
studies included)
Young children
School
RCT
Stice E, Shaw H, Marti CN. A meta-analytic review of obesity prevention
programmes for children and adolescents: the skinny on interventions that
work. Psychol Bull. 2006;132:667-91.
DARE
46
Children and
adolescents
RCT and non-
randomized trials
Summerbell CD, Waters E, Edmunds LD, Kelly S, Brown T, Campbell KJ.
Interventions for preventing obesity in children. Cochrane Database Syst
Rev. 2005:CD001871.
CDSR
22
Children and
adolescents
RCT and non-
randomized trials
Wofford LG. Systematic review of childhood obesity prevention. J Pediatr
Nurs. 2008;23:5-19.
PubMed
3 (only intervention
studies included)
Children and
adolescents
RCT and non-
randomized trials
CDSR: Cochrane Database of Systematic Reviews; DARE: Database of Abstracts and Reviews of Effects; RCT: Randomised controlled trial
130
Appendix 4
Studies included in the meta-analyses by type of intervention, outcome measure and age group
Age group
(Outcome
measure)
Physical activity
Education
Nutrition
Reduced television viewing in
combination with other
strategies
Physical activity
and education
Education and
nutrition
Physical activity,
education and
nutrition
Children
(BMI)
16, 22, 23, 28, 34, 38, 41, 49,
52, 66, 68
1, 8, 23, 25, 29, 31,
32, 37, 41, 46, 47,
49, 50, 63, 66
12
7, 8, 17, 25, 33, 48, 49, 53
2-5, 7, 15, 17, 18,
20, 24, 33, 41-43,
51, 57, 59, 61, 62,
64, 66
19, 46
6, 10, 21, 40, 53, 58
Children
(%BF)
14, 38, 52, 55, 66, 68
60, 66
60
3,15, 30, 51, 57, 66
65
6,9
Adults
(BMI)
9, 44
11, 13, 26, 27, 36,
39, 45, 56
44, 67
Adults
(%BF)
9, 54
13, 35, 45, 56
Study ID
References of 103 articles included (reporting on 68 studies)
1
Amaro S, Viggiano A, Di Costanzo A et al. Kaledo, a new educational board-game, gives nutritional rudiments and encourages healthy eating in children: A pilot cluster
randomized trial. European Journal of Pediatrics 2006;165:630-5.
2
Baranowski T, Baranowski JC, Cullen KW et al. The Fun, Food, and Fitness Project (FFFP): the Baylor GEMS pilot study. Ethn Dis 2003;13 1 Suppl 1:30-9.
3
Bayne-Smith M, Fardy PS, Azzollini A, Magel J, Schmitz KH, Agin D. Improvements in heart health behaviours and reduction in coronary artery disease risk factors in urban
teenaged girls through a school-based intervention: the PATH programme. Am J Public Health 2004;94:1538-43.
4
Beech BM, Klesges RC, Kumanyika SK et al. Child- and parent-targeted interventions: the Memphis GEMS pilot study. Ethn Dis 2003;13 1 Suppl 1:40-53.
5
Burke V, Milligan RA, Thompson C et al. A controlled trial of health promotion programmes in 11-year-olds using physical activity "enrichment" for higher risk children. J
Pediatr 1998;132:840-8.
6
Caballero B, Clay T, Davis SM et al. Pathways: A school-based, randomized controlled trial for the prevention of obesity in American Indian schoolchildren. American Journal of
Clinical Nutrition 2003;78:1030-8.
Going S, Thompson J, Cano S et al. The effects of the Pathways Obesity Prevention Programme on physical activity in American Indian children. Prev Med 2003;37:S62-9.
Steckler A, Ethelbah B, Martin CJ et al. Pathways process evaluation results: a school-based prevention trial to promote healthful diet and physical activity in American Indian
third, fourth, and fifth grade students. Prev Med 2003;37:S80-90.
131
Study ID
References of 103 articles included (reporting on 68 studies)
Stevens J, Story M, Ring K et al. The impact of the Pathways intervention on psychosocial variables related to diet and physical activity in American Indian schoolchildren.
Preventive Medicine 2003;37 Suppl 1:70-9.
Stone EJ, Norman JE, Davis SM et al. Design, implementation, and quality control in the Pathways American-Indian multicenter trial. Prev Med 2003;37:S13-23.
Davis CE, Hunsberger S, Murray DM et al. Design and statistical analysis for the pathways study. American Journal of Clinical Nutrition 1999;69 Suppl 4:760-3.
Caballero B, Davis S, Davis C et al. Pathways: A school-based programme for the primary prevention of obesity in American Indian children. Journal of Nutritional Biochemistry
1998;9:535-43.
Gittelsohn J, Evans M, Helitzer D et al. Formative research in a school-based obesity prevention programme for Native American school children (Pathways). Health Educ Res
1998;13:251-65.
Davis S, Gomez Y, Lambert L, Skipper B. Primary prevention of obesity in American Indian children. Ann N Y Acad Sci 1993;699:167-80.
7
Chavarro JE, Peterson KE, Sobol AM, Wiecha JL, Gortmaker SL. Effects of a school-based obesity-prevention intervention on menarche (United States). Cancer Causes Control
2005;16:1245-52.
Gortmaker SL, Peterson K, Wiecha J et al. Reducing obesity via a school-based interdisciplinary intervention among youth: Planet Health. Arch Pediatr Adolesc Med
1999;153:409-18.
8
Dennison BA, Russo TJ, Burdick PA, Jenkins PL. An intervention to reduce television viewing by preschool children. Arch Pediatr Adolesc Med 2004;158:170-6.
9
Donnelly JE, Hill JO, Jacobsen DJ et al. Effects of a 16-month randomized controlled exercise trial on body weight and composition in young, overweight men and women: the
Midwest Exercise Trial. Arch Intern Med 2003;163:1343-50.
Kirk EP, Jacobsen DJ, Gibson C, Hill JO, Donnelly JE. Time course for changes in aerobic capacity and body composition in overweight men and women in response to long-
term exercise: the Midwest Exercise Trial (MET). Int J Obes Relat Metab Disord 2003;27:912-9.
10
Donnelly JE, Jacobsen DJ, Whatley JE et al. Nutrition and physical activity programme to attenuate obesity and promote physical and metabolic fitness in elementary school
children. Obesity Research 1996;4:229-43.
11
Dzator JA, Hendrie D, Burke V et al. A randomized trial of interactive group sessions achieved greater improvements in nutrition and physical activity at a tiny increase in cost. J
Clin Epidemiol 2004;57:610-9.
Burke V, Giangiulio N, Gillam HF, Beilin LJ, Houghton S. Physical activity and nutrition programmes for couples: a randomized controlled trial. J Clin Epidemiol 2003;56:421-
32.
12
Ebbeling CB, Feldman HA, Osganian SK, Chomitz VR, Ellenbogen SJ, Ludwig DS. Effects of decreasing sugar-sweetened beverage consumption on body weight in adolescents:
a randomized, controlled pilot study. Pediatrics 2006;117:673-80.
13
Eiben G, Lissner L. Health Hunters--an intervention to prevent overweight and obesity in young high-risk women. Int J Obes (Lond) 2006;30:691-6.
14
Eliakim A, Makowski GS, Brasel JA, Cooper DM. Adiposity, lipid levels, and brief endurance training in nonobese adolescent males. Int J Sports Med 2000;21:332-7.
15
Eliakim A, Nemet D, Balakirski Y, Epstein Y. The effects of nutritional-physical activity school-based intervention on fatness and fitness in preschool children. J Pediatr
Endocrinol Metab 2007;20:711-8.
16
Ewart CK, Young DR, Hagberg JM. Effects of school-based aerobic exercise on blood pressure in adolescent girls at risk for hypertension. Am J Public Health 1998;88:949-51.
17
Fitzgibbon ML, Stolley MR, Schiffer L, Van Horn L, Kauferchristoffel K, Dyer A. Two-year follow-up results for Hip-Hop to Health Jr.: A randomized controlled trial for
overweight prevention in preschool minority children. Journal of Pediatrics 2005;146:618-25.
Fitzgibbon ML, Stolley MR, Dyer AR, VanHorn L, KauferChristoffel K. A community-based obesity prevention programme for minority children: rationale and study design for
Hip-Hop to Health Jr. Prev Med 2002;34:289-97.
132
Study ID
References of 103 articles included (reporting on 68 studies)
Stolley MR, Fitzgibbon ML, Dyer A, Van Horn L, KauferChristoffel K, Schiffer L. Hip-Hop to Health Jr., an obesity prevention programme for minority preschool children:
baseline characteristics of participants. Prev Med 2003;36:320-9.
18
Flores R. Dance for health: improving fitness in African American and Hispanic adolescents. Public Health Rep 1995;110:189-93.
19
Foster GD, Sherman S, Borradaile KE et al. A policy-based school intervention to prevent overweight and obesity. Pediatrics 2008;121:e794-802.
20
Graf C, Koch B, Falkowski G et al. School-based prevention: Effects on obesity and physical performance after 4 years. Journal of Sports Sciences 2008;26:987-94.
Graf C, Koch B, Falkowski G et al. Effects of a school-based intervention on BMI and motor abilities in childhood. Journal of Sports Science and Medicine 2005;4:291-9.
21
Haerens L, Deforche B, Maes L, Stevens V, Cardon G, De Bourdeaudhuij I. Body mass effects of a physical activity and healthy food intervention in middle schools. Obesity
2006;14:847-54.
22
Hansen HS, Froberg K, Hyldebrandt N, Nielsen JR. A controlled study of eight months of physical training and reduction of blood pressure in children: the Odense schoolchild
study. BMJ 1991;303:682-5.
23
Harrell JS, Gansky SA, McMurray RG, Bangdiwala SI, Frauman AC, Bradley CB. School-based interventions improve heart health in children with multiple cardiovascular
disease risk factors. Pediatrics 1998;102:371-80.
