Too early to ditch the campaign

We believe that James et al's conclusion that an intervention to reduce children's consumption of carbonated drinks and prevalence of overweight was not effective two years after completion¹ is not warranted for two reasons.

Firstly, they base their conclusions on the proportion of overweight children, which was significantly different between the two groups at 12 months, but not at three years. However, average values of body mass index (BMI), Z score, and waist circumference would be better outcome measures.^{2 3}

Changes in BMI, Z score, and waist circumference moved towards significance. Thus, the intervention did not have a significant effect on overweight after 12 months, but it was moving in the right direction.

If the children continued to consume fewer carbonated drinks as a result of the intervention, they would put on less weight each year. James et al should measure the children's BMIs in a few years—they may find the desired significant results.

Secondly, the trial characteristics were flawed. The trial was powered to detect differences in consumption of carbonated drinks, not proportion of overweight. Power was further reduced by loss to follow-up at three years. Large (though not significant) differences occurred at baseline—average BMI, Z score, and waist circumference were lower in the intervention group than in controls.

Childhood obesity and how to tackle it is a huge problem, with few solutions. James et al dismiss what could be a promising result, on the basis of an inappropriate outcome measure from an insufficiently powered and poorly randomised trial.