Response to school‐based interventions for overweight and obesity: A systematic scoping review

Summary Heterogeneity of response to paediatric obesity interventions is one of the greatest challenges to obesity care. While evaluating school‐based interventions by mean changes compared to control is important, it does not provide an understanding of the individual variability in response to intervention. The objective of this study was to comprehensively review school‐based interventions that reported study results in terms of response and identify definitions of response used. A scoping review was conducted using a systematic search of five scientific databases from 2009 to 2021. Inclusion criteria included randomized controlled trial design, school‐based setting, weight‐based outcomes (e.g., BMI, BMI z‐score), weight‐based outcomes analysed among youth with overweight/obesity, a study conducted in a developed country and publication in English. A total of 26 reports representing 25 unique studies were included. Overall, 19% (5/26) of articles reported response. Response was defined in three ways: maintenance/decrease in BMI z‐score, decrease in BMI z‐score ≥0.10, and decrease in BMI z‐score ≥0.20. Few school‐based interventions identified an a priori intervention goal or identified the proportion of participants who responded to the intervention. Without such evaluation participants who do not benefit are likely to be overlooked.


Funding information
Inclusion criteria included randomized controlled trial design, school-based setting, weight-based outcomes (e.g., BMI, BMI z-score), weight-based outcomes analysed among youth with overweight/obesity, a study conducted in a developed country and publication in English. A total of 26 reports representing 25 unique studies were included. Overall, 19% (5/26) of articles reported response. Response was defined in three ways: maintenance/decrease in BMI z-score, decrease in BMI z-score ≥0.10, and decrease in BMI z-score ≥0. 20. Few school-based interventions identified an a priori intervention goal or identified the proportion of participants who responded to the intervention. Without such evaluation participants who do not benefit are likely to be overlooked.

K E Y W O R D S
children, community-based, nutrition, response heterogeneity

| INTRODUCTION
The scope and severity of childhood obesity are well documented, warranting the prevalence of obesity to be referred to as an 'epidemic' and a 'public health crisis'. 1 Evaluation of paediatric obesity intervention traditionally relies on comparing mean changes between intervention and control groups. 2 This programmatic evaluation has played an important role in improving paediatric obesity intervention. For example, although schools were identified as an important focal point for intervention, 3 initially many school-based interventions did not improve child weight-related outcomes. 4,5 Through comparing mean changes between intervention and control conditions, intervention programs were revised and the effectiveness of school-based interventions has improved. 6 Despite the benefit of comparing mean changes between groups, real world conditions of community-based and public health interventions do not have control groups to which intervention outcomes can be compared. Identifying an a priori goal for intervention response is important to determine intervention utility and if an intervention is having a meaningful impact. This approach is important at all stages of intervention research and is consistent with the ORBIT model which emphasizes the need for preliminary evidence that intervention outcomes are likely to have clinically meaningful outcomes before doing larger scale efficacy trials. 7 Similarly, understanding the proportion of participants who reach clinically meaningful outcomes can help identify which interventions should be disseminated as well as to track outcomes of an intervention once disseminated. For example, changes in the proportion of participants who reach clinically meaningful outcomes overtime from an intervention could signal issues with implementation fidelity.
Currently, there is no agreed upon definition or criteria for clinically meaningful response to paediatric obesity interventions. 8 An increasing number of weight loss interventions among adults report the proportion of individuals who achieve the 5%-10% threshold of weight loss associated with cardiometabolic improvements. 2 This type of evaluation parallels the efficacy standards needed for the FDA to approve a weight loss drug. Specifically, at least 35% of individuals taking the drug must reach ≥5% weight loss for the drug to be considered effective. 9 While not used as an a priori intervention goal or evaluation metric, a wide range of paediatric response definitions have been developed out of necessity as part of secondary analyses investigating characteristics predictive of response in clinic-based paediatric intervention. Definitions used include the maintenance or decrease in standardized body mass index (zBMI) [10][11][12] or BMI represented as a percentage of the 95th BMI Percentile, 13 ≥5% reduction in zBMI, [14][15][16] ≥10% reduction in zBMI, 16 ≥0.20 reduction in zBMI, 17 or a 5% reduction in weight. 18 Response time frames ranged from post intervention 16,18 to 2 years follow-up, 13,16,17 with most studies analysing response at 1 year. 10,11,[13][14][15] The purpose of this scoping review was to (1) identify the proportion of school-based interventions that report study results in terms of response and to (2) examine definitions used for response to school-based intervention. Understanding how response to schoolbased intervention is evaluated and reported is particularly important as there is likely to be greater variation in response to school-based intervention than to clinical interventions among only treatmentseeking participants. Similarly, understanding how many students reach meaningful improvements from school-based intervention is important to determine if the intervention should be continued and/or disseminated further. With school-based interventions as an example, the findings from this review will shed light on the current state of the literature with regards to how meaningful change is being defined and reported in community-based paediatric obesity interventions.