24
Harrell JS, McMurray RG, Bangdiwala SI, Frauman AC, Gansky SA, Bradley CB. Effects of a school-based intervention to reduce cardiovascular disease risk factors in
elementary-school children: the Cardiovascular Health in Children (CHIC) study. J Pediatr 1996;128:797-805.
25
Harrison M, Burns CF, McGuinness M, Heslin J, Murphy NM. Influence of a health education intervention on physical activity and screen time in primary school children:
'Switch Off--Get Active'. J Sci Med Sport 2006;9:388-94.
26
Hivert MF, Langlois MF, Berard P, Cuerrier JP, Carpentier AC. Prevention of weight gain in young adults through a seminar-based intervention programme. Int J Obes (Lond)
2007;31:1262-9.
27
Howard BV, Manson JE, Stefanick ML et al. Low-fat dietary pattern and weight change over 7 years: the Women's Health Initiative Dietary Modification Trial. JAMA
2006;295:39-49.
28
Jago R, Jonker ML, Missaghian M, Baranowski T. Effect of 4 weeks of Pilates on the body composition of young girls. Prev Med 2006;42:177-80.
29
James J. School anti-"fizzy drinks" programme helps to prevent obesity in children. Evidence-Based Healthcare and Public Health 2004;8:368-9.
James J, Thomas P, Cavan D, Kerr D. Preventing childhood obesity by reducing consumption of carbonated drinks: cluster randomised controlled trial. BMJ 2004;328:1237.
James J, Thomas P, Kerr D. Preventing childhood obesity: two year follow-up results from the Christchurch obesity prevention programme in schools (CHOPPS). BMJ
2007;335:762.
30
Jamner MS, Spruijt-Metz D, Bassin S, Cooper DM. A controlled evaluation of a school-based intervention to promote physical activity among sedentary adolescent females:
project FAB. J Adolesc Health 2004;34:279-89.
31
Kafatos I, Manios Y, Moschandreas J, Kafatos A. Health and nutrition education programme in primary schools of Crete: changes in blood pressure over 10 years. Eur J Clin Nutr
2007;61:837-45.
Manios Y, Kafatos A, Apostolaki I et al. Health and nutrition education in primary schools in Crete: 10 years' follow-up of serum lipids, physical activity and macronutrient
intake. British Journal of Nutrition 2006;95:568-75.
Manios Y, Kafatos A, Mamalakis G. The effects of a health education intervention initiated at first grade over a 3 year period: physical activity and fitness indices. Health Educ
Res 1998;13:593-606.
Manios Y, Moschandreas J, Hatzis C, Kafatos A. Evaluation of a health and nutrition education programme in primary school children of Crete over a three-year period. Prev Med
1999;28:149-59.
Manios Y, Moschandreas J, Hatzis C, Kafatos A. Health and nutrition education in primary schools of Crete: changes in chronic disease risk factors following a 6-year
intervention programme. Br J Nutr 2002;88:315-24.
133
Study ID
References of 103 articles included (reporting on 68 studies)
32
Killen JD, Telch MJ, Robinson TN, Maccoby N, Taylor CB, Farquhar JW. Cardiovascular disease risk reduction for tenth graders. A multiple-factor school-based approach.
JAMA 1988;260:1728-33.
33
Kipping RR, Payne C, Lawlor DA. Randomised controlled trial adapting US school obesity prevention to England. Arch Dis Child 2008;93:469-73.
34
Lazaar N, Aucouturier J, Ratel S, Rance M, Meyer M, Duche P. Effect of physical activity intervention on body composition in young children: Influence of body mass index
status and gender. Acta Paediatrica, International Journal of Paediatrics 2007;96:1315-20.
35
Leermarkers EA, Jakicic JM, Viteri J, Wing RR. Clinic-based vs. home-based interventions for preventing weight gain in men. Obes Res 1998;6:346-52.
36
Levine MD, Klem ML, Kalarchian MA et al. Weight gain prevention among women. Obesity (Silver Spring) 2007;15:1267-77.
37
Lionis C, Kafatos A, Vlachonikolis J, Vakaki M, Tzortzi M, Petraki A. The effects of a health education intervention programme among Cretan adolescents. Prev Med
1991;20:685-99.
38
Martinez Vizcaino V, Salcedo Aguilar F, Franquelo Gutierrez R et al. Assessment of an after-school physical activity programme to prevent obesity among 9- to 10-year-old
children: a cluster randomized trial. Int J Obes (Lond) 2008;32:12-22.
39
Matvienko O, Lewis DS, Schafer E. A college nutrition science course as an intervention to prevent weight gain in female college freshmen. J Nutr Educ 2001;33:95-101.
40
McKenzie TL, Stone EJ, Feldman HA et al. Effects of the CATCH physical education intervention: teacher type and lesson location. Am J Prev Med 2001;21:101-9.
Luepker RV, Perry CL, McKinlay SM et al. Outcomes of a field trial to improve children's dietary patterns and physical activity. The Child and Adolescent Trial for
Cardiovascular Health. CATCH collaborative group. JAMA 1996;275:768-76.
Lytle LA, Stone EJ, Nichaman MZ et al. Changes in nutrient intakes of elementary school children following a school-based intervention: results from the CATCH Study. Prev
Med 1996;25:465-77.
Nader PR, Stone EJ, Lytle LA et al. Three-year maintenance of improved diet and physical activity: the CATCH cohort. Child and Adolescent Trial for Cardiovascular Health.
Arch Pediatr Adolesc Med 1999;153:695-704.
Perry CL, Stone EJ, Parcel GS et al. School-based cardiovascular health promotion: the child and adolescent trial for cardiovascular health (CATCH). J Sch Health 1990;60:406-
13.
Webber LS, Osganian SK, Feldman HA et al. Cardiovascular risk factors among children after a 2 1/2-year intervention-The CATCH Study. Prev Med 1996;25:432-41.
41
McMurray RG, Harrell JS, Bangdiwala SI, Bradley CB, Deng S, Levine A. A school-based intervention can reduce body fat and blood pressure in young adolescents. J Adolesc
Health 2002;31:125-32.
42
Nader P, Sallis J, Abramson I et al. Family-based cardiovascular risk reduction education among Mexican- and Anglo-Americans. Family Commun Health 1992;15:57–74.
43
Neumark-Sztainer D, Story M, Hannan PJ, Rex J. New Moves: A school-based obesity prevention programme for adolescent girls. Preventive Medicine 2003;37:41-51.
44
Pritchard JE, Nowson CA, Wark JD. A worksite programme for overweight middle-aged men achieves lesser weight loss with exercise than with dietary change. J Am Diet Assoc
1997;97:37-42.
45
Proper KI, Hildebrandt VH, Van der Beek AJ, Twisk JWR, Van Mechelen W. Effect of individual counseling on physical activity fitness and health: a randomized controlled trial
in a workplace setting. Am J Prev Med 2003;24:218-26.
46
Puska P, Vartiainen E, Pallonen U et al. The North Karelia youth project: evaluation of two years of intervention on health behaviour and CVD risk factors among 13- to 15-year
old children. Prev Med 1982;11:550-70.
47
Robbins LB, Gretebeck KA, Kazanis AS, Pender NJ. Girls on the move programme to increase physical activity participation. Nurs Res 2006;55:206-16.
48
Robinson TN. Reducing children's television viewing to prevent obesity: a randomized controlled trial. JAMA 1999;282:1561-7.
49
Robinson TN, Killen JD, Kraemer HC et al. Dance and reducing television viewing to prevent weight gain in African-American girls: the Stanford GEMS pilot study. Ethn Dis
2003;13 1 Suppl 1:65-77.
Robinson TN. Can a school - based intervention to reduce television use decrease adiposity in children in grades 3 and 4 ? West J Med 2000;173:40.
134
Study ID
References of 103 articles included (reporting on 68 studies)
50
Rodgers D, Johnson S, Tschann J, Chesterman E, Mellin L. The evaluation of a school-based obesity programme. 2001 [08 March 2010]; Available from: http://www.just-for-
kids.org/links.htm
51
Rosenbaum M, Nonas C, Weil R et al. School-based intervention acutely improves insulin sensitivity and decreases inflammatory markers and body fatness in junior high school
students. J Clin Endocrinol Metab 2007;92:504-8.
52
Sadowsky HS, Sawdon JM, Scheiner ME, Sticklin AM. Eight week moderate intensity exercise intervention elicits body composition change in adolescents. Cardiopulmonary
Physical Therapy Journal 1999;10:38-44.
Bell AC, Simmons A, Sanigorski AM, Kremer PJ, Swinburn BA. Preventing childhood obesity: the sentinel site for obesity prevention in Victoria, Australia. Health Promot Int
2008;23:328-36.
53
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 programme, Be Active Eat Well. Int J Obes (Lond) 2008;32:1060-7.
54
Schmitz KH, Jensen MD, Kugler KC, Jeffery RW, Leon AS. Strength training for obesity prevention in midlife women. Int J Obes Relat Metab Disord 2003;27:326-33.
55
Siegel J, Camaione D, Manfredi T. The effects of upper body resistance training on prepubescent children Pediatric Exercise Science 1989;1:145–15.
56
Simkin-Silverman LR, Wing RR, Boraz MA, Kuller LH. Lifestyle intervention can prevent weight gain during menopause: results from a 5-year randomized clinical trial. Ann
Behav Med 2003;26:212-20.
57
Simon C, Schweitzer B, Oujaa M et al. Successful overweight prevention in adolescents by increasing physical activity: a 4-year randomized controlled intervention. Int J Obes
(Lond) 2008;32:1489-98.
Simon C, Wagner A, DiVita C et al. Intervention centred on adolescents' physical activity and sedentary behaviour (ICAPS): concept and 6-month results. Int J Obes Relat Metab
Disord 2004;28 Suppl 3:96-103.
