Obesity in both adult and children is fast becoming one of the most serious public health problems of the 21st century in developed and developing countries alike. It is estimated that approximately 10% of school age children. The prevalence of childhood overweight and obesity is ever on the increase in the UK as in the rest of the world. It is estimated that the prevalence of overweight and obesity among 2 - 10 year old children in the UK rose from 22.7%-27.7% and 9.9%-13.7% respectively between 1995 and 2003; these figures are set to increase unless something is done. School-based interventions offer a possible solution in halting obesity prevalence, because the school setting provides an avenue for reaching out to a high percentage of children (especially in the western world), opportunity for constant monitoring of children and the resources for anti-obesity interventions.
To systematically review the evidence of the impact of school-based interventions to prevent childhood obesity on:
The review was done following the Cochrane collaboration guidelines.
In addition to searching electronic databases, first authors of all included studies were contacted. A recognised critical appraisal tool was used to assess the quality of included studies.
Three RCTs and one CCT met the inclusion criteria for the review. All four studies had a control and intervention group; with various study limitations.
While none of the studies found statistically significant BMI changes in intervention groups when compared with control group post-intervention, all of them recorded either a significant change in diet, or an increase in physical activity levels.
Obesity is generally understood as abnormal accumulation of fat to the extent that presents health risk (Kiess, Marcus et al. 2004), and was added to the international classification of diseases for the first time in 1948 (Kipping, Jago et al. 2008). The worldwide clinical definition of adult obesity by the WHO is body mass index (BMI) ≥ 30kg/m2 (WHO 2006). In children however, because of the significant changes in their BMI with age (Cole, Bellizzi et al. 2000), there is no universally accepted definition of obesity (Parizkova and Hills 2004; Bessesen 2008) and it therefore varies from country-to-country. The most commonly used definition of childhood obesity is the US definition which measures overweight and obesity in a reference population using the cut off points of 85th and 95th centiles of BMI for age (Ogden, Yanovski et al. 2007). In the UK, overweight and obesity are diagnosed using a national reference data from a 1990 BMI survey of British children (Stamatakis, Primatesta et al. 2005). Children whose weights are above the 85th centile are classed as overweight and over the 95th centile are considered obese (Reilly, Wilson et al. 2002). Recent estimates suggest that obesity has reached epidemic proportions globally with about 400 million adults being clinically obese, a figure projected to rise to about 700 million by 2015 (WHO 2006). In children, the current WHO estimates are that about 22 million children globally under age 5 are overweight (WHO 2008). In the UK, evidence suggests that obesity is set to be the number one preventable cause of disease in a matter of time (Simon, Everitt et al. 2005).
In the last three decades, the scale as well as the prevalence of obesity have grown rapidly amongst all age, social and ethnic groups in the UK, as well as globally (Table 1)(Kipping, Jago et al. 2008). Estimates suggest that in the UK, between 1984 and 2002/2003, the prevalence of obesity in boys aged 5-10 rose by 4.16%, and by 4.8% in girls (Stamatakis, Primatesta et al. 2005). There is therefore there is an urgent need for the development and implementation of effective intervention strategies to halt the ever increasing obesity prevalence (Summerbell Carolyn, Waters et al. 2005).
The primary risk factors associated with the increase in prevalence of childhood obesity are ever increasing involvement in sedentary lifestyles and an increase also in the consumption of high energy dense food and drink (Ebbeling, Pawlak et al. 2002; Sekine, Yamagami et al. 2002; Speiser, Rudolf et al. 2005; Topp, Jacks et al. 2009). The underlying mechanism of obesity formation is an imbalance between energy input and expenditure (Moran 1999; Kipping, Jago et al. 2008)
Genetic and environmental factors greatly influence the body's energy balance. Nevertheless, genetic conditions which either cause production of excessive fat in the body or reduce the rate at which it is broken down, of which Prader-Willi syndrome is an example account for less than 5% of obese individuals (Speiser, Rudolf et al. 2005), with environmental factors accounting for a very high percentage (French, Story et al. 2001).
The major cause of the rising obesity problem is arguably changes in physical and social environments (French, Story et al. 2001). In recent times, there has been a remarkable shift towards activities that do not promote energy expenditure, for example, most children would travel to school in cars rather walk, in contrast to what obtained in the 1970s (Popkin, Duffey et al. 2005; Anderson and Butcher 2006). There is evidence to suggest that obese children are less active than their non-obese counterparts, hence promoting physical activity such as walking or exercising will help prevent obesity in children (Hughes, Henderson et al. 2006).
Media time (television viewing, playing video games and using the computer) has been identified as one of the significant environmental changes responsible for the surge in childhood obesity. Besides promoting physical inactivity, it encourages energy input via excessive snacking and inappropriate food choices as a result of television advertisements (Ebbeling, Pawlak et al. 2002; Speiser, Rudolf et al. 2005). Robinson in his study reveals that “between ages 2 and 17, children spend an average of 3 years of their waking lifetime watching television alone” (Robinson 1998).
