Impact of a community-based pilot intervention to tackle childhood obesity: a ‘whole-system approach’ case study

The Go-Golborne intervention was developed by the Royal Borough of Kensington and Chelsea’s (RBKC) public health team to promote healthy lifestyles amongst children and families, as part of a broader programme to tackle childhood obesity. A third of year six children across RBKC are overweight or obese, and prevalence is above the London and national averages in several of the most deprived wards [1, 2]. Variations in prevalence are strongly linked to income and socio-economic status; higher rates of obesity tend to be concentrated in areas with high levels of deprivation. The RBKC council chose to pilot a targeted approach to identifying and addressing barriers to a healthy lifestyle at a community level within Golborne ward: an area with a diverse population and relatively high deprivation and obesity prevalence. There were around 900 children living in the ward, and over 1700 children attending six local schools [2].

There is a great deal of literature on behaviour modification or lifestyle change in the prevention and management of childhood obesity, influenced by several different theories, concepts and accounts of behaviour and behaviour change. Evidence of effectiveness for behavioural interventions has been mixed with small, short term effects on weight loss and Body Mass Index (BMI) [3, 4]. However research has highlighted key behaviour change techniques that are most likely to promote positive changes (e.g. provide information on the consequences of behaviour and environmental restructuring) [5]. Health behaviours are influenced by a range of socio-economic, cultural and environmental conditions, social and community networks and individual factors such as age and sex. Therefore, a combination of interventions that tackle population, community and individual-level factors are needed to help people change their behaviour in the longer term [6]. Systematic reviews of interventions and clinical guidelines indicate that successful interventions are complex and multi-component - aimed at changing both physical (or sedentary) activity and diet or healthy eating, and comprise multiple, potentially interacting methods of changing behaviour [3, 7‐10]. In general, interventions which involve the whole community in complex interventions that target environments and upstream determinants appear to be more effective than those which simply target children [9].

The increased recognition of the complexity of obesity causation and prevention, and a frustration with the lack of success of efforts over the last few decades, has led to a growing interest in whole systems approaches (WSAs) [11, 12]. In theory, WSAs draw on understandings of complexity science and of complex adaptive systems that help to explain particular problematic situations and identify ways in which they might be improved. However, what is meant by a whole system is interpreted in different ways. In practice, they are often described in terms of ‘big picture’ thinking, where efforts are made to link together actions in a coordinated and integrated effort, across multiple sectors, to bring about change [13]. According to Public Health England, “a local whole systems approach responds to complexity through an ongoing, dynamic and flexible way of working … stakeholders agree actions and decide as a network how to work together in an integrated way to bring about sustainable, long-term systems change” (p.17) [12]. Whilst community-based and local WSAs to health and wellbeing are not the same thing, they share many common features, such as community engagement, long-term commitment, a focus on relationships and networks and dynamic understandings of causes and effects [14].

There is a paucity of evidence on the effectiveness of community-wide programmes displaying features of a WSA to prevent obesity. A systematic review of population-based whole-of-community obesity prevention interventions published in 2014 identified eight trials, none of which were undertaken in the UK [15]. The review suggested that such interventions can be effective in achieving modest reductions in population weight gain among children, but there is a paucity of evidence, particularly for the UK context. Since that review, there have been important additions to the evidence base, particularly from Australia where experiences in implementing community-based childhood obesity prevention projects in different contexts and communities found that the effectiveness of intervention strategies is dependent on individual and community factors. This reinforces the call for a systems approach whereby existing systems are modified [16]. The language, theory and practice of WSAs – certainly within the public health field - is still young. Understandings of how best to apply systems thinking and what a WSA to obesity looks like in practice are newly emerging [12] and there is little knowledge yet of what is most likely to work.