Simon C, Wagner A, Platat C et al. ICAPS: a multilevel programme to improve physical activity in adolescents. Diabetes Metab 2006;32:41-9.
58
Singh AS, Paw MJMCA, Brug J, Van Mechelen W. Short-term effects of school-based weight gain prevention among adolescents. Archives of Pediatrics and Adolescent
Medicine 2007;161:565-71.
Singh AS, Chin A Paw MJM, Kremers SPJ, Visscher TLS, Brug J, van Mechelen W. Design of the Dutch Obesity Intervention in Teenagers (NRG-DOiT): systematic
development, implementation and evaluation of a school-based intervention aimed at the prevention of excessive weight gain in adolescents. BMC Public Health 2006;6:304.
59
Spiegel SA, Foulk D. Reducing overweight through a multidisciplinary school-based intervention. Obesity (Silver Spring) 2006;14:88-96.
60
Spruijt-Metz D, Nguyen-Michel ST, Goran MI, Chou CP, Huang TTK. Reducing sedentary behaviour in minority girls via a theory-based, tailored classroom media intervention.
International Journal of Pediatric Obesity 2008;3:240-8.
61
Stock S, Miranda C, Evans S et al. Healthy Buddies: a novel, peer-led health promotion programme for the prevention of obesity and eating disorders in children in elementary
school. Pediatrics 2007;120:e1059-68.
62
Story M, Sherwood NE, Himes JH et al. An after-school obesity prevention programme for African-American girls: the Minnesota GEMS pilot study. Ethn Dis 2003;13 1 Suppl
1:54-64.
Rochon J, Klesges RC, Story M et al. Common design elements of the Girls health Enrichment Multi-site Studies (GEMS). Ethn Dis 2003;13 1 Suppl 1:6-14.
Story M, Sherwood NE, Obarzanek E et al. Recruitment of African-American pre-adolescent girls into an obesity prevention trial: the GEMS pilot studies. Ethn Dis 2003;13 1
Suppl 1:78-87.
63
Tamir D, Feurstein A, Brunner S, Halfon ST, Reshef A, Palti H. Primary prevention of cardiovascular diseases in childhood: changes in serum total cholesterol, high density
lipoprotein, and body mass index after 2 years of intervention in Jerusalem schoolchildren age 7-9 years. Prev Med 1990;19:22-30.
64
Taylor RW, McAuley KA, Barbezat W, Strong A, Williams SM, Mann JI. APPLE Project: 2-y findings of a community-based obesity prevention programme in primary school
age children. Am J Clin Nutr 2007;86:735-42.
135
Study ID
References of 103 articles included (reporting on 68 studies)
Taylor RW, Mcauley KA, Williams SM, Barbezat W, Nielsen G, Mann JI. Reducing weight gain in children through enhancing physical activity and nutrition: the APPLE project.
Int J Pediatr Obes 2006;1:146-52.
65
Trevino RP. Effect of the Bienestar health programme on physical fitness in low-income Mexican–American Children. Hispanic J Behav Sci 2005;27:120-32.
Trevino RP, Yin Z, Hernandez A, Hale DE, Garcia OA, Mobley C. Impact of the Bienestar school-based diabetes mellitus prevention programme on fasting capillary glucose
levels: a randomized controlled trial. Arch Pediatr Adolesc Med 2004;158:911-7.
66
Vandongen R, Jenner DA, Thompson C et al. A controlled evaluation of a fitness and nutrition intervention programme on cardiovascular health in 10- to 12-year-old children.
Prev Med 1995;24:9-22.
67
Velthuis-te Wierik EJ, van den Berg H, Weststrate JA, van het Hof KH, de Graaf C. Consumption of reduced-fat products: effects on parameters of anti-oxidative capacity. Eur J
Clin Nutr 1996;50:214-9.
68
Yin Z, Gutin B, Johnson MH et al. An environmental approach to obesity prevention in children: Medical College of Georgia FitKid Project year 1 results. Obes Res
2005;13:2153-61.
Gutin B, Yin Z, Johnson M, Barbeau P. Preliminary findings of the effect of a 3-year after-school physical activity intervention on fitness and body fat: the Medical College of
Georgia Fitkid Project. Int J Pediatr Obes 2008;3 Suppl 1:3-9.
136
Appendix 5
a) Details and outcome estimates of 68 studies included in the meta-analysis (see appendix 4 for Study ID references)
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
1
Italy
C-RCT
x
241
BMI
-0.2
0.12
mf
1
Outcome data were
provided by the
author.
12-18
6
0
2
USA
RCT
x
x
31
BMI
0.6
1.6
f
3
6-11
3
0
3
USA
Q-E
x
x
442
BMI
-0.1
0.15
f
2
SFT
12-18
3
0
433
%BF
-0.8
0.2
f
3
4
USA
RCT
x
x
39
BMI
-0.38
0.25
f
3
6-11
3
0
5
Australia
C-RCT
x
x
485
BMI
0.02
0.05
mf
com
Outcome data were
provided by the
author. Combined
outcome estimates
of both intervention
groups were used in
the meta-analysis.
6-11
6
0
7
USA
C-RCT
x
x
x
508
BMI
-0.3
0.1
f
3
BIA
Outcome data were
taken from
Chavarro et al.
2005.
6-11
19
0
6
USA
C-RCT
x
x
x
1409
BMI
-0.2
0.17
mf
3
Data were taken
from Caballero
2003.
6-11
36
0
1409
%BF
0.2
0.55
mf
3
8
USA
C-RCT
x
x
77
BMI
-0.36
0.44
mf
3
0-5
10
0
9
USA
RCT
x
31
BMI
-1.4
0.4
m
1
HW
Outcome data a
were taken from
Donnelly 2003 and
were combined for
male and female
participants.
>18
16
0
31
%BF
-2.9
0.62
m
1
43
BMI
-1
0.54
f
1
43
%BF
-1.9
0.35
f
1
74
BMI
-1.07
0.38
mf
com
74
%BF
-2.17
0.44
mf
com
137
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
10
USA
Q-E
x
x
x
108
BMI
-0.2
0.2
mf
1
HW
6-11
24
0
108
%BF
1.3
0.48
mf
1
11
Australia
C-RCT
x
274
BMI
-0.15
0.12
mf
com
Outcome data were
taken from Dzator
2004 and were
combined for both
intervention groups
>18
4
8
13
Sweden
RCT
x
30
BMI
-2.2
0.77
f
2
DXA
>18
12
0
30
%BF
-3.9
1.98
f
2
14
USA
RCT
x
38
%BF
-0.7
0.64
m
1
SFT
12-18
1
0
15
Israel
C-RCT
x
x
101
BMI
-0.3
0.13
mf
1
SFT
0-5
3
0
101
%BF
-2.72
0.98
mf
2
12
USA
RCT
x
103
BMI
-0.14
0.21
mf
3
12-18
6
0
16
USA
RCT
x
88
BMI
0.3
0.28
f
1
12-18
6
0
17
USA
C-RCT
x
x
x
300
BMI
-0.54
0.22
mf
3
Outcome data were
taken from
Fitzgibbon 2005.
0-5
3
24
18
USA
C-RCT
x
x
49
BMI
-1.1
0.37
f
1
Outcome data for
male participants
were lacking
standard deviations
and could not be
used in the meta-
analysis.
12-18
3
0
19
USA
C-RCT
x
x
844
BMI
-0.04
0.12
mf
3
6-11
24
0
20
Germany
C-RCT
x
x
586
BMI
0.6
0.17
mf
2
6-11
48
0
21
Belgium
C-RCT
x
x
x
2291
BMI
-0.15
0.04
mf
com
Combined outcome
estimates of both
intervention groups
were used in the
meta-analysis.
12-18
21
6
851
BMI
-0.48
0.06
f
com
1440
BMI
0.18
0.05
m
com
22
Denmark
RCT
x
132
BMI
0.2
0.12
mf
1
6-11
8
0
24
USA
C-RCT
x
x
1274
BMI
0.11
0.07
mf
3
6-11
2
0
138
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
23
USA
C-RCT
x
314
BMI
0.3
0.16
mf
2
6-11
2
0
x
258
BMI
0.17
0.16
mf
2
25
Ireland
Q-E
x
x
284
BMI
-0.08
0.15
mf
3
6-11
4
0
26
Canada
RCT
x
115
BMI
-0.4
0.28
mf
2
>18
24
0
27
USA
RCT
x
41173
BMI
-0.3
0.03
f
3
>18
90
0
28
USA
RCT
x
30
BMI
-0.4
0.46
f
1
6-11
1
0
29
UK
C-RCT
x
418
BMI
-0.26
0.16
mf
2
Outcome data were
taken from James
2007.
6-11
12
24
30
USA
Q-E
x
x
47
%BF
0.08
0.32
mf
1
DXA
12-18
4
0
31
Greece
Q-E
x
641
BMI
-0.6
0.22
mf
2
Data were taken
from Manios 2002.
6-11
72
0
32
USA
C-RCT
x
1130
BMI
-0.24
0.05
mf
com
Combined outcome
estimates for male
and female
participants were
used in the meta-
analysis.
12-18
2
0
518
BMI
-0.2
0.08
f
1
612
BMI
-0.3
0.06
m
1
33
UK
C-RCT
x
x
x
472
BMI
0.1
0.19
mf
3
6-11
5
0
34
France
Q-E
x
420
BMI
-0.38
0.04
mf
com
Combined outcome
estimates of both
study groups were
used in the meta-
analysis.
6-11
6
0
210
BMI
-0.45
0.06
f
com
210
BMI
-0.32
0.06
m
com
35
USA
RCT
x
38
%BF
-0.8
0.99
m
2
BIA
>18
4
0
36
USA
RCT
x
205
BMI
-0.08
0.27
f
2
Outcome data were
provided by the
author and were
combined for both
intervention groups.
