Muckelbauer 2010.

Study characteristics
Methods	Study design: cluster‐RCT Intervention period: 11 months Follow‐up period (post‐intervention): nil Differences in baseline characteristics: reported Reliable outcomes: reported Protection against contamination: reported Unit of allocation: city Unit of analysis: individual accounting for clustering by school
Participants	N (control baseline) = 1839 N (control follow‐up) = 1309 N (intervention baseline) = 1978 N (intervention follow‐up) = 1641 Setting (and number by study group): elementary schools in deprived areas (16 control and 17 intervention schools) Recruitment: NR Geographic region: 2 neighbouring cities, Dortmund and Essen, Germany Percentage of eligible population enrolled: random sample of schools, 84% of 3817 children attending the participating schools, with a higher rate in the intervention group (88%) than in the control groups (80%; P = 0.004) Mean age: intervention: 8.26 ± 0.73; control: 8.34 ± 0.76 Sex: intervention: 49.8% female; control: 49.7% female
Interventions	To test whether a combined environmental and educational intervention solely promoting water consumption was effective in preventing overweight among children in elementary school In each intervention school, 1 water fountain, or 2 for schools with > 150 participants, was installed. The fountains provided cooled, filtered, plain or optionally carbonated water. In addition, each child received a plastic water bottle (500 mL), and teachers were encouraged to organise filling of the water bottles each morning for all children in the corresponding classes. The educational intervention consisted of four 45‐min classroom lessons dealing with the water needs of the body and the water circuit in nature. At the beginning of the study, teachers received a booklet with the prepared curriculum and necessary materials to implement the lessons in the formal school curriculum. The lessons were developed by using the results of empirical teaching research and were intended to improve the constructs of intention, attitudes, and perceived behavioural control, on the basis of the theory of planned behaviour. 3 months after the beginning of the study, teachers introduced a motivation unit (i.e. booster sessions) that used a goal‐setting strategy to reach a sustained increase in water consumption by giving quantitative targets and feedback. In month 5 after the baseline assessment, each participant received a new water bottle with an improved handling design. Control schools did not receive any intervention. Diet (water consumption only) intervention vs control
Outcomes	Outcome measures Primary outcome: overweight prevalence Secondary outcomes: BMI SD scores Process evaluation: reported (implementation)
Implementation‐related factors	Theoretical basis: TPB Resources for intervention implementation: reported Who delivered the intervention: reported PROGRESS categories assessed at baseline: gender PROGRESS categories analysed at outcome: NR Outcomes relating to harms/unintended effects: reported Intervention included strategies to address diversity or disadvantage: NR, intervention only included deprived schools Economic evaluation: reported: costs
Notes	NCT00554294 Funding: this study was supported by grant no. 05HS026 of the German Federal Ministry of Food, Agriculture, and Consumer Protection. Intervention material (water fountains, bottles, print of the lesson booklet) was provided by the Association of the German Gas and Water Industries. The initial costs per water fountain were EUR 2500 and the long‐term costs per enrolled child were EUR 13/year. The educational intervention was presented by the teachers; therefore, no additive costs emerged.
Risk of bias
Bias	Authors' judgement	Support for judgement
Random sequence generation (selection bias)	Unclear risk	Randomisation was performed at the city level to minimise contamination between neighbouring schools in 1 city, no further details
Allocation concealment (selection bias)	High risk	NR. However the model to test for intervention effects on the primary outcome prevalence of overweight at the follow‐up assessment included significant confounders, besides the fixed intervention effect, although randomisation was conducted
Blinding (performance bias and detection bias) All outcomes	High risk	Blinding of outcome assessment for BMI not reported. Participants not blind Quote: "participants were aware of the behavioural intervention aim."
Incomplete outcome data (attrition bias) All outcomes	Low risk	Of 3190 children screened at baseline, a total of 2950 children (92%) were also measured at the follow‐up assessment and were considered for analysis. Dropouts (N = 240) were similar to analysed participants with respect to the prevalence of overweight (24.6% vs 24.5%; P = 0.741), mean BMI SDs (0.26 vs 0.26; P = 0.807), mean age (8.27 vs 8.30 years; P = 0.574), proportion of boys (50.4% vs 50.2%; P = 0.772), and proportion of children with migrational background (42.1% vs 44.3%; P = 0.568). > 30% loss to follow‐up ‐ therefore high risk?
Selective reporting (reporting bias)	Low risk	Protocol/trial registration document seen. All outcomes reported
Other bias	Low risk	No additional threats to validity
Other bias‐ timing of recruitment of clusters	Low risk	Figure indicates recruitment happened prior to randomisation