Go-Golborne was designed in 2014 and implemented over 3 years (2015–2018). It sought to engage all those with a role in shaping the environments in which children live, learn and play in Golborne. The programme team described it as a WSA in that it combined “‘bottom-up’ community empowerment actions with ‘top-down’ interventions in a single initiative” and aimed to “use and optimize existing systems, build on local assets, connect multiple stakeholders, synchronize ongoing activities across multiple settings and stimulate further actions” (p.110) [17]. Its methodology and design are detailed in a separate paper [17], but it was informed by the best practice principles for community-based obesity prevention developed in Australia [18], the World Health Organisation Good Practice Appraisal Tool [19] and the EPODE approach to childhood obesity prevention [20]. The PESTEL framework (distinguishing political, economic, sociocultural, technological and physical and legal environments) was used to explore and describe the influences that hinder or support the adoption of healthy lifestyles in the community. From this, and through stakeholder engagement, a programme plan was developed that included: social marketing campaigns every 6 months, covering specific nutrition and physical activity themes; training and development opportunities for people working with children and families; the distribution and promotion of consistent messages on nutrition and physical activity; environmental improvements, working with council departments, local agencies and other stakeholders such as local retailers; and community development activities, including a grant scheme for each theme, local events and other ad hoc support. The high-level programme theory was that by engaging the whole community and stakeholders within the ward and across the council in a geographically-focused initiative, locally appropriate and co-developed activities would be designed and delivered to raise awareness and understanding of the issues, and encourage and support behaviour change amongst children and their families (see Theory of Change, Fig. 1).

Researchers at the University of Kent were commissioned to design and conduct a robust research evaluation (from May 2015 to April 2019). This paper provides a summary of the overarching findings. A thorough analysis of the quantitative data, and a more detailed report of the process evaluation findings will be available in separate articles (in progress).

The evaluation was designed to answer questions associated with process, outcomes, and implications for future programmes and policy. It took a theory of change approach [23‐24], which clarified the programme’s aims, objectives and outcomes and articulated the assumptions underlying the programme’s design (see Logic Model, Fig. 2). Data collection, management and analysis was guided by the RE-AIM framework [25], which focuses on essential programme elements (reach, efficacy, adoption, implementation, and maintenance) that can improve the sustainable adoption and implementation of evidence-based interventions.

Data was collected and analysed to measure programme reach, assess implementation fidelity, and examine programme context, from: eight steering group meetings; ten stakeholder group meetings; event log forms (completed by the programme team); progress reports; eight newsletters; three in-depth interviews with the programme co-ordinator; and attendance records and other programme documentation.

A non-experimental case study design and mixed methods were used to evaluate a range of indicators, in accordance with the logic model, at baseline, mid-term and follow-up where possible. Data sources included:

All interviews were either face-to-face or via telephone. Most were audio-recorded with consent from participants; for two parent interviews, detailed notes were taken. Focus group discussions were audio-recorded, except the one with mothers at a children’s centre, for which detailed notes were taken by a scribe.

A thorough review of all existing validated questionnaires identified none that would meet our objectives and be appropriate for primary school-aged children. For example, when exploring how to capture active play, few of the existing questionnaires dealt with physical activity or exercise that might be considered ‘active play’ (e.g. Physical Activity Questionnaire for Older Children). The Day in the Life Questionnaire (DILQ) [26] asked what children did yesterday at morning break, lunchtime, and after school, but there was no attempt to measure frequency, intensity or duration. All questionnaires therefore were collaboratively designed by the evaluation and programme teams. Subject experts were consulted, and the survey structure and some specific questions were drawn from our review of existing validated questionnaires. The child questionnaire was designed to measure any significant change, over time, in the population health behaviours of children in Golborne. Questions were in a simple and suitable format (adapting those used in existing questionnaires for young children such as the DILQ), with embedded audio files and clear graphics to aid comprehension. Children in years 5 and 6 were asked 14 additional questions taken directly from the Child Nutrition Questionnaire [27] to assess attitudes towards eating fruit and vegetables. The questionnaire was pre-tested with seven children aged seven to 11 in order to explore comprehension, retrieval/recall, judgement and response. It was then piloted in a primary school with similar pupil profile (N = 91) and subsequently refined by the evaluation and programme teams.

The parent/carer questionnaire collected additional information on children’s behaviours, assessed parent awareness and knowledge around key themes, and assessed parents’ behaviours in relation to supporting healthy eating/activity in their children. The partner questionnaire explored the support partners received from the Go-Golborne team, the extent of partner engagement, how information was being received and used, whether capacity had been strengthened, and whether community partners were doing anything new or different to support children in making positive behaviour changes.

Height and weight data for school years one to five were combined with the routine NCMP dataset to add in Reception and year six. For the extended NCMP data, the LMSgrowth tool was used to calculate BMI, BMI Standard Deviation (z-score) and BMI percentile based on sex, date of birth, date of measurement and height and weight values [28]. Weight classifications were determined using the UK90 BMI reference curves [29]. For routine NCMP the validated percentiles as provided by Public Health England were used (LMS results in the same BMI groups for these measures). For clinical BMI groups the following centiles were used as cut-offs: underweight: ≤2.3, healthy weight: 2.4 to < 90.9, overweight 90.9 and over, very overweight 97.7 and over. The analysis consisted of a series of pupil counts under different variables, e.g. by BMI classification.

