The figure outlines the study design. Each of 29 classes (two of the 31 clusters were excluded because they were mixed age classes) was considered as a cluster. Fifteen were randomised to the intervention group and 14 to the control group. At the time of consent, parents and children were unaware of randomisation group. The average class size was 22 (SD 5) children (). In total, 644 of 914 (70.5%) parents and children (320 girls) gave written consent. The average age at baseline was 8.7 (SD 0.9) years (range 7.0 to 10.9 years). Consent was withdrawn for six children (five boys) over the school year. Both groups were similar at baseline for distributions of age, sex, consumption of sweetened carbonated drinks, and percentage overweight or obese ().14
Body mass index was measured in 602 (93.5%) children at six months and 574 (89.1%) at 12 months.
Age, prevalence of overweight and obese children, and mean number of glasses of carbonated drinks consumed daily at baseline
shows the body mass indices, z scores (SDS), and percentage of children above the 91% centile at baseline and 12 months and change in anthropometric measurements over 12 months. The intracluster correlation coefficient for body mass index was 0.01 (95% confidence interval -0.01 to 0.06). After 12 months there was no significant change in the difference in body mass index (mean difference 0.13, -0.08 to 0.34) or z score (0.04, -0.04 to 0.12). At 12 months the mean percentage of overweight and obese children increased in the control clusters by 7.5%, compared with a decrease in the intervention group of 0.2% (mean difference 7.7%, 2.2% to 13.1%; ). Assessing and comparing prevalence between studies is influenced by the methods used to classify overweight children ().
Body mass indices, z scores (standard deviation scores), and mean percentages >91st centile at baseline and 12 months
Mean change in prevalence of overweight and obese children from baseline to follow up at 12 months according to clusters
Prevalence of overweight and obese children (individual analysis). Values are numbers (percentages)
Overall, 55.0% (338 of 615) of the children returned the first drink diary and 56.0% (321 of 574) returned the second; 36% (235) returned both. Body mass indices between those children who returned the diaries and those who did not were similar (17.3 (2.3) v 17.5 (2.4), respectively, P = 0.3 using the t test) Overall, 19.0% of the children who did or did not return diaries at baseline were overweight. Baseline consumption of carbonated drinks was similar between children who did or did not return diaries at 12 months (1.8 v 1.9 glasses, -0.7 to 0.3 glasses).
The intracluster correlation for consumption of carbonated drinks was -0.009 (-0.03 to 0.05), suggesting independence between members of each cluster ().13
At 12 months, consumption decreased in the intervention group compared with the control group (mean difference 0.7, 0.1 to 1.3). Water intake increased in both groups, but there was no difference between intervention and control clusters.
Changes in consumption of drinks over 12 months in control clusters (n=14) and intervention clusters (n=15)*. Values are means (SDs) unless stated otherwise