Parents play a significant role in where, what and how much their children eat and to an extent, how physically active their children are. In most homes, children make their food choices based on the options they are presented with by their parents, and they characteristically would go for wrong option, more so if they have an obese parent (Strauss and Knight 1999).
Other changes within the family such as physical inactivity and working patterns of parents have contributed somewhat to the obesity epidemic. In a family where the parents work full-time, there tends to be very little time for them to prepare wholesome home-made meals and this could possibly explain the increasing demand for eating out (Anderson and Butcher 2006) thereby increasing intake of high energy dense food.
Children's attitude to and participation in physical activities depends largely on how physically active their parents are. Thus children of sporty parents embrace exercise heartily and are therefore less prone to becoming obese.(Sallis, Prochaska et al. 2000).
In addition to these family factors, societal factors such as high crime rate, access to safe sports/recreational facilities, transportation and fewer physical education programs in schools significantly impact on energy balance (Koplan, Liverman et al. 2005; Popkin, Duffey et al. 2005; Topp, Jacks et al. 2009).
French summarizes the environmental influence on obesity by opining that “The current epidemic of obesity is caused largely by an environment that promotes excessive food intake and discourages physical activity” (French, Story et al. 2001)
CONSEQUENCES OF OBESITY
Evidence suggests that childhood obesity and/or overweight has a great impact on both physical and psychological health; causing effects such as behavioral problems and low self esteem, with a higher risk in girls than in boys (Reilly, Methven et al. 2003). Although most of the serious consequences do not become evident until adulthood, research has shown childhood obesity to be linked to metabolic disorders such as insulin resistance and type 2 diabetes, stroke and heart attacks, sleep apnea, nonalchoholic fatty liver disease, higher incidence of cancers, depression, dyslipidaemia, increased blood clotting tendency, etc (Ebbeling, Pawlak et al. 2002; Reilly, Methven et al. 2003; Kiess, Marcus et al. 2004; D. A. Lawlor, C. J. Riddoch et al. 2005; Daniels 2006; WHO 2006).
One of the long-term serious consequences of childhood obesity is that obese children are twice more likely to grow into obese adults than their non-obese counterparts (Moran 1999); however, this largely depends on factors such as age of onset, severity of the disease and the presence of the disease in one parent (Moran 1999; Campbell, Waters et al. 2001; Kiess, Marcus et al. 2004; WHO 2006). Other long term consequences include early death and adverse socio-economic consequences such as poor educational attainment and low/no income in adulthood (Reilly, Methven et al. 2003; Fowler-Brown and Kahwati 2004; Kiess, Marcus et al. 2004).
Obesity-related morbidity places a huge and growing financial demand on governments. In the UK alone, the Department of Health has reported that obesity costs the NHS and the UK economy as a whole about £1b and between £2.3b - £2.6b annually respectively, with the cost to the NHS projected to rise to £3.6b by 2010 (DH 2007).
TREATMENT AND PREVENTION
The treatment of obesity requires a multidisciplinary approach due to the multi-faceted nature of the condition (Parizkova and Hills 2004). This is aimed at reducing caloric intake and increasing energy expenditure through physical activity (Ebbeling, Pawlak et al. 2002). These interventions are more likely to be successful if the patients' family is involved and the treatment tailored to individual needs and circumstances (Fowler-Brown and Kahwati 2004). In extreme cases, options such as surgical and pharmacological treatments could be exploited. These options are very unpopular and usually not recommended because the associated health risks outweigh the benefits by far (Epstein, Myers et al. 1998; Ebbeling, Pawlak et al. 2002).
Considering the huge costs and high levels of treatment failure associated with obesity treatment (Stewart, Chapple et al. 2008), the axiom by Benjamin Franklin cannot describe any other condition better than it describes obesity management.“An ounce of prevention is worth a pound of cure”
Dietz et al confirm this by saying that prevention remains the best and most effective management of obesity (Dietz and Gortmaker 2001).
Obesity prevention interventions are usually set either in the home or at school with an objective of eliminating peer pressure and, by so doing effect behavioral change (Ebbeling, Pawlak et al. 2002). Literature suggests that the school has so far remained the choice setting for these preventive interventions despite the very limited evidence on its effectiveness (Birch and Ventura 2009).
Why is the school setting a good focus of intervention?
In a nut shell, “Schools offer many other opportunities for learning and practicing healthful eating and physical activity behaviors. Coordinated changes in the curriculum, the in-school advertising environment, school health services, and after-school programs all offer the potential to advance obesity prevention” (Koplan, Liverman et al. 2005).