>18
24
12
37
Greece
Q-E
x
148
BMI
-0.51
0.2
mf
2
12-18
9
0
139
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
38
Spain
C-RCT
x
514
BMI
0.07
0.1
m
3
BIA
Combined outcome
estimates for male
and female
participants were
used in the meta-
analysis.
6-11
9
514
%BF
-0.37
0.25
m
3
530
BMI
-0.12
0.1
f
3
530
%BF
-0.58
0.24
f
3
1044
BMI
-0.02
0.09
mf
com
1044
%BF
-0.48
0.17
mf
com
39
USA
RCT
x
33
BMI
-1.6
0.66
f
1
>18
4
16
40
USA
C-RCT
x
x
x
3748
BMI
0
0.1
mf
3
Outcome data were
taken from Nader
1999.
6-11
24
36
41
USA
C-RCT
x
566
BMI
-0.02
0.08
mf
2
Outcome data for
males and females
could not be used in
the subgroup
analysis, because
the number of
participants in each
group was not
reported.
6-11
2
0
x
513
BMI
0.03
0.09
mf
2
x
x
555
BMI
0
0.08
mf
2
42
USA
C-RCT
x
x
497
BMI
0.22
0.26
mf
com
Combined outcome
estimates of both
study groups were
used in the meta-
analysis.
6-11
12
48
BMI
-0.05
0.43
f
com
43
USA
C-RCT
x
x
179
BMI
0.03
0.25
mf
1
Outcome data were
provided by the
author.
12-18
4
8
44
Australia
RCT
x
40
BMI
-1.3
0.23
m
1
>18
12
0
x
37
BMI
-2.5
0.22
m
1
45
Netherlands
C-RCT
x
190
BMI
-0.22
0.13
mf
3
SFT
>18
9
0
192
%BF
-0.79
0.32
mf
3
46
Finland
Q-E
x
557
BMI
-0.2
0.1
mf
com
Combined outcome
estimates for male
12-18
2
0
281
BMI
-0.2
0.16
f
2
140
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
276
BMI
-0.2
0.14
m
2
and female
participants were
used in the meta-
analysis.
x
x
568
BMI
-0.05
0.1
mf
com
281
BMI
0
0.14
f
2
287
BMI
-0.1
0.14
m
2
47
USA
C-RCT
x
77
BMI
0.96
0.44
f
com
Combined outcome
estimates of both
study groups were
used in the meta-
analysis.
12-18
3
0
48
USA
C-RCT
x
x
192
BMI
-0.45
0.14
mf
3
Outcome data were
taken from
Robinson 1999.
6-11
7
0
49
USA
RCT
x
x
52
BMI
-0.32
0.23
f
3
6-11
3
0
50
USA
Q-E
x
109
BMI
-0.09
0.19
mf
1
6-11
3
9
51
USA
C-RCT
x
x
73
BMI
-1.2
0.1
mf
1
BIA
12-18
3
0
%BF
-2.9
0.16
mf
1
53
Australia
Q-E
x
x
x
x
1807
BMI
-0.28
0.21
mf
3
6-11
36
0
57
France
C-RCT
x
x
954
BMI
-0.2
0.23
mf
com
BIA
Combined outcome
estimates of both
study groups were
used in the meta-
analysis.
6-11
48
0
%BF
0.18
0.92
mf
com
52
USA
RCT
x
61
BMI
-0.86
0.13
mf
com
SFT
Combined outcome
estimates for male
and female
participants were
used in the meta-
analysis.
12-18
2
0
45
BMI
-0.95
0.13
f
2
16
BMI
-0.08
0.17
m
2
61
%BF
-0.77
0.44
mf
com
45
%BF
-1.7
0.49
f
2
16
%BF
1.34
0.67
m
2
54
USA
RCT
x
56
%BF
-1.63
0.58
f
3
DXA
>18
4
6
55
USA
Q-E
x
96
%BF
-1.69
0.38
mf
com
SFT
Combined outcome
estimates for male
6-11
3
0
40
%BF
-2.7
0.65
f
1
141
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
56
%BF
-1
0.4
m
1
and female
participants were
used in the meta-
analysis.
56
USA
RCT
x
509
BMI
-0.91
0.17
f
2
DXA
>18
60
0
509
%BF
-1.6
0.36
f
2
58
Netherlands
C-RCT
x
x
x
894
BMI
-0.04
0.05
mf
com
Outcome data were
taken from Singh
2007.
12-18
8
0
449
BMI
-0.05
0.07
f
3
445
BMI
-0.02
0.07
m
3
59
USA
C-RCT
x
x
1013
BMI
-0.36
0.06
mf
2
12-18
6
0
60
USA
C-RCT
x
x
459
%BF
0.08
0.39
f
1
BIA
12-18
1
3
61
Canada
Q-E
x
x
199
BMI
-0.3
0.13
mf
2
Outcome data were
only reported for
one study group
(4th grade to 7th
grade)
6-11
9
0
62
USA
RCT
x
x
53
BMI
0.2
0.2
f
3
Outcome data were
taken from Story
2003a.
6-11
3
0
63
Israel
Q-E
x
371
BMI
-0.27
0.1
mf
com
Combined outcome
estimates of all
study groups were
used in the meta-
analysis.
6-11
26
0
153
BMI
-0.36
0.15
f
com
218
BMI
-0.2
0.14
m
com
64
New
Zealand
Q-E
x
x
282
BMI
-0.6
0.12
mf
1
6-11
24
0
65
USA
C-RCT
x
x
1419
%BF
0.18
0.98
mf
3
BIA
6-11
7
0
66
Australia
C-RCT
x
303
BMI
-0.23
0.08
mf
com
SFT
Combined outcome
estimates of
participants in all
educational
interventions were
used in the meta-
6-11
9
0
144
BMI
-0.4
0.11
f
1
159
BMI
-0.1
0.11
m
1
303
%BF
-0.22
0.22
mf
com
144
%BF
-0.6
0.28
f
1
159
%BF
0.1
0.35
m
1
142
Study
ID
Country
Design*
PA
EDU
NUTR
TV
Sample
size
Outcome
measure*
Outcome
estimate
Standard
Error
Gender*
Type of
outcome
data**
Method
used to
assess
%BF*
Data used in the
meta-analysis
Age
group
Length of
intervention
in months
Months of
follow-up
after
intervention
ending
x
651
BMI
0.03
0.09
mf
com
analysis. Outcome
estimates were also
combined for male
and female
participants.
327
BMI
-0.15
0.15
f
com
324
BMI
0.17
0.09
m
com
651
%BF
0.13
0.183673469
f
com
x
x
307
BMI
0.32
0.09
mf
com
148
BMI
0.2
0.13
f
1
159
BMI
0.4
0.12
m
1
307
%BF
-0.37
0.19
mf
com
67
Netherlands
RCT
x
76
BMI
-0.19
0.53
mf
2
>18
6
0
68
USA
C-RCT
x
447
BMI
-0.16
0.12
mf
3
DXA
Outcome data were
taken from Yin
2005.
6-11
8
12
447
%BF
-0.76
0.34
mf
3
*ABBREVIATIONS: BMI: Body Mass Index; BIA: Bioelectrical impedance analysis; C: Cluster-randomised controlled trial; com: Data were combined for groups; DXA: Dual-energy X-Ray
Absorptiometry; f: Data for females; HW: Hydrostatic weighing m: Data for males; mf: Data for males and females; Q-E: Quasi-experimental; RCT: Randomised controlled trial; SFT: Skinfold thickness;
%BF: Percentage of body fat.
143
b) The studies were checked against the following items:
Q7 – Q14 refer to the quality appraisal
Question number
Q1
Was the control group active?
Q2
Was the intervention school- based?
Q3
Was the family involved in the intervention?
Q4
Was additional cluster adjustment required?
Q5
Did the study report the proportion of overweight and obese participants?
Q6
Was the study population described as being disadvantaged or at risk?
Q7
Were inclusion/exclusion criteria clearly defined?
Q8
Was the intervention clearly described?
Q9
Was the power calculated a priori?
Q10
Was there blinding of the outcome assessment?
Q11
Was the method of randomization described (where applicable)?
Q12
Were the baseline characteristics similar between intervention and control group?
Q13
Was the attrition rate described?
Q14
Was the statistical analysis reproducibly described?
