Background Childhood obesity is a major public health concern worldwide, yet the best way to prevent it remains unknown. Objective(s) To determine what types of intervention strategy are most effective at preventing the development of obesity in children aged 5–18 years, as measured by change in body mass index, and to determine whether interventions work differentially in children with different characteristics associated with inequities. Design Systematic reviews and statistical evidence syntheses. Eligibility criteria Randomised controlled trials of dietary and/or activity interventions that aimed to prevent overweight or obesity in children and young people aged 5–18 years and reported outcomes at least 12 weeks after baseline. Non-randomised evidence was identified through an overview of systematic reviews. Sources of inequity of interest were those defined by the PROGRESS (place of residence, race/ethnicity/culture/language, occupation, gender/sex, religion, education, socioeconomic status, social capital) acronym: place, race/ethnicity, occupation (of parents), gender/sex, religion, education (of parents), socioeconomic status and social capital. Data sources Updating of an existing Cochrane Review, searching bibliographic databases up to February 2023, including MEDLINE, EMBASE, PsycInfo® (American Psychological Association, Washington, DC, USA) and Cochrane Central Register of Controlled Trials on the Cochrane Library, international trial registers and grey literature databases, and examining reference lists. Results subgrouped by inequity factors were sought directly from trialists. Review methods Cochrane Reviews followed standard Cochrane procedures. The main statistical synthesis was informed by a novel analytic framework developed iteratively through discussions with children and young people, schoolteachers and public health professionals. Methodology was developed to analyse the data using multilevel metaregression. To examine the impact of inequity factors, we performed a two-stage meta-analysis of interactions, based on subgroup-level aggregate data collected directly from the trialists. We collected available information on intervention costs. Results We included 172 trials in ages 5–11 and 74 in ages 12–18. In the main synthesis, of 204 trials from both reviews, we found interventions were effective on average (mean difference in standardised body mass index −0.037, 95% credible interval −0.053 to −0.022, which would correspond to a reduction in a proportion of 2.3% with obesity to a proportion of approximately 2.1%). Larger effects were associated with physical activity rather than dietary interventions, after 12 months of follow-up and in the older age group. The overview of non-randomised evidence included 24 systematic reviews, yielding mixed results. The investigation of inequity did not identify substantial differences in effectiveness according to place of residence, race/ethnicity/culture/language, occupation, gender/sex, religion, education, socioeconomic status, social capital characteristics, and there was very limited information about costs. Limitations We were able to examine only the interventions that had been evaluated in studies identified for inclusion in the systematic reviews, which does not cover all possible intervention approaches. Conclusions Interventions to prevent obesity in children aged 5–18 have a small beneficial effect on body mass index on average but with considerable variation. A novel re-analysis of existing randomised trials failed to identify general intervention characteristics driving this variation. No evidence was identified to suggest that interventions increase (or decrease) health inequities. Future work Future studies of the effects of interventions to prevent childhood obesity should routinely collect baseline characteristics around potential inequities. Funding This synopsis presents independent research funded by the National Institute for Health and Care Research (NIHR) Public Health Research programme as award number NIHR131572.
Spiga et al. (Tue,) studied this question.