PREVIOUS SYSTEMATIC REVIEWS
Systematic reviews have been conducted on the effectiveness of school-based interventions in the prevention of childhood obesity. Campbell et al (2001), conducted a systematic review of 7 randomised control trials (RCTs) (6 were school-based, varying in length of time, target population, quality of study and intervention approach). The review found that dietary and physical education interventions have an effect on childhood obesity prevalence. However, success varied with different interventions amongst different age groups. Two of the three long term studies that focused on a combination of dietary education and physical activity, and dietary education respectively reported an effect on obesity prevalence reduction. Similarly, 1 out of the 3 school based short-term interventions that focused only on reducing sedentary activity also found an effect on obesity prevalence. While this review shows that dietary and physical activity interventions based at school are effective against the risk factors of obesity, the question of generalisability and reproducibility arises as the review reports the majority of the included primary studies were carried out in the US. Most of the studies used BMI as a measure of adiposity, and BMI as has been documented varies across ethnic and racial groups (Rush, Goedecke et al. 2007), thus, it will be inappropriate to apply the findings of US-based obesity prevention interventions to children in middle and low income countries where conditions are different. There are also concerns about the methodology and study design. For example the school-based study by Gotmaker et al (1999) had limitations such as low participation rate (65%) and the researchers were unable to adjust for maturity in boys and there was also poor assessment of dietary intake. All these limitations could have been responsible for a high percentage of the reported intervention effect thus affecting the validity of the results of the study (Gortmaker, Peterson et al. 1999). The authors of the review however concluded that there is currently very limited high quality evidence on which to draw conclusions on the effectiveness of anti-obesity programmes.
A Cochrane review which is an update of the Campbell et al (2001) study by Summerbell et al (2005) has examined the impact of diet, physical activity and/or lifestyle and social support on childhood obesity prevention. Their review examined the effectiveness of childhood obesity prevention interventions which included school based interventions. Their study included 10 long-term (a minimum duration of 12 months) and 12 short-term (12weeks - 12 months) clinical trials (randomised and controlled). 19 out of the 22 studies that met their inclusion criteria were school/pre-school based. The study chose the appropriate study type; more than one reviewer was involved in the entire process of data collection, extraction and selection of included studies. In general, the study found that most of the school-based interventions (dietary and/or physical activity) reported some positive changes in targeted behaviours, but however had very little or no statistically significant impact on BMI. The reviewers stated that none of the 22 studies fulfilled the quality criteria because of some form of methodological weakness which includes measurement errors. For instance, the study by Jenner et al (1989) had no valid method of measuring food intake. The studies by Crawford et al (1994), Lannotti et al (1994) and Sallis et al (2000) had similar measurement errors. Reporting error was identified in studies by Little et al (1999) and Macdiarmid et al (1998). There were also reliability concerns about the secondary outcomes measurement in some of the included studies. The reviewers therefore expressed the need for further high quality research on effectiveness.
Kropski et al (2008) reviewed 14 school-based studies that were designed to effect a life style change, a change in BMI, decrease overweight prevalence through a change in nutrition, physical activity or a combination of both. Of the 14 studies, three were done in the UK, one in Germany and 10 in the US. The right type of studies were chosen for this review and the whole process was done by more than one reviewer, however they were unable to draw strong conclusions on the efficacy of school-based interventions because of the limited number of primary studies available and methodological or design concerns which include: small sample size (Luepker, Perry et al. 1996; Mo-suwan, Pongprapai et al. 1998; Nader, Stone et al. 1999; Warren, Henry et al. 2003), no intention-to treat analysis (Danielzik, Pust et al.; Sallis, McKenzie et al. 1993; Sahota, Rudolf et al. 2001; Warren, Henry et al. 2003), possibility of type I (Coleman, Tiller et al. 2005) and type II errors (Warren, Henry et al. 2003), unit of analysis errors (Sallis, McKenzie et al. 1993) and inconsistent results (Mo-suwan, Pongprapai et al. 1998; Caballero, Clay et al. 2003; Coleman, Tiller et al. 2005). Despite their inability to draw a conclusion on effectiveness, overall, the review found that a combination of nutritional and physical activity interventions had the most effect on BMI and prevalence of overweight, with the result largely varying from community-to-community. The nutrition only and physical activity only interventions appeared to have had a change on lifestyles of participants but either had no significant effect on the measures of overweight or no BMI outcomes were measured.
Another systematic review on the effectiveness of school-based interventions among Chinese school children was carried out by M.Li et al (2008). The authors included 22 primary studies in their review. The review reported that the primary studies showed that there are some beneficial effects of school-based interventions for obesity prevention; the reviewers however expressed their concerns that most of the studies included in the review had what they considered to be serious to moderate methodological weaknesses. Sixteen of the 22 studies included studies were cluster control trials, and there was no mention by any of the researchers that cluster analysis was applied to any of the 16 studies. In addition to lack of cluster analysis, no process evaluation was conducted in any of the studies. Only one study performed an intention to treat analysis. Twelve studies experienced dropouts, but there was incomplete information on the study population at the end of the trial and the reason for the dropouts. Additionally, none of the studies explained the theory upon which they based their intervention. There was also potential recruitment and selection bias in all the primary studies as identified by the reviewers. They stated that none of the studies reported the number of subjects that were approached for recruitment into the study. As none of the RCTs included described the method they used in randomization, neither did they state if the studies were blinded or not. The methodological flaws in a high percentage of the included primary studies could impact on the validity of the findings of the review. Again, the authors failed to reach a conclusion on the effectiveness of the interventions because of the intrinsic weaknesses found in the primary studies, and as a result state the need for more primary studies that would address the methodological weaknesses that is highly present in nearly all existing primary studies conducted on this topic so far.
