BackgroundPreventing childhood obesity remains an important public health issue worldwide. Since visceral fat in particular is understood as an important risk factor for many chronic diseases, waist circumference is recommended as a measurement parameter for global obesity surveillance. This systematic review and meta-analysis focused on waist circumference as an outcome parameter for studies of school-based interventions to prevent overweight and obesity.MethodsA systematic literature search was conducted at the end of 2019 in nine data bases, including Medline and Embase, in order to identify relevant studies evaluating interventions in schools aimed at preventing obesity. Eligibility criteria admitted randomised and non-randomised controlled trials. After screening titles, abstracts and full texts, the data of the identified studies were systematically extracted. Risk of bias was assessed according to study type with the appropriate Cochrane Risk of Bias Tool. The review gives a qualitative overview over all included studies structured by extracted data. Separate meta-analyses were done for the outcome mean difference in change in waist circumference, measured in cm or reported as z-score value, using an inverse variance random-effects model due to study design.ResultsA total of 2421 publications were screened based on titles, abstracts and full texts. Complemented by results of a former systematic literature search 44 studies were identified for inclusion, comprising a total of 39.837 participants (age range: 6 to 18 years). Nearly half of the studies were conducted in Europe, two-thirds combined diet and exercise-based interventions. Likewise two thirds of the studies were conducted as cluster-randomised trials. Most of the reported effects favoured the experimental groups, indicating the basic effectiveness of school-based measures. Based on reported data, only one third of the studies could be included in the meta-analyses. For the difference in mean change of the outcome parameter waist circumference measured in cm (95% CI), we found a pooled effect estimate of -0.95 (-1.87; -0.46). For the difference in mean change of the outcome parameter waist circumference reported as z-score value (95% CI), the pooled effect estimate was -0.10 (-0.15; -0.05). Both effect estimates were in favour of the experimental group. The overall effect sizes were small with a p-value < 0.05.ConclusionsPooled effect estimates were small but in favour of the experimental groups. The same applies to the majority of the effects reported in the included studies. The included cluster of randomised controlled trials demonstrated an especially sound methodological standard. The possibility of achieving larger effects in studies of preventive interventions and health promotion is limited. Schools can only realise their full potential in preventing overweight and obesity in children and adolescents if they are accompanied by measures in other areas of the obesogenic environment.