| Study design
This scoping review intended to analyze the state of current literature on school-based obesity interventions, particularly how response is defined and reported, which falls under Kirksey and O 0 Malley's first scoping review purpose, "to examine the extent, range and nature of research activity." 19

| Inclusion/exclusion criteria
Inclusion criteria were a school-based setting, an outcome measuring weight or BMI, school-aged paediatric participants, randomized controlled trial, and publication in English in 2010 or later. With this being the initial review to examine how response to intervention was being reported, it was important to include only randomized controlled trials to keep the data synthesis as clean as possible. Included articles were limited to those published in 2010 or later because the focus on precision medicine and heterogeneity in response to obesity intervention is relatively new among paediatric populations. 2 Accordingly the most relevant articles are likely to be captured in the past decade of research. Exclusion criteria were any settings outside of school such as churches or community centres, pre-school settings, primary prevention studies or studies that did not measure weight or BMI, non-randomized controlled trial study designs, publication in languages other than English, and locations in Africa, Antarctic or Arctic regions, or Asia. These geographic region exclusions were applied to primarily limit results to westernized nations. Articles were also excluded if they assessed weight-based outcomes on samples with mixed weight statuses (i.e., only articles with an analysis for 'secondary prevention' populations were included).

| Article screening
The database searches yielded a combined 2214 records. A first round of deduplication was conducted in EndNote, and a second round was done in Rayyan, a free web-based systematic review screening tool developed by the Qatar Computing Research Institute. 21 Rayyan was then used to screen the remaining 1608 unique records for inclusion and exclusion. For the first level of screening, two researchers reviewed the titles and abstracts of the citations to eliminate articles that did not meet the minimum inclusion criteria. When opinions between these two researchers differed, a third researcher reviewed the article. Reviewers met to resolve any conflicts and ensure consistency. Disagreements were resolved by a majority vote. This same process was followed for the full-text review of articles, and articles not meeting eligibility criteria were excluded. Reasons for exclusion were documented within Rayyan. See Figure 1 for the PRISMA flow diagram.

| Data extraction
The characteristics of each article were extracted by two reviewers.
Reviewers extracted the following information regarding the reporting of response: was any definition of response reported (yes or no), definition of response if it was reported, was the difference in the proportion of participants who met the response definition tested between intervention conditions (yes or no), and the response outcome reported. Specifically, the term 'response' refers to a threshold of meaningful change in weight outcomes as defined and reported (or not) in each study. After researchers independently extracted data, it was compiled into a single spreadsheet. Data synthesis was performed through researcher discussion examining the similarities and differences of each study. As part of this discussion, tables were created to map out the characteristics of the study population (Table 1), intervention description and outcomes (Table 2), and response reporting (Table 3). These tables were used to identify patterns in how meaningful change is being defined and reported in school-based obesity interventions.