The study of the efficacy of school-based interventions aimed at preventing childhood obesity or reducing the risk factors is a rather complex one. Pertinent issues on effectiveness of school-based interventions to prevent the risk factors of obesity remain that there is very limited/weak evidence on which to base policies on. Heterogeneity of primary research (in terms if age of study population, duration of intervention, measurement of outcomes and outcomes measured) makes further statistical analysis nearly impossible. BMI is currently the most widely used measure of overweight and obesity in children. However, BMI has no way of distinguishing between fat mass and muscle mass in the body and might therefore misdiagnose children with bigger muscles as obese. Another disadvantage of using BMI in overweight measurement is its inability of depicting the body fat composition (Committee on Nutrition 2003), other surrogate indicators of adiposity may be needed.
Most authors that have carried out a review on this topic so far have expressed the need for further research on this topic to add to the existing body of evidence.
RATIONALE FOR THIS STUDY
All the systematic reviews on this subject so far have focused mainly on the United States. Lifestyle differences such as eating habits between American and British children possibly affect generalisability and reproducibility of US findings to the UK. For example, in the US, research has shown that 0.5% of all television advertisements promote food, and that about 72% of these food advertisements promote unhealthy food such as candy and fast food (Darwin 2009). In the UK paradoxically, the government in 2007 enforced regulations banning television advertisement of unhealthy foods (foods with high fat, salt, and sugar content) during television programmes aimed at children below 16 years of age (Darwin 2009). Thus US children are at a higher risk of becoming obese than their UK counterparts as a result of higher rate of exposure to TV junk food advertisements.
Another lifestyle difference between American and British children is physical activity. In the UK, a high percentage of children aged 2 to 15 achieve at least 60 minutes of physical activity daily (about 70% of males and 60% of females) (DoH 2004), as opposed to the US where only about 34% of school pupils achieve the daily recommended levels of physical activity daily (CDC 2008). These differences highlight the importance of public health policies being based on the local population characteristics rather than on imported overseas figures. There is therefore need to review the evidence of UK school-based obesity interventions to inform policy relevant to the UK population.
To the best of my knowledge following an extensive literature search, no systematic review has been conducted on the effectiveness of school-based intervention in preventing childhood obesity in the UK, despite the high prevalence of the condition and its public health significance in this country. This research aims to bridge this gap in knowledge by focusing on UK based studies to evaluate the efficacy of school-based interventions in the UK population.
This study therefore stands out insofar as it will be assessing the effectiveness of school-based interventions in the reducing the risk factors of obesity in the UK, with a hope of providing specific local recommendations based on UK evidence. This type of review is long overdue in the UK, considering that the government's target to reduce childhood obesity to its pre-2000 levels by the year 2020 (DoH 2007) will require local evidence of effective interventions to succeed.
The next stage of this review will describe in detail the research methodology to be used to conduct the proposed systematic review. Also included will be research strategy details to be adopted, study selection criteria, data collection and analysis.
AIMS AND OBJECTIVES
The aim of this research is to:
CRITERIA FOR INCLUDING STUDIES IN THIS REVIEW
This review was performed as a Cochrane review. The Cochrane guidance on systematic reviews and reporting format were as far as possible adhered to by the author (Green, Higgins et al. 2008). The entire review process was guided by a tool for assessing the quality of systematic reviews, alongside the accompanying guidance (health-evidence.ca 2007a; health-evidence.ca 2007b).
TYPES OF STUDY
In the search for the effectiveness of an intervention, well conducted randomised control trials (which are the best and most credible sources of evidence) will be the preferred source of studies for this review. However, because of the limited number of RCTs conducted on this topic so far, this study will include controlled clinical trials if there is insufficient availability of RCTs.
TYPES OF PARTICIPANTS
School children under 18 years of age
TYPES OF INTERVENTIONS
Interventions being evaluated are those that aim to:
Interventions not included in this study are:
For each intervention, the control group will be school children not receiving the intervention(s).
TYPES OF OUTCOMES MEASURED
SEARCH METHODS FOR IDENTIFICATION OF STUDIES
The electronic databases OVID MEDLINE® (1950-2009), PsycINFO (1982-2009), EMBASE (1980-2009) and the British Nursing Index (1994-2009) were all searched using the OVID SP interface. The Wiley Interscience interface was used to search the following databases: Cochrane Central Register of Controlled Trials and Database of Abstracts of Reviews of Effects.
There was also a general search of internet using Google search engine, in an attempt to identify any ongoing studies or unpublished reports before proceeding to search grey literature sources.