144
c) Results of the quality appraisal (see appendix 4 for Study ID references)
Study ID
Q1
Q2
Q3
Q4
Q5
Q6
Q7
Q8
Q9
Q10
Q11
Q12
Q13
Q14
1
n
y
n
n
n
n
n
y
p
nr
n
n
y
y
2
y
n
y
n
n
y
y
y
p
nr
y
n
y
y
3
n
y
n
y
n
y
n
y
nr
nr
n
y
y
4
y
n
y
n
n
y
y
y
p
nr
n
(y)
y
y
5
n
y
n
n
y
y
n
y
y
nr
n
nr
y
y
7
n
y
n
n
y
n
y
y
nr
nr
y
y
y
y
6
nr
y
y
n
y
y
y
y
nr
nr
n
y
y
y
8
y
y
n
y
y
n
y
y
nr
nr
y
y
y
y
9
n
n
n
n
n
n
y
y
y
nr
n
y
y
y
10
y
y
n
y
y
n
n
y
y
nr
y
n
y
11
n
n
n
n
y
n
y
y
y
nr
y
y
y
y
13
n
n
n
n
n
y
y
y
p
nr
n
y
y
y
14
y
y
n
n
n
n
n
y
nr
y
n
y
y
y
15
n
y
n
y
n
n
n
y
nr
nr
n
y
y
y
12
n
n
y
n
y
n
y
y
p
nr
y
y
y
y
16
y
y
n
n
n
y
y
y
nr
y
n
y
y
y
17
y
y
y
n
y
y
n
y
y
nr
n
n
y
y
18
y
y
n
y
n
y
n
y
nr
nr
n
nr
n
y
19
n
y
y
n
y
y
y
y
nr
n
n
(y)
y
y
20
nr
y
y
y
y
n
n
y
n
nr
n
n
y
y
21
nr
y
n
n
y
y
y
y
y
nr
n
nr
y
y
22
n
y
n
n
n
y
y
y
nr
nr
n
n
y
y
24
y
y
n
n
n
y
y
y
nr
nr
n
y
n
y
23
y
y
n
n
y
y
y
y
nr
nr
n
(y)
n
y
25
n
y
y
n
y
y
n
y
y
nr
y
y
y
26
n
n
n
n
y
n
y
y
y
nr
n
y
y
y
27
y
n
n
n
y
y
y
y
nr
y
n
y
y
n
28
n
n
n
y
y
n
n
y
p
nr
n
y
y
y
29
nr
y
n
n
y
n
y
y
y
nr
y
y
y
y
30
n
y
n
y
n
y
y
y
nr
nr
(y)
y
y
31
n
y
y
n
y
n
y
y
nr
nr
n
y
y
145
32
nr
y
n
y
n
n
n
y
nr
nr
n
n
y
y
33
n
y
n
n
y
n
n
y
p
y
n
(y)
y
y
34
n
y
n
y
y
n
y
y
nr
nr
y
y
y
35
n
n
n
n
y
y
y
y
nr
nr
n
y
y
y
36
y
n
n
n
n
n
y
y
y
nr
n
(y)
y
y
37
n
y
n
y
n
n
n
y
nr
nr
y
y
y
38
n
y
n
n
y
n
y
y
y
n
n
n
y
y
39
n
n
n
n
y
y
y
y
nr
nr
n
(y)
y
y
40
nr
y
y
n
n
n
y
y
y
y
n
y
y
y
41
y
y
n
y
n
y
y
y
nr
nr
n
n
n
y
42
nr
y
y
n
n
y
y
y
nr
nr
n
y
y
y
43
y
y
y
n
n
y
y
y
nr
nr
n
nr
y
y
44
n
n
n
n
n
n
y
y
nr
nr
n
y
y
y
45
n
n
n
n
n
n
y
y
y
y
n
y
y
y
46
n
y
y
y
y
n
n
y
nr
nr
n
y
y
47
y
y
n
y
n
y
y
y
n
nr
y
nr
y
n
48
n
y
y
n
n
n
y
y
y
y
n
y
y
y
49
y
n
n
n
n
y
y
y
p
y
n
y
y
y
50
n
y
n
y
n
n
y
y
nr
nr
nr
y
y
51
nr
y
n
y
n
y
n
y
nr
nr
n
y
y
y
53
n
y
y
n
y
n
y
y
nr
n
n
y
y
57
n
y
y
n
y
n
y
y
y
y
n
(y)
y
y
52
y
y
n
n
n
n
y
y
nr
nr
nr
y
y
54
n
n
n
n
n
n
y
y
nr
nr
n
(y)
y
y
55
n
y
n
n
n
n
n
y
nr
nr
y
y
n
56
n
n
n
n
y
y
y
y
y
y
y
y
y
y
58
n
y
y
n
y
y
y
y
y
n
y
n
y
y
59
n
y
y
y
y
n
n
y
nr
nr
n
nr
y
y
60
nr
y
n
n
y
y
y
y
nr
nr
n
y
y
y
61
n
y
n
n
n
n
n
y
n
y
y
n
y
62
y
y
y
n
y
y
y
y
p
nr
n
n
y
y
63
nr
y
n
y
n
n
y
y
nr
nr
y
y
n
64
n
y
n
n
y
n
y
y
y
nr
n
y
y
65
nr
y
y
n
n
y
y
y
nr
nr
y
(y)
y
y
146
n: no; nr: not reported; p: pilot study; y: yes; (y): baseline characteristics were similar in anthropometric measures
66
n
y
n
n
y
n
n
y
y
n
n
nr
y
y
67
n
n
n
n
n
n
y
y
nr
nr
n
nr
y
y
68
nr
y
n
n
n
y
y
y
y
nr
n
n
y
y
147
Appendix 6
Forest plot of seven intervention groups with low or moderate levels of heterogeneity
Figure 1: Forest plot of the intervention group ‘physical activity, education and nutrition’ for children (0 - 18
years) measured in BMI
Figure 2: Forest plot of the intervention group ‘reducing television viewing’ for children (0 - 18 years) measured in
BMI
Study or Subgroup
Caballero 2003
Donnelly 1996
Haerens 2006
Nader 1999
Sanigorski 2008
Singh 2007
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 5.29, df = 5 (P = 0.38); I² = 5%
Test for overall effect: Z = 3.35 (P = 0.0008)
Mean Difference
-0.2
-0.2
-0.15
0
-0.28
-0.04
SE
0.17
0.2
0.04
0.1
0.21
0.05
Weight
3.3%
2.4%
49.0%
9.4%
2.2%
33.7%
100.0%
IV, Random, 95% CI
-0.20 [-0.53, 0.13]
-0.20 [-0.59, 0.19]
-0.15 [-0.23, -0.07]
0.00 [-0.20, 0.20]
-0.28 [-0.69, 0.13]
-0.04 [-0.14, 0.06]
-0.10 [-0.17, -0.04]
Mean Difference Mean Difference
IV, Random, 95% CI
-0.5 -0.25 0 0.25 0.5
Favours experimental Favours control
Study or Subgroup
Chavarro 2005
Dennison 2004
Fitzgibbon 2005
Harrison 2006
Kipping 2008
Robinson 1999
Robinson 2003
Sanigorski 2008
Total (95% CI)
Heterogeneity: Tau² = 0.01; Chi² = 8.74, df = 7 (P = 0.27); I² = 20%
Test for overall effect: Z = 3.90 (P < 0.0001)
Mean Difference
-0.3
-0.36
-0.54
-0.08
0.1
-0.45
-0.32
-0.28
SE
0.1
0.44
0.22
0.15
0.19
0.14
0.23
0.21
Weight
27.4%
2.4%
8.5%
15.9%
11.0%
17.6%
7.9%
9.3%
100.0%
IV, Random, 95% CI
-0.30 [-0.50, -0.10]
-0.36 [-1.22, 0.50]
-0.54 [-0.97, -0.11]
-0.08 [-0.37, 0.21]
0.10 [-0.27, 0.47]
-0.45 [-0.72, -0.18]
-0.32 [-0.77, 0.13]
-0.28 [-0.69, 0.13]
-0.27 [-0.40, -0.13]
Mean Difference Mean Difference
IV, Random, 95% CI
-1 -0.5 00.5 1
Favours experimental Favours control
148
Figure 3: Forest plot of the intervention group ‘education and nutrition’ for children (0 - 18 years) measured in
BMI
Figure 4: Forest plot of the intervention group ‘education’ for children (0–18 years) measured in %BF
Figure 5: Forest plot of the intervention group ‘physical activity’ for adults (>18 years) measured in BMI
Study or Subgroup
Foster
Puska 1982
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 0.00, df = 1 (P = 0.95); I² = 0%
Test for overall effect: Z = 0.60 (P = 0.55)
Mean Difference
-0.04
-0.05
SE
0.12
0.1
Weight
41.0%
59.0%
100.0%
IV, Random, 95% CI
-0.04 [-0.28, 0.20]
-0.05 [-0.25, 0.15]
-0.05 [-0.20, 0.10]
Mean Difference Mean Difference
IV, Random, 95% CI
-0.5 -0.25 0 0.25 0.5
Favours experimental Favours control
Study or Subgroup
Spruijt-Metz 2008
Vandongen 1995
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 0.02, df = 1 (P = 0.90); I² = 0%
Test for overall effect: Z = 1.45 (P = 0.15)
Mean Difference
0.08
0.13
SE
0.39
0.09
Weight
5.1%
94.9%
100.0%
IV, Random, 95% CI
0.08 [-0.68, 0.84]
0.13 [-0.05, 0.31]
0.13 [-0.04, 0.30]
Mean Difference Mean Difference
IV, Random, 95% CI
-1 -0.5 0 0.5 1
Favours experimental Favours control
Study or Subgroup
Donnelly 2003
Pritchard 2007
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 0.27, df = 1 (P = 0.60); I² = 0%
Test for overall effect: Z = 6.29 (P < 0.00001)
Mean Difference
-1.07
-1.3
SE
0.38
0.23
Weight
26.8%
73.2%
100.0%
IV, Random, 95% CI
-1.07 [-1.81, -0.33]
-1.30 [-1.75, -0.85]
-1.24 [-1.62, -0.85]
Mean Difference Mean Difference
IV, Random, 95% CI
-2 -1 012
Favours experimental Favours control
149
Figure 6: Forest plot of the intervention group ‘education’ for adults (>18 years) measured in %BF
Figure 7: Forest plot of the intervention group ‘physical activity’ for adults (>18 years) measured in %BF
Study or Subgroup
Eiben 2006
Leermakers 1998
Proper 2003
Simkin-Silvermann 2003
Total (95% CI)
Heterogeneity: Tau² = 0.18; Chi² = 4.88, df = 3 (P = 0.18); I² = 38%
Test for overall effect: Z = 3.42 (P = 0.0006)
Mean Difference
-3.9
-0.8
-0.79
-1.6
SE
1.98
0.99
0.32
0.36
Weight
3.1%
10.9%
45.0%
41.0%
100.0%
IV, Random, 95% CI
-3.90 [-7.78, -0.02]
-0.80 [-2.74, 1.14]
-0.79 [-1.42, -0.16]
-1.60 [-2.31, -0.89]
-1.22 [-1.92, -0.52]
Mean Difference Mean Difference
IV, Random, 95% CI
-4 -2 0 2 4
Favours experimental Favours control
Study or Subgroup
Donnelly 2003
Schmitz 2003
Total (95% CI)
Heterogeneity: Tau² = 0.00; Chi² = 0.55, df = 1 (P = 0.46); I² = 0%
Test for overall effect: Z = 5.63 (P < 0.00001)
Mean Difference
-2.17
-1.63
SE
0.44
0.58
Weight
63.5%
36.5%
100.0%
IV, Random, 95% CI
-2.17 [-3.03, -1.31]
-1.63 [-2.77, -0.49]
-1.97 [-2.66, -1.29]
Mean Difference Mean Difference
IV, Random, 95% CI
-4 -2 0 2 4
Favours experimental Favours control
150
Appendix 7
Search strategies in chapter 5
PubMed
(("Physical activity” OR "Physical education” OR "exercise”[MeSH] OR exercise[All Fields]
OR exercising[All Fields]) OR "nutritional“ OR "nutrition“ OR KEY("diet”) OR "healthy
