A six-week programme in Massachusetts after-school clubs set out to get children off sugary drinks. A year later their BMI had edged down against the comparison clubs. Their drinks hadn't: the gap was 20 mL a day

Programmes that try to change everything a child eats and does have a patchy record, partly because they are hard to run and harder to keep running. So a team led from Boston University picked one habit: sugary drinks. Their programme, called H2GO!, was written with the Boys & Girls Clubs of America and delivered by the clubs' own staff after school, to children aged 8 to 13. The results were published in JAMA Network Open on 9 October 2026.

Ten clubs across Massachusetts were matched in pairs by region and size, and one of each pair was picked at random to run the programme. An eleventh joined later, when one comparison club stopped enrolling because it ran short of staff. That left five clubs with 235 children on the programme and six with 202 children on the clubs' usual activities, which included health sessions on things like online safety but nothing on diet. About two in three of the children had free or reduced-price school lunches, 43% were Hispanic or Latino and 16% were Black.

H2GO! was twelve sessions over about six weeks: taste tests, measuring out the sugar in a drink, children writing and telling their own stories about what they drink, activities to take home, and a community event the children ran themselves. Staff got a 90-minute video training. Sessions were meant to last an hour and averaged 45 minutes, and children turned up to half of them.

The main result was BMI z score, measured by study staff with a proper scale and height rod. A z score says how far a child's BMI sits from the average for their age and sex: 0 is average, 1 is roughly the 84th percentile. After 12 months it had gone down by 0.11 in the programme clubs and up by 0.08 in the comparison clubs. Neither of those changes was clear of chance by itself. The difference between them, 0.19, was, with a p value of 0.04 and a confidence interval that runs from 0.38 to 0.01, so it stops just short of zero. In an exploratory count, 28.9% of programme children had moved into a lower weight category by 12 months against 17.8% in the comparison clubs.

Now the drinks, which is where the programme was aimed. Children in the programme clubs said they were drinking 3.66 fluid ounces (about 108 mL) a day less of sugary drinks a year later. Children in the comparison clubs, who had no programme, said they were drinking 3.00 fluid ounces (about 89 mL) less. The difference is 0.67 of a fluid ounce, about 20 mL, four teaspoons of liquid, and nowhere near statistical significance (p = 0.77). Water went up more in the programme clubs, by about 144 mL a day more than in the comparison clubs, but that could also be chance (p = 0.11). The third planned explanation, that children would feel more in charge of their choices, did not hold either: a questionnaire score for that did not change with the programme and did not predict weight.

So the weight result has no visible cause, and there are reasons to look hard at it. The two sets of clubs did not start level. In the programme clubs 43.0% of children had obesity at the start; in the comparison clubs it was 29.7%, and the average z score was 1.0 against 0.8. With only eleven clubs, randomising clubs cannot even that out. Groups that start further from the average tend to drift back towards it. The paper's line chart shows this: the programme clubs come down from roughly 1.07 to 0.96, the comparison clubs go up from roughly 0.80 to 0.89, and after a year the programme children are still the heavier group. When the authors ran a standard analysis that adjusts for each child's starting z score, the result pointed the same way but was no longer statistically significant. The unadjusted result missed too (p = 0.07). It is the model adjusted for household income, parents' education and one question about sports drinks that gets to 0.04.

Dropout was also uneven. On the paper's flow chart, 89 of 235 programme children (38%) were not there at 12 months, against 46 of 202 (23%) in the comparison clubs. The authors report that children who stayed and children who left looked similar at the start, and that results held when they left out the 2020 intake, recruited during the first Covid year, and when they allowed for brothers and sisters. They also say plainly that the drinks findings call for 'cautious interpretation' of how the weight change came about, and suggest children may have cut sugar in food or slept better, neither of which this paper shows.

Our reading: this is a fair test of a cheap, realistic programme in clubs that serve families who are usually hard to reach, and almost every family asked said yes. Its headline result is in the right direction and sits on the edge of chance, in groups that started unequal, with no change in the habit that was meant to drive it. That is a reason for a bigger trial with more clubs, not proof that twelve sessions about soda slim children down. None of it changes the ordinary advice. Water and plain milk are the best everyday drinks for children, and the less sugar in the glass the better, for teeth as much as for weight.