For references to childhood obesity prevention in schools, the following grey literature sources were searched:
Additionally, current control trials database at http://www.controlled-trials.com/ was searched for any ongoing research. The UK national research register was also searched at https://portal.nihr.ac.uk/Pages/NRRArchive.aspx. All the links to the grey literature databases were tested at the time of this review and found to be working.
It was not possible to conduct a hand search of journals due to pragmatic reasons.
Reference lists of retrieved studies were searched for other potential relevant studies that might have been omitted in the earlier search.
First author of all included studies were contacted with a view to seeking more references.
DATA COLLECTION AND ANALYSIS
Selection of studies
The abstracts and titles of the hits from the electronic databases searched were screened for relevance by a single assessor. Those that were thought to be potentially relevant were retrieved and downloaded unto EndnoteTM to make the results manageable and also avoid loss of data. At the end of the search, all databases were merged into one single database and duplicated records of the same study were removed.
Subsequently, the assessor then sought and obtained the full text of, and reviewed the relevant studies that were considered eligible for inclusion. Multiple reports of same study were linked together. No further data were sought for studies not included in the review.
Data extraction from included studies was done by a single reviewer and the data recorded on a data extraction form. A summary of each included study was described according to these characteristics: Participants (age, ethnicity etc.), study design, description of school-based interventions, study quality and details such as follow-ups and date, location, outcomes measured, theoretical framework, baseline comparability and results
Assessment of methodological quality of included studies
A number of researchers (Jackson, Waters et al. 2005) and the Cochrane guidelines for systematic reviews of health promotion and public health interventions (Rebecca Armstrong, Waters et al. 2007) strongly advise using the Quality Assessment Tool for Quantitative Studies (2008a) developed by the Effective Public Health Practice Project in Canada and the accompanying dictionary (to act as a guideline) (2008b) in assessing methodological quality. Based on criteria such as selection bias, study design, blinding, cofounders, data collection methods, withdrawals and drop-outs and intervention integrity, the tool which is designed to cover any quantitative study employs the use of a scale (strong, moderate or weak) to assess the quality of each study included in the review.
Considering the small number of studies included in the review and heterogeneity in terms of interventions, delivery methods, intensity of interventions, age of participants, duration of intervention and outcomes measured, it was not statistically appropriate to undertake a Meta analysis, which admittedly would have been the preferred method of analysing and summarising the results of the studies. A narrative synthesis of the results was done instead.
DESCRIPTION OF STUDIES
Results of the search
The search of electronic sources identified 811 citations out of which 97 potential studies were retrieved. A reference management software EndnoteTM was used to search for and remove duplicate citations. Further screening of title and abstract reduced the number of citations to 17 potential studies. Full texts of the 17 studies were sought, 13 were excluded, and four met the inclusion criteria and were therefore included in the review. Authors of the four studies were then contacted in view to obtaining additional references. No relevant papers were retrieved through the grey literature search.
There were no ongoing studies at the time of this review
Four school based intervention studies carried out in the UK were included in the review.
The aim of the Active Programme Promoting Lifestyle Education in Schools (APPLES) project in Leeds (Sahota, Rudolf et al. 2001), a multidisciplinary and multiagency programme was to reduce risk factors of obesity in primary schools by influencing dietary and physical activity behaviour, by promoting lifestyle education. The intervention was underpinned by the Health Promoting Schools philosophy and involved the whole school community including parents. In the Southwest of England, the Christ Church Obesity Prevention Programme (CHOPPS) (James, Thomas et al. 2004) aimed to prevent excess weight gain by discouraging the consumption of carbonated drinks amongst school children. On the other hand, a pilot study, the “Active for life year 5” project in the South Gloucestershire aimed to examine the effects of lessons on physical activity, nutrition and screen viewing on time spent involving in sedentary activities. It also evaluated the feasibility of adapting lessons from a US intervention (“Eat well and keep moving”) for use in the UK. Fourthly, the aims of the “Be Smart” intervention in Oxford were to promote healthy diet and/or physical activity in school children and prevent childhood obesity. The development of this intervention was based on the Social Learning Theory.
The four intervention programmes employed various media for the delivery of the interventions. The APPLES intervention, over one academic year in September 1996 to July 1997 targeted obesity by promoting healthy eating and physical activity via the school curriculum. The intervention was a multidisciplinary and multiagency programme that embarked on teacher training sessions, modification of school meals to exclude unhealthy foods, and the development of school action plans. Whilst the intervention group received this intervention, the comparison received no intervention.
One of the two school based intervention project to involve parents was conducted in Oxford. The “Be Smart“ intervention began in January 2000 and lasted for 20 weeks over four school terms. The intervention involved delivering a 25-minuite interactive and age-appropriate lesson to each intervention group at lunch-time clubs and targeted behavioural change. Four of the authors were involved in the delivery of the lessons, which was delivered weekly in term one and fortnightly in subsequent terms. To ensure continuity, the same author taught the same intervention to the same intervention group for the entire duration of the project.