eating"[All Fields]) AND ("schools"[MeSH Major Topic] OR school*[All Fields] OR
"School Health Services"[MeSH]) AND (policy[All Fields] OR policies[All Fields] OR
"nutrition policy"[MeSH] OR "health promoting school"[All Fields] OR "school health
promotion"[All Fields] OR "healthy school"[All Fields] OR ((holistic[All Fields] OR
whole[All Fields]) AND "school approach"[All Fields]) OR "wellness"[All Fields]) AND
("weight"[All Fields] OR "body composition"[All fields] OR ("obesity"[MeSH Terms] OR
"obesity"[All Fields]) OR ("overweight"[MeSH Terms] OR "overweight"[All Fields]))
Scopus
(TITLE-ABS-KEY("weight") OR TITLE-ABS-KEY("body composition") OR TITLE-ABS-
KEY("obesity") OR TITLE-ABS-KEY("overweight"))
AND
((TITLE-ABS-KEY(policy) OR TITLE-ABS-KEY(policies) OR KEY("nutrition policy")) OR
ALL("health promoting school") OR ALL("school health promotion") OR ALL("healthy
school") OR ALL((holistic OR whole) AND "school approach") OR TITLE-ABS-
KEY("wellness"))
AND
(TITLE-ABS-KEY(school*) OR KEY("school health services")) AND (TITLE-ABS-
KEY("physical activity") OR TITLE-ABS-KEY("physical education") OR (TITLE-ABS-
KEY("exercise")) OR (TITLE-ABS-KEY("nutrition*")) OR KEY("diet") OR ALL("healthy
eating"))
EMBASE
'physical activity, capacity and performance'/exp OR 'physical education'/exp OR
'nutrition'/syn OR 'child nutrition'/exp OR 'meal'/exp
AND
'school'/exp
151
AND
'healthcare policy'/exp OR 'policy'/exp OR 'school policy' OR 'school-based policy' OR 'health
promoting school' OR 'school health promotion' OR 'healthy school' OR (holistic OR whole
AND 'school approach') OR 'wellness'/syn
CINAHL
(MH "Nutrition+") OR "nutrition" OR (MH "Nutrition Services+") OR (MH "Nutritional
Support+/ED/EV/LJ") OR (MM "Physical Education, Adapted/ED/EV") OR (MM "Physical
Activity/EV/LJ/ED")
AND
(MH "Schools+")
AND (MH "Organizational Policies") OR (MH "School Policies") OR (MH "Nutrition
Policy/ED/EV/LJ/OG/PC") OR "policy" OR “health promoting school” OR “school health
promotion” OR “healthy school” OR (holistic OR whole AND “school approach”) OR
“wellness”
Cochrane
Search
#1
MeSH descriptor Schools, this term only
#2
MeSH descriptor Physical Education and Training, this term only
#3
MeSH descriptor Exercise, this term only
#4
MeSH descriptor Child Nutrition Sciences explode all trees
#5
MeSH descriptor Nutrition Policy explode all trees
#6
(#1 AND ( #2 OR #3 OR #4 OR "nutrition" OR ( physical AND activity ) ) AND ( #5 OR "policy" OR
"health promoting school" OR "school health promotion" OR "healthy school" OR ( ( holistic OR whole )
AND "school approach" ) OR "wellness" ))
152
Appendix 8
Information on the cost calculation in chapter 7
a) PHYSICAL ACTIVITY COORDINATORS
Main source of reference: APPLE study, New Zealand (McAuley et al. 2010)
Description: The coordinator spends one hour each weekday to encourage children to be active during breaks and after school. A monthly blanket
amount covers ongoing expenses for materials and equipment needed.
Resource use in ‘Physical activity coordinators’
Costs component
Assumption for Transfer
Cost ingredients (quantities per
school or as indicated)
Value employed in
this study
Original value
Original source
Employment of a
physical activity
instructor
The instructors were employed
on a half-time base in the APPLE
study, however no exact number
of hours was reported other than
that the programme was
delivered for one hour per school
day. Therefore, it was assumed
that the number of paid hours
equalled the hours of sessions
held plus 30 min. preparation for
each session.
Frequency of sessions
5 days/week
5 days/week
McAuley et al. 2010
Session length
60 minutes
60 minutes
McAuley et al. 2010
Preparation time per session
30 minutes
-
Estimate
Coordinators needed per school
1
2
McAuley et al. 2010
Wage rate instructor
AU $44
-
Pay rate for school services officer level 4 First year,
South Australia School Services Officers (Government
Schools) Award URL:
www.industrialcourt.sa.gov.au/index.cfm?objectid=A3
B69692-E7F2-2F96-369614F673DDD81D (last
accessed 05 April 2012) , includes 60% on-costs and
office use
Programme
supplies
The blanket amount for
equipment provided in the
APPLE study was taken and
adjusted for the year and
currency.
Number of equipment sets per year
1
1
McAuley et al. 2010
Equipment, vouchers and rewards for
schools
AU $822
Amount in NZD
(2006)
McAuley et al. 2010
153
b) EVENTS
Main source of reference: Shape up Summerville, USA (Economos et al. 2007); Hartslag Limburg, Netherlands (Ronckers et al. 2006)
Description: Regular one-day events are held on the school premises that promote physical activity including competitions. Recruited volunteers who
received police checks and a liaising teacher are responsible for operating the event. Volunteer recruitment and training continues at 25% intensity
compared to the set-up phase.
Resource use in ‘Events’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed in
this study
Original value
Original source
Volunteer time in
recruiting
The only information available for this
component was the number of events per
year and the blanket amount to cover
expenses. Therefore, the number of
volunteers and the time of the responsible
teacher were assumed including recruiting
and police checks. The blanket amount was
adjusted for year and currency.
Recruiting time
150 minutes
Estimate
Volunteers recruited
20
Estimate
Volunteer travel
time for
recruiting
Number of return trips of volunteer
20
Estimate
Police checks
Police checks
20
Estimate
Teacher time in
event
Time per event for teachers
480 minutes
Estimate
Volunteer time in
event
Sport events per year
10
10
Economos et al. 2007
Volunteers per event
5
Estimate
Volunteer travel
cost for event
Number return trips of volunteers per year
50
Estimate
Event expenses
Budget per event
AU $392
Amount in EUR
(2004)
Ronckers et al. 2006
154
c) ACTIVE TRANSPORT TO SCHOOL
Main source of reference: Walking School Bus, Australia (Victoria) (Moodie et al. 2009a)
Description: Children walk to school in a bus formation of eight students and two adults. Recruited parents act as volunteers who receive police
checks and training. A liaising teacher helps with the coordination. The routes are assessed for safety. Participating children receive a resource kit and
the bus theme is also embedded in the curriculum. Volunteer recruitment and training continues at 25% intensity compared to the set-up phase.
Resource use in ‘Active transport to school’
Costs
component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value
employed in
this study
Original value
Original source
Liaison
teacher
Based on description in original study
and adjusted for local prices.
Liaison school teacher needed per year
1
1
Moodie et al. 2009b
Liaison teacher time load per week
60 minutes
60 minutes
Moodie et al. 2009b
Volunteer
time routine
operation
The frequency of running the bus was
assumed to be higher in South Australia
than in the original study due to weather
conditions. In addition, the number of
busses per school was adjusted
participation observed in the DoT’s
Walking School Bus programme
(Western Australia).