Qualitative data was analysed using thematic analysis of either full audio transcripts or detailed notes, using the theory of change as an analytical framework, to which sub-themes were added inductively [30]. Two researchers independently coded a sub-set of transcripts until agreement and confidence was reached. One researcher conducted the remaining coding, bringing any arising issues to the research team for discussion and consensus. Analysis templates were populated in Microsoft Word for each data source to identify the key data organised into themes; data within each theme were synthesised into thematic statements. This enabled a close link to the data to be maintained to ensure analysis remained grounded in the data, and to ensure that a range of data sources, contributed to building explanatory models. For the overarching analysis across data sets, prominent and recurring themes from across the data were extracted, matched and cross-compared to develop an explanatory case for the propositions at the heart of the Go-Golborne programme [31]. Rival explanations were also sought and interrogated.

This section presents a summary of the main evaluation findings in relation to the elements in the RE-AIM framework.

Go-Golborne was developed as a pilot approach to identifying and addressing barriers to a healthy lifestyle at community level, and a potentially effective way of reducing child obesity. The challenge set by the council was ambitious. It aimed, within 3 years, to engage the whole community and stakeholders within the ward and across the council, to design and deliver locally appropriate and co-developed activities to raise awareness and understanding of the issues, and encourage and support behaviour change amongst children and their families. The scale and complexity of this challenge, and the importance of context in shaping the success of the programme, was acknowledged early on by the programme and evaluation teams.

The design and conduct of the evaluation was fraught with challenges that are now comprehensively discussed in the literature [21, 23, 32‐35]. In particular, these were multiple programme components, action at multiple levels, the importance of context, the flexible and evolving nature of the programme, the breadth and long-term nature of the outcomes being pursued, and the absence of appropriate control groups for comparison purposes [36, 37]. The case study design sought to take account of these challenges by taking a theory-of-change approach with mixed methods and a strong process evaluation. It enabled an in-depth empirical investigation of the situation to understand the how and why questions within the evaluation. However, the evaluation was inherently political [38] and there were expectations to manage about what the evaluation would and wouldn’t be able to ‘prove’. The case study design enabled the integration of qualitative and quantitative data from a variety of sources to give an in-depth analysis of the situation and the context. Whilst qualitative data was inevitably from a small sample, efforts were made to reduce response bias, particularly by using purposive sampling methods and pro-active approaches (e.g. working with community groups to help engage specific participants) to ensure we incorporated perspectives of diverse audiences. The evaluation provided a detailed picture of programme operations and resulted in a rich understanding of how and why programme operations related to outcomes. However, in the measuring of behaviour-related outcomes, the team were obliged for pragmatic reasons to rely on self-reported data, which has obvious limitations in terms of recall and social desirability bias. Moreover, since there were no existing tools that suited our purpose given the need to assess change across six behaviour change themes, the evaluation used bespoke (and therefore unvalidated) questionnaires. Low sample sizes from the parent surveys limited our ability to detect small intervention impacts and generalise the findings to the whole community. Another limitation is sampling bias (participation was voluntary). Behaviour data reported by young children should be treated with particular caution. Evaluation findings should be considered, therefore, with these caveats in mind.

In Golborne, throughout the programme period, data collected via child surveys indicated that children (at a population level) did not make significant changes to their eating and physical activity behaviours. This is perhaps unsurprising given the time-frame, the scale of the programme, the complexity of the issue, and the limitations in the evaluation methodology. Similar findings, demonstrating inconsistent and limited success in changing healthy eating and physical activity-related behaviours, have been found in the evaluation of other comparable interventions [16]. As with the experience in Australia [16], aspects of the local context were important in relation to the achievement of behaviour change objectives. In particular, many factors outside of the programme’s sphere of control, such as the relative poverty of many Golborne residents and the Grenfell fire, affected both implementation and context.