In South Gloucestershire, the Active for life year 5 intervention was conducted over a five-month period from February 2006 to June 2006. This intervention was a multi-component one which was adapted from the “Eat Well Keep Moving” project in the US. It involved delivering of interactive lessons on nutrition, physical activity and screen viewing by trained primary school teachers.
Lastly, the CHOPPS intervention lasted for one school year from August 2001 to October 2002. The intervention was delivered by one of the authors, and was targeted at behavioural change by discouraging the consumption of carbonated drinks. Each class received a one-hour session each term. The first session focused on balance of good health and the ill-effects of carbonated drinks. The subsequent sessions comprised of music competition, presentation of art and a quiz based on a popular television game show.
All the four studies had RCT designs. However, the “Be smart” project (Warren, Henry et al. 2003) did not specify how randomisation was done. All programmes had an intervention and a control group for baseline and post-intervention comparison. The APPLES project (Sahota, Rudolf et al. 2001) was single-blinded and compared outcomes in primary school children aged 7-11 years in Leeds (intervention: n = 314, and control group: n = 322, Boys: 51% and girls: 49%) at baseline and at the end of the intervention. Evaluating outcomes in a slightly similar age group, the “Active for Life Year 5” project, a double-blinded cluster RCT compared outcomes in 9-10 year old primary school children in SW England [Intervention schools = 10 clusters (n = 331), control = 9 clusters (n = 348)]. In Oxford, the “Be Smart” project (Warren, Henry et al. 2003) evaluated its intervention in 5-7 year old school children via a group RCT. There were 3 intervention groups (Eat smart: n = 56, Play smart: n = 54 and Play/Eat Smart: n = 54) and a control group (Be Smart: n = 54). Finally, the “CHOPPS” (James, Thomas et al. 2004) project in SW England was a cluster RCT that assessed its intervention in 7-11 year old school children. There were a total of 19 clusters in the study [Intervention: 15 clusters (n=325); Boys: 169, Girls: 156, and control: 14 clusters (n = 319); Boys: 155, Girls: 164.
In all the four studies, the effect of the intervention was assessed by collecting the data on growth, measured in terms of BMI (height and weight). There was repeat measure of height and weight in all four studies at baseline, and at the end of the study for “Active for life year 5” (Kipping, Payne et al. 2008), APPLES (Sahota, Rudolf et al. 2001) and CHOPPS interventions (James, Thomas et al. 2004). For the “Be smart” intervention, the repeat measure was taken a month post-intervention (Warren, Henry et al. 2003).
Besides weight and height, the individual studies assessed other different outcomes. The “Active for life year 5” intervention assessed time spent doing screen-viewing activities (Watching DVDs, television, Videos and playing computer games) and mode of transport to school using questionnaires completed by the children (at baseline and at the end of the study).
The APPLES interventions additionally assessed the children's diet (using a 24hour recall and 3-day food diaries), their knowledge on nutrition and physical activity (via focus group), their psychological state and how physical activity they are (both using questionnaires).
In SW England, the CHOPPS project in addition to height and weight assessment, consumption of carbonated drink and water were also assessed (using 3-day diaries completed by the children).
Nutrition knowledge, physical activity and diets were also assessed using questionnaires in the “Be smart” intervention in Oxford.
Thirteen published studies were excluded from this review. The reasons for excluding these studies are provided in Appendix 5 and include location (studies carried out outside the UK) and study design (non-RCTs or CCTs).
Methodological quality of included studies
In terms of overall quality, one of the four included studies has been rated as strong (James, Thomas et al. 2004), two as moderate (Sahota, Rudolf et al. 2001; Kipping, Payne et al. 2008) and one as weak (Warren, Henry et al. 2003).
Three of the four studies had cluster RCT study designs (Sahota, Rudolf et al. 2001; James, Thomas et al. 2004; Kipping, Payne et al. 2008) whilst one had a CCT design (Warren, Henry et al. 2003). Of the three RCTs, only one reported blinding of both participants and assessors (Kipping, Payne et al. 2008). Number of participants in all four studies ranged from 213 to 679 school children, and all studies reported follow-up of up to at least 80% of participants.
In terms of cofounding, all four included studies reported that there were no significant baseline differences between the intervention and control groups, and as such rated were rated as strong (Sahota, Rudolf et al. 2001; Warren, Henry et al. 2003; James, Thomas et al. 2004; Kipping, Payne et al. 2008).
Validity and reliability of data collection tools were addressed by all four studies, with a few of them identifying issues with validity. Collection tools used for height and weight in all four studies were reliable and valid. However, there were a few validity issues with the questionnaires used for assessing other outcomes such as physical activity and dietary intake. For example, the “Active for Life Year 5” project expressed concerns that the questionnaires used for assessing physical activity though reliable, might not have been sufficiently valid (Kipping, Payne et al. 2008). Similarly, in the “CHOPPS” intervention, there were issues around validity of self collected diary data owing to the possibility of under-reporting by the children (James, Thomas et al. 2004). Again in the “APPLES” intervention project, problems with dietary and behaviour change assessments were reported (Sahota, Rudolf et al. 2001)
Effects of interventions
Only one of the four studies compared effectiveness of different types of school based interventions (Warren, Henry et al. 2003).