Number of walking busses
2/school
16 children per
term, translates
into 2
busses/school
(>1.6 in WSB
from Moodie et
al. 2009b)
DoT’s Walking School Bus
programme (WA)
URL:
http://www.transport.wa.gov.au/
AT_TS_P_wsb_IndividualSchool
Summaries_2010.pdf (last accessed
06 April 2012)
Bus size
8 children
8
Moodie et al. 2009b
Number of volunteers needed
2/bus
2/bus
Moodie et al. 2009b
Time routine operation
40 minutes
40 minutes
Moodie et al. 2009b
Frequency
2/week
1/week
Moodie et al. 2009b
Proportion of volunteers already walking
0.5
0.5
Moodie et al. 2009b
Volunteer
recruiting
Based on description in original study
Volunteers recruited
15
15
Moodie et al. 2009b
Length of recruitment session
150 minutes
150 minutes
Moodie et al. 2009b
Recruiting
travel time
Based on description in original study
Number of return trips for volunteers
15
15
Moodie et al. 2009b
Police checks
Based on description in original study
Number of volunteers checked
8
8
Moodie et al. 2009b
Volunteer
training
Based on description in original study
Volunteers trained
4/bus
4/bus
Moodie et al. 2009b
Number of volunteer training sessions
1
1
Moodie et al. 2009b
155
Costs
component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value
employed in
this study
Original value
Original source
Length of training session
150 minutes
150 minutes
Moodie et al. 2009b
Volunteer
training
travel time
Based on description in original study
Number of return trips for volunteers
8
8
Moodie et al. 2009b
Volunteer
trainer
Based on description in original study
Number of trainers needed
1/school
1/school
(presumably)
Moodie et al. 2009b
Trainer time
180 minutes
180 minutes
Moodie et al. 2009b
Wage rate volunteer trainer
AU $25
S.A. PUBLIC SECTOR
SALARIED EMPLOYEES
INTERIM AWARD Public Service
officer Operational Stream Level 2
– URL:
http://www.decd.sa.gov.au/docs/doc
uments/1/SaPublicSectorSalariedE
mp.pdf
(last accessed 05 April 2012)
Includes 30% on-costs
Catering
Based on description in original study
Number of catering serves needed
9
9 (presumably)
Moodie et al. 2009b
Venue hire
Based on description in original study
Number of training venues needed
1/school
1/school
(presumably)
Moodie et al. 2009b
Photocopied
materials
Based on description in original study
Training printed material needed
8
8
Moodie et al. 2009b
Price of printing promotion material
AU $1
Commercial
rate
Estimate based on commercial rate
Route
assessment
Based on description in original study
Number of assessors needed
1/route
1/route
Moodie et al. 2009b
Assessor time needed
120 minutes
120 minutes
Moodie et al. 2009b
Wage rate route assessor
AU $25
S.A. PUBLIC SECTOR
SALARIED EMPLOYEES
INTERIM AWARD Public Service
officer Operational Stream Level 2
URL:
http://www.decd.sa.gov.au/docs/doc
uments/1/SaPublicSectorSalariedE
mp.pdf
(last accessed 05 April 2012)
Includes 30% on-costs
Volunteer
Based on description in original study
Number of volunteer assessors needed
2/bus
2/bus
Moodie et al. 2009b
156
Costs
component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value
employed in
this study
Original value
Original source
route
assessment
Time for assessment
60
minutes/bus
60 minutes/bus
Moodie et al. 2009b
Resource kits
for travellers
Based on description in original study
Number of resource kits needed
1/traveller
1/traveller
(presumably)
Moodie et al. 2009b
Price for resources kit
AU $71
Amount in AU $
(2001)
Moodie et al. 2009b
Curriculum
manual
Number of manuals needed was assumed
to be one for each school year. Manual
price was taken from the original study
and adjusted for year.
Number of manuals
7
Estimate
Curriculum manual price
AU $57
Amount in AU $
(2001)
Moodie et al. 2009b
Supplies for
promotion
Based on price in original study and
adjusted for year.
WSB promotion event per year
1
1
Moodie et al. 2009b
Blanket amount for promotion
AU $425
Amount in AU $
(2001)
Moodie et al. 2009b
157
d) TRAINING FOR TEACHERS
Main source of reference: Planet Health, (Wang et al. 2003), CATCH, (Brown and Summerbell 2009)
Description: All schoolteachers receive training to enhance their curriculum with health and life-style themes. Relief teachers compensate for the
absence from class.
Resource use in ‘Training for teachers
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed in
this study
Original value
Original source
Relief teacher
All teachers attend the training in Planet
Health. Therefore, relief teachers are
required to fill in the absence from work.
Number of teachers receiving training
14/school
all teachers in the
intervention school
Average number of
teachers in South
Australian public
primary schools
(ABS 2010c)
Length of training
480 minutes
480 minutes
Brown et al. 2007a
Venue hire
No description in the original study,
therefore venue hire was assumed based
on other programmes (Moodie et al.
2009a,b).
Venue (full day)
1/school
Estimate
Trainer
Transferred from the original study. Time
for preparation was added to the training
session time.
Time in training session
720 minutes
Estimate
Material
Transferred from the original study and
adjusted for year and currency.
Training materials needed
14/school
1/teacher
(presumably)
Wang et al. 2003
Price of training materials
AU $111
Wang et al. 2003
Catering
No description in the original study,
therefore catering was assumed based on
other programmes (Moodie et al.
2009a,b).
Total number for catering
15
Estimate
158
e) PARENT FORUMS
Main source of reference: Shape up Summerville, USA (Economos et al. 2007)
Description: Four times per year a two-hour forum is organised by a liaising teacher. Speakers could be the local community coordinator, a general
practitioner, a dietician or the physical activity coordinator.
Resource use in ‘Parent forums’
Costs component
Assumption for Transfer
Cost ingredients (quantities per
school or as indicated)
Value employed
in this study
Original value
Original source
Time cost to
parents
No detailed information for the
sessions was available. Therefore,
frequency, session length and the
type of speakers were assumed.
Frequency of sessions per year
4
Estimate
Length per session
150
Estimate
Assumed average attendance rate
0.3
Estimate
Number of parents attending per session
63
Assumed that 1/3 of parents and caregivers
would attend the sessions
Travel cost to
parents
The average attendance rate of
parents was assumed to be 30% of
all parents/caregivers (one per
child).
Number of return trips for parents
254
Estimate
Time cost GP
Profession of speakers was
assumed
Time cost to General Practitioner
150
Estimate
Hourly rate General Practitioner
AU $196
SADI claims policy, URL:
http://www.gpsa.org.au/media/docs/newsletter/g
psa_board
_communique_13_march_2009.pdf (last
accessed 06 April 2012); includes 30% on-costs
Venue hire
Based on description in original
study and adjusted for local prices.
Forum venue required
4/year
Estimate
Liaising teacher
Assume that teacher dedicates one
hour per week for activities in the
education component
Time for component
2400 minutes/year
Estimate
159
f) SOCIAL MARKETING CAMPAIGN
Main source of reference: Be active eat well, Australia (Victoria) (Sanigorski et al. 2008)
Description: The campaign promotes healthy life-style on TV, radio and with posters in public spots. It also includes the promotion of an annual
event to reduce TV-viewing ("switch off" week). Health promotion resources are distributed in schools.
Resource use in ‘Social marketing campaign’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed
in this study
Original value
Original source
Resources for schools
Value approximated from expenses stated in
Be active Eat Well programme.
Price of resource kit for one school
AU $100
Pettman et al. 2010
State-wide campaign
Data were taken from a previous social
marketing campaign in South Australia.
Production of resources (set-up phase)
AU $315,000
Communication with Health
Promotion Branch,
Department of Health,
Government of South
Australia
Media buy (annually)
AU $550,000
Communication with Health
Promotion Branch,
Department of Health,
Government of South
Australia
160
g) FRUIT AND WATER PROGRAM
Main source of reference: APPLE study, New Zealand (McAuley et al. 2010); National fruit programme, Norway (Bere et al. 2010)
Description: Every student receives a free piece of fruit on every school day. The fruit theme is also incorporated into the curriculum. A liaising
teacher coordinates the scheme. In addition, water dispensers are installed in classrooms.
Resource use in ‘fruit and water program’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed
in this study
Original value
Original source
Water dispenser
instalment
A budget per school was assumed to cover
the costs of the dispensers as well as their
instalment based on Be Active Eat Well
experience.
Budget for water supply upgrade
AU $800
c. AU $ 800
Pettman et al. 2010
Fruit supply
It was assumed that the fruits are provided
on a daily base throughout the school year.
Fruit price was adopted from an Australian
scheme in Victoria and adjusted for year.
Number of fruit needed
5/week
5/week
Bere et al. 2010
Fruit price
AU $0.77
Amount in AU
$ (2007)
Free Fruit Friday Victoria, URL:
http://www.education.vic.gov.au
/studentlearning/programmes/
freefruitfriday/resources.htm (last
accessed 06 April 2012)
Liaising teacher
Assume that liaising teacher dedicates one
hour per week for activities in the nutrition
component
Teacher time for component
2400 minutes
Estimate
Materials
Based on description in original study and
adjusted for year and currency.
Price of teaching material
AU $34
Amount in
NZD (2006)
McAuley et al. 2010
161
h) CANTEEN FOOD IMPROVEMENT AND DIETITIAN SUPPORT:
Main source of reference: CATCH, USA (Brown et al. 2007a), Be Active Eat Well, Australia (Victoria) (Sanigorski et al. 2008)
Description: Canteen staff members and volunteering parents receive training (on site). A dietician dedicates one hour per school week to consult
with canteen staff and with students in class
Resource use in ‘canteen food improvement and dietitian support’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value
employed in
this study
Original value
Original source
Training time staff
member
The number of staff members and
volunteers receiving training was
adjusted to the South Australian context,
where canteens commonly operated by
volunteering parents.
Number of canteen staff trained
1
Estimate
Length of training
2880 minutes
2880 minutes
Brown et al. 2007a
Wage rate canteen staff
AU $22
National canteen award
scheme (level 3) URL:
http://www.fwa.gov.au/
documents/modern_awards/
award/ma000076/default.htm
(last accessed 05 April 2012)
Includes 30% on-costs
Training time
volunteers
Number of volunteers receiving training
15
Estimate
Training time for volunteers
3060 minutes
3060 minutes
Brown et al. 2007a
Trainer costs
Additional preparation time for the
trainer was assumed.
Trainer time needed (including preparation time)
4320
2880 minutes
Brown et al. 2007a
Catering during
training
Additional catering expenses were
assumed to be included (Moodie et al.
2009a,b).
Training attendees for catering
102
Estimate
Ongoing support by
a school dedicated
dietician
Assumed to be an existing position.
Hence, no additional load for office space
was added to the wage rate.
Dietician per school
1
1
Sanigorski et al. 2008
Dietician time per week
60 minutes
60 minutes
Sanigorski et al. 2008
162
i) EDUCATION TO REDUCE TV-VIEWING
Main source of reference: (Harrison et al. 2006, Robinson 1999)
Description: A designated number of teachers receive additional training to incorporate reduction of TV-viewing as a theme into the curriculum.
Resource use in ‘education to reduce TV-viewing’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed in
this study
Original value
Original source
Relief teacher
Proportion of teachers receiving training
was assumed based on the number
Harrison et al. 2006
Time additional teacher training
180 minutes
180 minutes
Harrison et al.