| RESULTS
A total of 26 records, representing 25 unique studies, met the inclusion/exclusion criteria to be included in this scoping review. Two studies occurred in Australia, [22][23][24] one in Canada, 25 and one in the United Kingdom. 26  No n/a n/a n/a Bogart 2016 37 No n/a n/a n/a Daly 2016 41 No n/a n/a n/a Davis 2021 31 No a n/a n/a n/a Foster 2010 36 No a n/a n/a n/a No n/a n/a n/a Kong 2013 42 No n/a n/a n/a Kubik 2021 30 No n/a n/a n/a Love-Osborne 2014 43 Yes 0.1 decrease in zBMI at 10 months yes, chi square 18.2% of intervention group compared to 40.3% of control group met response criteria at end of academic year (difference statistically significant, in favour of control) Lubans 2016 23 No a n/a n/a n/a Mabli 2020 39 No a n/a n/a n/a Madsen 2021 47 No n/a n/a n/a Pbert 2013 44 No n/a n/a n/a Pbert 2016 45 No n/a n/a n/a Robbins 2020 40 No n/a n/a n/a Santos 2014 25 No n/a n/a n/a Smith 2014 22 No a n/a n/a n/a Staiano 2013 46 No n/a n/a n/a Williamson 2012 29 No n/a n/a n/a Wright 2012 28 No n/a n/a n/a Abbreviations: zBMI, standardized body mass index. a Reports movement between weight classification (e.g., proportion of participants who moved from overweight to healthy weight classification).
Two studies only addressed nutrition or eating behaviours 31,41 and five primarily focused on physical activity. [22][23][24]38,40,46 Four studies included individual school-based health clinic visits focused on goal setting, motivational interviewing, and other behaviour change techniques. [42][43][44][45] One of these school-based health clinic studies also incorporated group physical activity sessions. 45 Two studies did not provide direct intervention with youth: One study utilized BMI screening procedures, 47 and the other solely provided materials and technical assistance to school staff. 27 Intervention duration ranged from 5 weeks 37 to 3 years, 36 with the majority (11/25) of studies lasting 6 months [32][33][34][35]38 to an academic year (8-10 months). 25,30,31,42,43,45 Improved weight-based outcomes among the intervention group compared to the control group were reported in 16/26 articles. Table 3 provides information regarding the inclusion of intervention response as an evaluation metric. Response was reported in 19% (5/26) of articles. [32][33][34][35]43 Response was defined in three ways: maintenance or decrease in zBMI at 6 months and 1 and 2 years, 32,33,35 decrease in zBMI of ≥0.10 at 10 months, 43  The infrequent reporting of response among school-based obesity interventions may be due to a lack of consensus in a definition for response to paediatric obesity intervention. Few of the studies in this review defined a weight goal for the intervention, 27,35,45 and only one study explained this goal prior to reporting results. 45 This observation indicates that the a priori weight goals for school-based obesity interventions is often to improve weight-related outcomes more than doing nothing (control group). While this is a first step, the dissemination of programs that are "better than nothing" (i.e., have statistically significant improvements compared to control) but do not achieve meaningful reductions in weight outcomes among a substantial proportion of participants is unlikely to impact the prevalence of obesity.
In the context of community-based and public health settings, a priori intervention goals are important to determine which interventions merit dissemination, use of resources, and to evaluate intervention impact when a control group is not available. Establishing specific goals for an intervention and tracking the proportion of participants who meet the goal would provide a meaningful mechanism to continue the evaluation of intervention programs in the absence of a control group to ensure they are still beneficial. The Endocrine Society identified a 1.5 kg/m 2 decrease in BMI as of important benefit to youth with overweight and obesity and a 7% decrease in weight as a realistic goal for youth with severe obesity. 63 Notably the reasoning behind this definition was not discussed, this type of response definition appears to emphasize the importance of preventing further weight gain, rather than being concerned with a set amount of improvement. Given the challenges of weight maintenance following intervention, a priori goals for follow-up assessment may also need to differ from the primary endpoint of the intervention.
Research regarding how to meaningfully define response to paediatric obesity intervention is clearly needed. Given the complexity of understanding the relationship between weight and cardiometabolic outcomes among youth, it is possible that alternative metrics need to be considered to define meaningful response besides the weight loss needed to see cardiometabolic improvement.
The primary reason for article exclusion was because weightrelated outcomes were not reported separately for participants classified with overweight or obesity from those with a healthy weight classification (57% of excluded articles were excluded because analyses were among a population of youth with a mix weight statuses).
A unique challenge of school-based obesity intervention is that interventions often include children of all weight classifications.
The inclusion of students of all weight statuses prevents potential stigmatization and is feasible because behaviours promoted are similar for both primary and secondary prevention of obesity. However, because the youth of varying weight classifications inherently have different weight-related goals, the definition of response to intervention needs to vary by weight classification. Analysing the weight related outcomes of all participants together regardless of weight status is difficult to interpret and prevents understanding for how school-based interventions may contribute to the secondary prevention of obesity.
This study is the first to examine response definitions and reporting among school-based obesity interventions. This study has been reported according to the PRISMA-ScR, which lends strength to the methodology. Additionally, the inclusion of five databases covering biomedicine, psychology, and education helped to ensure a comprehensive search was conducted for this interdisciplinary topic. Limitations of this review include that the protocol was not registered prior to the study being carried out and the authors were not contacted for more details. Additionally, research records written in languages other than English were excluded due to a lack of resources for translation.