Only two of the four studies measured adiposity using indices other than BMI. The “Be smart” programme (Warren, Henry et al. 2003) measured skin-fold thickness at five sites using a Holtain skinfold calipers. Waist circumferences were also measured at four sites using a standard tape measure. However, there was no comparison either at baseline or post intervention of adiposity between the intervention and control groups using these measures.
Similarly, the CHOPPS programme (James, Thomas et al. 2004) measured waist circumference at the point of flexure as the child bends to one side (deducting 1cm to account for clothing). The scores were converted to z scores and comparisons were made between the intervention and control groups. However, no significant changes were observed.
All four studies reported results for BMI in terms of height and weight, and found no significant differences between control and intervention groups at the end of the studies (Sahota, Rudolf et al. 2001; Warren, Henry et al. 2003; James, Thomas et al. 2004; Kipping, Payne et al. 2008). Height and weight measurements were taken using standardised measurement tools in all four studies. Although all the studies reported no significant BMI differences, in terms of the number of overweight children, the “CHOPPS” intervention (James, Thomas et al. 2004) recorded a 7.5% increase in number of overweight children in the control clusters, compared with a 0.2% decrease in the intervention group (Mean difference 7.7%, 95% CI: 2.2% to 13.1%) at 12 months. At three-year follow-up however, the prevalence of overweight had increased in all the groups (intervention and control), which meant that the significant difference previously recorded at the end of the study was no longer evident.
Two of the four studies assessed the children's knowledge about physical and nutritional education at baseline and post intervention. The “Be smart intervention”(Warren, Henry et al. 2003) assessed nutrition knowledge using a questionnaire where children were shown pictorial representations of different kinds of food and asked to choose the one they thought was healthiest. The end of study analysis showed an increase in nutrition knowledge in both control and intervention groups when compared to the initial stages (p<0.01, p<0.001).
Although unquantifiable, the “APPLES” programme (Sahota, Rudolf et al. 2001) through a focus group discussion found that when compared to the control group, children in the intervention group had a greater understanding of the health benefits of staying active and healthy eating and were also more able to recall all most of the lessons they were taught during the intervention.
Dietary intake was assessed by three of the four studies (Sahota, Rudolf et al. 2001; Warren, Henry et al. 2003; James, Thomas et al. 2004). Sahota 2001, assessed dietary intake using a 24hour recall (using a checklist, where children were required to tick the foods eaten from a list of possible foods) and a free form three day food diary. At the end of the study, an analysis of the 24hour recall showed a 50% increase in vegetable consumption amongst intervention children when compared to the control group (weighted mean difference of 0.3, 95% CI 0.2 to 0.4). However, the three day diary did not show any significant difference; possibly because of the low completion rate of the food diaries.
The “Be smart” intervention (Warren, Henry et al. 2003) similarly reported an overall increase in vegetable (p<0.05) and fruit (p<0.01) consumption, with no significant differences between the control and intervention groups or genders at baseline or final stage. Intervention group analysis showed that the “Eat smart” and “Be smart” groups recorded a significant increase (p<0.05) in fruit and vegetable consumption when compared to the other intervention groups. A significant increase in fresh fruit consumption was recorded amongst males (p<0.01) when compared to females. No significant changes in consumption of foods high in fat were observed amongst the groups.
The CHOPPS intervention (James, Thomas et al. 2004) measured change in diet by assessing the children's consumption of carbonated drinks using a three day diary. The children were required to record their carbonated drink consumption over two weekdays and one weekend day in a diary. At the end of the study, an analysis of the carbonated drink diaries showed a reduction in the consumption of carbonated drinks in the intervention group compared with the control group (mean difference 0.7 95% CI: 0.1 to 1.3). There was also an increase in water consumption in both the control and intervention groups, but no significant difference between the two groups was recorded.
Physical activity levels
The “Active for life year 5” programme measured physical activity levels by assessing mode of transportation to school and time spent on screen-viewing activities (Kipping, Payne et al. 2008). A questionnaire about the length of time spent of screen-viewing activities (watching televisions, DVDs or playing computer games) was given to the children to complete. The end of study analysis revealed that although the children from intervention groups spent less time on screen- viewing activities when compared to children from the control group, however, the differences between the two groups did not reach a statistically significant level (mean difference at the end of intervention between the two groups adjusted for clustering and baseline: -11.6 minutes, 95% CI: -42.7 to 19.4 for weekday and -15.4 minutes 95% CI: -57.5 to 26.8 for Saturday). The study also found that at the end of the study, children from the control school had higher odds of walking/cycling to school (after adjusting for baseline difference).