Although population behaviours did not appear to change significantly, there was some evidence that a supportive environment was starting to be developed in Golborne’s homes, schools and neighbourhoods. This was created by giving community stakeholders information, skills and motivation to support children in making healthy choices. The programme sought to raise awareness and knowledge of healthy eating and physical activity through social marketing campaigns, at the same time as making micro-environmental changes through informing, engaging and supporting a range of stakeholders. The locally-designed campaigns were developed with strong community involvement and used established social structures of the community. Existing evidence suggests that such approaches are more likely to be implemented and sustained [39, 40]. However, evaluation data suggested that the raised awareness and knowledge brought about by the campaigns and the changes made at micro-level were not sufficient to achieve a reduction in child overweight. This is consistent with other literature that highlights the importance of being realistic about the potential of such programmes to alter the outcomes of a system as complex and extensive as that driving the weight status of populations, especially within a three-year period [41, 42].

Go-Golborne demonstrated that it is possible to bring stakeholders together to develop a shared commitment to tackling overweight, to recognise the part they can play, and to start to make changes in their services/behaviours. There are strengths in this design that relate to the principles of practice for collective impact, which has proven to be a powerful approach in tackling a wide range of issues in communities all over the world [43]. The findings demonstrate that capacity to tackle overweight within the Golborne community was strengthened in a number of ways, including knowledge, skills, resources and opportunities. This is significant as existing evidence suggests that a community-wide, capacity-building approach to reducing child obesity is flexible, cost-effective, sustainable, equitable and safe, and has the potential to influence the underlying social and economic determinants of health [44]. Go-Golborne strengthened and leveraged the interaction of human capital, organisational resources and social capital to help tackle child overweight as a collective problem. Most importantly, it did this in a way that strengthened community identity, built frameworks to facilitate sustainable change, and empowered the community through a strength-based approach and inclusive practice. In the context of the tragic fire at Grenfell Tower, a wide range of issues related to inequity and mental health came to the fore which couldn’t help but affect neighbouring Golborne. In this context, Go-Golborne not only managed to maintain its momentum throughout the period, but also demonstrated the value of its approach in terms of building trust, strengthening networks and reinforcing a community identity.

Go-Golborne aimed to take a ‘whole systems’ approach at a local level, which is consistent with a growing body of evidence and current thinking around how best to tackle obesity. A significant four-year action research project carried out during the same period as Go-Golborne resulted in a ‘whole systems approach to obesity’ guide and resources, published in July 2019 [12]. Using the guide as a framework of best practice, Go-Golborne could be considered as having done a good job of implementing a ‘whole systems approach’ to obesity, albeit at a very local (ward) level. It secured senior-level support and established the necessary governance and resource structure to implement the approach; it built a compelling narrative and a shared understanding of why obesity matters locally and how it can be addressed; it brought stakeholders together to understand the local system and agree a shared vision; it oversaw a number of collaborative and aligned actions; it maintained momentum by developing a stakeholder network; and it critically reflected on its approach and considered opportunities for strengthening the process. However, the explicit use of systems tools (like causal loop diagrams or group based modelling, for example) did not feature in attempts to map the local system. In addition, as a local community-centred project within a large borough, Go-Golborne placed more emphasis on Golborne-based actors and actions than on change within and driven by the Council. For improved impact, the Council should seek to scale up this systems approach to working across the whole borough – preferably in concert with a similar London-wide and indeed UK government-wide whole systems approach.

Go-Golborne represents an important attempt to implement an evidence-informed, community-based, WSA to childhood obesity prevention in a deprived inner-city ward, within a local government context that is experiencing some of the tightest financial restrictions in recent history [45].

The findings from the evaluation of the Go-Golborne intervention demonstrate that this kind of approach can establish firm foundations for supporting healthier diet and physical activity related behaviours amongst children, through engaging children and their families, schools, and the wider community. The intervention helped stakeholders and parents to develop a shared commitment to tackling overweight, to identify barriers to a healthy lifestyle, and to start to make changes in their services/behaviours. Key to this engagement was running a positive, fun and locally-tailored campaign with excellent reach into the community, broad adoption by partners, and flexible implementation plans that took account of the local context and adapted to changes and challenges. These foundations were deemed to be crucial for building trust (and therefore for acceptability of the intervention), and for maintaining the programme’s momentum in the longer-term. However, the findings also highlight the complexity of and time taken to significantly alter population behaviours, and consequently weight status. The campaigns and changes made at micro-level appeared to be not sufficient, in the face of counteracting forces and personal factors, to achieve significant behaviour change within 3 years. This highlights first, the need for local initiatives to be reinforced by supporting action at regional, national and global levels, and second, the need for all initiatives to be seen as part of a longer term vision for childhood obesity prevention.