2006
Proportion of teachers receiving TV training
20%
Estimate
Training venue
No description in original study,
therefore venue hire was assumed based
on other programmes (Moodie et al.
2009a,b).
Venue needed
1
Estimate
Trainer
Additional preparation time for the
trainer was assumed.
Trainer time
270 minutes
Harrison et al.
2006
Material
Materials needed were assumed based on
Planet Health (Wang et al. 2003)
No. Teachers receiving training
3
Estimate
Price of training materials
AU $111
Amount in US $
(1996)
Wang et al. 2003
Catering
No description in original study,
therefore venue hire was assumed based
on other programmes (Moodie et al.
2009a,b).
Number of catering serves
4
Estimate
163
j) ROUTINE MONITORING
Main source of reference: Australia's Active After-School Communities programme, Australia (Victoria) (Moodie et al. 2009b)
Description: Liaising teachers conduct routine evaluation of activities every school term.
Resource use in ‘routine monitoring’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value employed in this
study
Original
value
Original source
Time dedicated by
teachers
Time to evaluate activities per
term was assumed to be based
on the original source, but was
multiplied by the number of
components in this study.
Time for evaluation needed per component
30
minutes/term/component
30
minutes/ter
m
Moodie et al. 2009b
k) STEERING COMMITTEE
Main source of reference: Australia's Active After-School Communities programme, Australia (Victoria) (Moodie et al. 2009b)
Description: Four staff members sit on the steering committee, which is located at the State Department of Health. They manage the state-wide
implementation and are responsible for the social marketing campaign.
Resource use for ‘steering committee’
Costs component
Assumption for Transfer
Cost ingredients
Value employed in this
study
Original value
Original source
Central coordinator
Assumed the number of staff
members in the committee
based on other programmes
(Moodie et al. 2009b).
Number of committee members
4 (full-time equivalent)
2.25/state
Moodie 2009b
Salary project managers
AU $118,334
SA GOVERNMENT WAGES
PARITY (SALARIED)
ENTERPRISE AGREEMENT
2010, ASO 5 URL:
http://www.health.sa.gov.au/Portals
/0/sagovtwagesparityenterpriseagr
eement-100121.pdf (last accessed
06 April 2012), Includes 60% on-
costs and office use
164
l) LOCAL COORDINATION AND COMMUNITY ENGAGEMENT
Main source of reference: Australia's Active After-School Communities programme, Australia (Victoria) (Moodie et al. 2009b), Kids Go For Your
Life, Australia (Victoria) (de Silva-Sanigorski et al. 2010), APPLE study, (McAuley et al. 2010)
Description: In each local government council, a coordinator is employed (full-time) who has the task to initiate, coordinate and oversee the activities
in the community. This includes meetings with programme stakeholders, support for community capacity building and contributions to the parent
forums. Throughout the intervention, meeting with community stakeholders are continued, but at 25% intensity compared to the set-up phase.
Resource use in ‘local coordination and community engagement’
Costs component
Assumption for Transfer
Cost ingredients (quantities per school or as
indicated)
Value
employed in
this study
Original
value
Original source
Community
coordinator
Number of community coordinators
based on experience from Eat Well Be
Active in South Australia (Pettman et
al. 2010).
Coordinator per Community
1
1
Pettman et al. 2010
Annual salary local community coordinator
AU $86,299
SOCIAL AND COMMUNITY
SERVICES AWARD, level 6
community services, URL:
http://www.fwa.gov.au/consolidated_
awards/an/AN150140/asframe.html
(last accessed 06 April 2012),
includes 60% on-costs and office use
Time cost to
professionals
Average length of consultation with
community members during start-up
phase had to be assumed. Number of
participants based on description in Eat
Well Be Active (Pettman et al. 2010).
Estimate includes travel time all for
community members.
Time per meeting
120
Estimate
No of local professionals in meetings
70
70
Pettman et al. 2010
Time costs to
community
members
Number of community members in meetings
44
44
Pettman et al. 2010
Venue hire
Assumed that a venue would need to be
hired.
Average number of consultations
18
18
Pettman et al. 2010
Average number of venues needed
18
Estimate
165
m) Local government information sources
n) School information sources
School Info
Value used
Explanatory notes
Source
Number of public
primary schools in
South Australia
489
Excludes special
schools, language
centre, open access
schools
DECD School Statistics (2010) URL:
http://www.decd.sa.gov.au/docs/documents/1/AnnualRep
ort2010Stats.pdf, (last accessed 06 April 2012)
Average size of
primary school in
SA
212
REPORT ON GOVERNMENT SERVICES 2011 -
School Education [Table 4A.22 Full time student
enrolments and schools (number) (a)]
(Productivity Commission 2011)
Proportion of
schools with a
school canteen
90%
Oral communication with a school nutrition researcher in
South Australia
School weeks per
year in Australia
40
http://www.liveinvictoria.vic.gov.au/living-in-
victoria/education-and-childcare/primary-schools (last
accessed 05 April 2012)
Student teacher
ratio in public
primary schools
15.3
ABS 2010c Schools Australia
Number of public
primary school
students
103,506
ABS 2010c Schools Australia
Estimated student
per year number
12,500
Divided 100,000 by
8 years, assumption
that distribution
between years is
even
ABS 2010c Schools Australia [Table 9], DECD School
statistics (2010) DECD School Statistics URL:
http://www.decd.sa.gov.au/docs/documents/1/AnnualRep
ort2010Stats.pdf (last accessed 05 April 2012)
o) Common salaries and wage rates
Value used
Explanatory notes
Source
Teacher
AU $237
Daily earnings 2010
(Tier 4)
DECD Wages and Salaries 2010, URL:
http://www.decd.sa.gov.au/hrstaff/pages/employmentcond
itions/wages/ (last accessed 05 April 2012)
Relief Teacher
AU $269
Daily earning 25%
(Tier 4)
DECD Wages and Salaries 2010, URL:
http://www.decd.sa.gov.au/hrstaff/pages/employmentcond
itions/wages/ (last accessed 05 April 2012)
Trainer for
teachers
AU $50
Lecturing Rate &
Higher marking
rate-oncost load
included
RATES OF PAYMENT FOR CASUAL ACADEMIC
STAFF 2010, University of Adelaide URL:
http://www.adelaide.edu.au/hr/conditions/salary/cas_acad
_rates030710.pdf (last accessed 05 April 2012) includes
30% on-costs
National average
weekly earning
AU $1,257
August 2010, all
employees total
earning seasonal
adjustment
ABS 2010a
Valuation of
volunteer time
25%
Moodie et al. 2009a
% of volunteers
sacrificing work
10%
Moodie et al. 2009a
Value of volunteer
hour
AU $11
Calculation based on national average weekly earnings,
valuation of time for volunteering, % of volunteers
sacrificing work time
Value used
Explanatory notes
Source
Number of
communities/LGA
68
http://www.lga.sa.gov.au/site/page.cfm?u=208
166
Value used
Explanatory notes
Source
Dietician
AU $41
Public Service
officer Professional
Stream Level 4
S.A. Public Sector Salaried Employees Interim Award,
Professional Stream Level 4 URL:
http://www.decd.sa.gov.au/docs/documents
/1/SaPublicSectorSalariedEmp.pdf (last accessed 05 April
2012), includes 30% on-costs
p) Common unit costs
Value used
Explanatory notes
Source
Cost police check
AU $33
South Australian Police Department
http://www.police.sa.gov.au/sapol/services/information_re
quests.jsp (last accessed 05 April 2012)
Catering (small)
AU $15
Commercial rate
Catering (large)
AU $30
Commercial rate
Average distance
travelled to school
1.7 km
Estimate taken from
an active travel
assessment study in
South Australia
Harten et al. 2004
Travel cost per km
AU $0.74
Assumed to equal
rate for tax
deduction (medium
car size)
Australian Taxation Office, URL:
http://www.ato.gov.au/individuals/content.asp?doc=/content/338
74.htm&pc=001/002/013/008/004&mnu=&mfp=&st=&cy=1
(last accessed 05 April 2012)
Average travel
time
30 min
2x15 min
Moodie et al. 2009a
Venue hire half
day
AU $220
Commercial rate
Venue hire full day
AU $340
Commercial rate
167
Appendix 9
Additional information on scenarios adjustments
a) Geographic variation
Value used
Source
Average school
size adjusted for
geographic
variation
205
Adjustment o school size according to GEOGRAPHIC LOCATION
DATABASE (MCEECDYA 2009)
Number of
metropolitan
schools
324
Public primary schools in inner regional and metropolitan areas
(MCEECDYA 2009)
Number of
remote schools
165
Public primary schools in outer regional remote and very remote areas
(MCEECDYA 2009)
Average
kilometres
travelled to
school in remote
areas
14.14 km
Average distance to school in remote South Australia (CGC 2008)
Material
surcharge in
remote areas
10%
Indicative estimate based on average price difference between Adelaide and
Roxby Downs URL:
http://www.bhpbilliton.com/home/aboutus/regulatory/Documents/odxEisAp
pendixQSocialEnvironmentAndTrafficContents.pdf, last accessed 06 April
2012
Fruit surcharge
in remote areas
20%
Indicative estimate based on average price difference in remote areas in
Queensland compared to urban areas URL:
http://www.mja.com.au/public/issues/186_01_010107/har10516_fm.html,
last accessed 05 April 2012
b) Socio-economic variation
Value used
Source
Number of
schools with
socio-economic
disadvantage
168
Approximated thruogh the number of public primary schools participating in
the partnership for Low Socio-economic Status School Communities
(South Australia Smarter Schools National Partnerships 2009)
Proportion of
students with
socio-economic
disadvantage in
South Australia
22.8%
Approximated based on number of children under 16 years in low-income
families in in South Australia(PHIDU 2008)