Similarly, the “Be smart” intervention (Warren, Henry et al. 2003) assessed physical activity patterns rather than levels by asking the children questions about their mode of transport to school, and activities they undertake a break times. Questionnaires about how physically active the children are after official school hours were issued to their parents to complete. Information on the parental questionnaire included the frequency and duration of their child's habitual attendance of after-school clubs, screen-viewing activities and outdoor play. The post intervention analysis of the questionnaires revealed a slight increase in the number of children that walked to and fro school in both intervention and control groups. For playground activity, an increase was also recorded in all groups, with a higher increase in all intervention groups when compared with the control groups. Overall, there was no significant gender difference in playground activities at either baseline or post intervention. Similarly, the parental questionnaires reported no intervention effect on activity levels after school hours.
The APPLES intervention (Sahota, Rudolf et al. 2001) used a questionnaire to measure physical activity levels and sedentary behaviour in the children. The questionnaire was categorised by how frequent the children were involved in outdoor sporting activities such as swimming and frequency of sedentary activity such as watching television, in the past 24 hours. An analysis of the questionnaires showed no significant difference in physical activity levels in the intervention and control groups. What it however showed, was a 33.3% increase in sedentary activity in overweight children in the intervention group.
Summary and discussion of main results
None of the four included studies reported significant short-term changes in BMI at baseline and post intervention. The fact that no significant BMI changes were detected does not in any way imply evidence of ineffectiveness. Possibilities are that small sample/unit sizes and short intervention duration (in all included studies) might have resulted in the inability to detect any weight/height changes. Previous school based intervention studies that have reported significant anthropometric changes in school children both lasted for a minimum duration of two years (Dwyer, Coonan et al. 1983; Gortmaker, Peterson et al. 1999); which is a reasonable time frame to expect any anthropometric changes.
Despite the lack of significant anthropometric changes, changes were reported for some other outcomes measured in the primary studies. There was a modest increase in vegetable and fruit consumption in two of the studies (Sahota, Rudolf et al. 2001; James, Thomas et al. 2004). Also a significant reduction in the consumption of carbonated drinks and an increase in water consumption were reported in the CHOPPS project. Although the changes did not reach significant levels, the “Active for life year 5” intervention reported a reduction in the time spent on screen viewing activities in the intervention group when compared to the control group.
Quality of the evidence
Given that this study is a review of intervention effects, the study designs of the included studies were the appropriate types to answer the study question. In terms of global rating, one study was rated as strong, two as moderate and one as weak. However, some caution is required in interpreting findings from this review as all of the studies had some limitations such as small sample/unit sizes and issues around concealment. All these may have introduced a possible systematic measurement bias.
Another major issue with all the studies was the short intervention duration. Considering that all the studies measured change in adiposity in terms of weight and height, realistically, it takes a considerable length of time to actually notice a change in either weight or height following an intervention. This shortcoming may have possibly made statistically significant changes difficult to detect.
Potential biases in the review process
The guidance in Cochrane Handbook for Systematic Reviews of interventions (Higgins and Green 2008) was followed throughout the review process as far as possible. A quality assessment tool for assessing methodological quality of systematic reviews (health-evidence.ca 2007a; health-evidence.ca 2007b) was used in this review to assess the quality of included studies. Judging by the principles set in these resources, a number of potential limitations have been identified in this review.
Firstly, in order to minimise errors, limit bias and improve reliability of findings, the Cochrane guidance recommends that key steps of a systematic reviews such as selection of studies and data extraction should be undertaken by more than one reviewer. This was however not possible due to the nature of this piece of work.
Although there is strong evidence that RCT is “the least bias estimate of effect size” (Campbell, Waters et al. 2001) and the preferred method for estimating the effectiveness of interventions (Stephenson and Imrie 1998), there is still a lot of debate around its usefulness in assessing the effectiveness of lifestyle and behavioural interventions (Campbell, Waters et al. 2001). Nevertheless, a majority of the studies included in the review have Randomised controlled trial designs.
During the search for studies, efforts were made to comprehensively search all relevant sources such as RCT register and social science databases. Additionally, firs authors of all included studies were contacted. Grey literature sources were also searched. Despite these efforts, it is possible that hand searching of key journals may have identified additional potentially relevant studies.
No language restriction was imposed during the search, and as such, no potentially relevant studies were excluded on a language basis.
Agreements and disagreements with other studies or reviews
Implications for practice
Despite the need for more research identified by this review, some evidence that school based interventions could have some positive impact on lifestyle behaviours that places children at risk of becoming obese was also found. Although positive effects in terms of adiposity were not shown, no harmful effects of the interventions were shown either. School based anti-obesity interventions should therefore be promoted by local public health authorities and encouraged by schools for long term prevention of obesity and its associated adverse health effects.
Considering that children do not have any say at home in terms of purchasing food, involving parents actively in school based interventions could produce a sustained positive effect on children outside of the school setting.
Although the study concludes that there is insufficient evidence on the efficacy of school based intervention in preventing childhood obesity, this does not mean evidence of ineffectiveness of these interventions. Given that some positive changes were reported in all the included studies, promotion of school based anti-obesity interventions is greatly encouraged at this time, as these interventions have demonstrated the potential to be beneficial on the long run.
Implications for research