The starting point was to review the literature to identify (i) risk factors for childhood obesity and children's current eating/drinking and physical activity behaviours (ii) the determinants of these behaviours and (iii) apparently successful and unsuccessful components of previous school-based interventions to prevent and reduce obesity.
(i)Possible risk factors for obesity
Obesity results from an imbalance between consumption and expenditure of energy. Controlled experimental and epidemiological studies suggest a number of dietary risk factors associated with increased energy intake in children and adults. These included, diets with a high energy density  usually characterised by foods high in fat and low in fibre, including fast food [9, 10] and large habitual portion sizes . Experimental studies also report that liquid calories have lower satiating properties than solid food  and epidemiological studies report an increased risk of weight gain or obesity in consumers of sugar-rich drinks. A single carbonated drink per day can add 10% to a child's energy intake . According to the National Diet and Nutrition Survey (2008/9), in the UK children's intake of non milk extrinsic sugars (NMES) provides 15% of food energy , compared to a recommendation of not more than 11% . Carbonated soft drinks are a major source of NMES providing 19% of NMES intake in children aged 4-10 and over one-third in children aged 11-18 .
Reduced energy expenditure has also been associated with weight gain  and numerous studies in adults and children reported an association between lower weight gain and higher levels of physical activity . Stratton et al reported a decrease in the levels of cardiovascular fitness in 9-11 year olds in England between 1997 and 2003 while the prevalence of obesity increased over the same time period . Children's TV viewing time and time spent playing electronic games has been associated with overweight and obesity [18–20], total calorific intake  and the consumption of snack foods . Longitudinal data from the Avon Longitudinal Study of Parents and Children (ALSPAC), found strong associations between children's fat mass at age 14 and their physical activity at age 12 . We also know that today's children are spending more time in front of the television or computer screen than in previous generations - an average of two and a half hours of TV and 1 hour and 50 minutes online a day . (i.e. nearly 4 1/2 hours a day of screen time). An attempt to encourage children to replace screen-based sedentary behaviours with more active pursuits is clearly an appropriate aim in preventing obesity in children and promoting a healthy lifestyle.
(ii)Determinants of behaviours
A variety of family and social determinants affecting children's eating and activity behaviours have been identified. For eating, these include food preferences, food availability and accessibility, modeling (copying the behaviour of others), mealtime structure (social context of meals, the role of TV during mealtimes, eating out, portion size, school meals, snacking habits), feeding styles (the caregivers approach to maintain or modify children's behaviours with respect to eating) and socio-economic and cultural factors (e.g. family time constraints, education, income, ethnicity and culture) . In terms of children's physical activity, parental support (e.g. transporting the child, observing the activity, encouraging the child, providing equipment, participating with the child and reinforcing physical activity behaviours) has been identified as a key determinant both directly and indirectly through its positive association with self efficacy perceptions . Griew et al recently reported that children's school time physical activity varied according to the primary school they attended even after accounting for individual demographic and the school compositional factors with a 'school effect' explaining 14.5% of the variation in pupils' school-time physical activity . However, it is less clear that school based activities have a substantial effect on total, as opposed to school time, activity. In a study of 3 schools from one area, with different sporting facilities and opportunity for physical activity in the curriculum, Mallam et al (2003) reported large differences in school time activity levels but virtually no differences in the total activity of the children .
This research suggests that while it appears that schools have the potential to create a positive physical activity culture that can influence whether children engage in physical activity it will be crucial in intervention studies to assess whether any effects translate in to changes in total as opposed to only school time activity.
Drawing on the social ecological approach  we began from the theoretical perspective that, while both eating and activity behaviours in children are partly determined by choices made by the children, they are highly dependent both on direct intervention by parents (e.g. the food provided, opportunities for physical activity) and by patterns of behaviour within the family, within the school and within peer groups. As children get older the relative importance of self directed, as opposed to family directed, behaviours increases and these behaviours are influenced by wider social factors which include the school environment and peer group norms. Therefore any intervention we designed needed to affect behaviour through influencing the children, their families and the school environment. There is some evidence from previous studies of interventions in children that the use of drama/theatre can be an effective tool to engage children, increase knowledge and change behaviours [30–33]. For example, in an obesity prevention programme aimed at low income children and their parents, an after school theatre-based intervention was shown to motivate and engage both parents and children and increase awareness of the need for making changes. However, the authors did conclude that theatre alone is not enough to lead to behavioural change and that the next step should be to incorporate this delivery method into more comprehensive programmes with both educational and environmental components . Two small studies in primary schools in the UK based on drama/the arts reported increases in vegetable, salad and fruit juice consumption [32, 33]. Although both these studies had serious methodological weaknesses, the use of drama to engage children to change specific behaviours looked promising and was explored at length with experts from drama and education as a possible implementation strategy in step 3 of the intervention mapping process.
We were mindful that there were other key drivers including intrinsic factors such as genes and the wider social environment but these are less modifiable and so were not considered as potential points of intervention.
(iii) School-based interventions
The most recent systematic review (2009) of controlled trials of school-based interventions identified 38 studies; 3 dietary intervention only, 15 physical activity only and 20 combined diet and physical activity . The authors concluded that there was insufficient evidence to determine the effectiveness of dietary interventions alone, but suggested that interventions which increase activity and reduce sedentary behaviour may help children to maintain a healthy weight, although results were short-term and inconsistent. Results for combined diet and activity were also inconsistent, although there was a suggestion that the combined approach might be more effective in preventing children becoming overweight in the long term. Social Cognitive Theory (SCT), which proposes that a dynamic interaction exists between personal, behavioural and environmental factors, provides a basis for many of these programmes, particularly the constructs of self efficacy, behavioural capability (knowledge and skills to perform a behaviour), outcome expectations, self regulation and reinforcement . Environmental conditions of eating behaviour such as school lunch provision and parental/home environment were often targeted [36, 37]. A review of reviews of effective elements of school health promotion across behavioural domains (substance abuse, sexual behaviour and nutrition) found that five elements from the highest quality reviews were found to be effective for all three domains using two types of analysis. These were use of theory; addressing social influences (especially social norms); addressing cognitive behavioural skills; training of facilitators and multiple components. Using one type of analysis only, another two elements were identified: parental involvement and a large number of sessions .
The authors concluded that the 5 elements identified should be primary candidates to include in programmes targeting these behaviours.