Children at Massachusetts Boys & Girls Clubs that ran a six-week programme called H2GO! ended the year with a lower BMI z score than children at clubs that did not, according to a cluster randomised trial published in JAMA Network Open on 1 October 2026. The difference was small and only just statistically significant, and the programme did not measurably change the drinking habits it was designed to change. For anyone weighing community programmes against the drug options now being discussed for children, it is a modest, carefully registered result with a puzzle at its centre.
What the trial did and what it found
Eleven club sites were matched in pairs and randomly assigned to deliver H2GO! or their usual programming between September 2020 and June 2025. The programme was 12 group sessions over six weeks, run in person by trained club staff, aimed at cutting sugar-sweetened drinks and increasing water through health sessions, stories the children created themselves, and parent-child activities. The trial enrolled 437 children aged 8 to 13 (mean age 10.1) with their parents or caregivers. Two in three (68%) were eligible for free or reduced-price school lunch; 42.6% were Hispanic or Latino, 15.8% Black and 29.5% White. At the start, 16.9% had overweight and 36.8% had obesity.
The primary outcome was the 12-month change in BMI z score, which compares a child’s BMI with children of the same age and sex, calculated from height and weight measured by staff. Over the year the intervention group’s average z score fell by 0.11 (95% confidence interval −0.25 to 0.02) and the comparison group’s rose by 0.08 (−0.04 to 0.20). Neither change was statistically significant on its own. The adjusted difference between the groups was −0.19 (95% CI −0.38 to −0.01; P = .04). That interval runs from a worthwhile effect at one end to almost no effect at the other: an upper bound of −0.01 is close to zero. Self-reported intake of sugar-sweetened drinks and water, the secondary outcomes, did not differ significantly between the groups.
Where it fits, and what is not yet known
The trial was registered in February 2020 as NCT04265794, with BMI z score as the primary outcome and drink intake as secondary, which is how the paper reports it. That matters: the headline result is the one the team committed to before seeing any data, not one picked afterwards. It also sits at the opposite end of the intensity scale from the drug trials in children this site has covered, such as semaglutide tested in children aged 6 to 11. A six-week, staff-delivered club programme costs little and reaches children who are not in a clinic at all. More of this beat is collected on our research news page.
Several things limit what the result can bear. Only 11 sites were randomised, so a single unusual club could move the result, and the confidence interval only just excludes zero. The mechanism is unclear: the drinks the programme targeted did not measurably change, which could mean the one-day recall questionnaire was too noisy to detect a real shift, or that something else differed between sites. Nobody was blinded, and the abstract does not say whether the staff who measured children were the same staff who ran the sessions. The trial ran through the pandemic years, when children’s BMI was rising, in one state, for one year. We could not reach the full paper and wrote this from the published abstract and the registry record, so drop-out and site-level detail are unchecked. The abstract does not convert the z-score difference into kilograms, and we have not either, because that conversion depends on each child’s age and sex.
Sources: Wang ML, Alatorre S, Schwarz AG, et al. A Community-Based Behavioral Intervention and Body Mass Index in School-Aged Children: A Cluster Randomized Clinical Trial. JAMA Network Open, 1 October 2026; 9(10): e2638188. Trial registration: ClinicalTrials.gov NCT04265794, first posted 12 February 2020.
