This association was independent of the affluence of the country and of similar magnitude in boys and girls. By contrast, phase 3 adolescents who self-reported that they frequently ate fast-foods tended to have lower BMIs, with the exception of male adolescents from low-GNI countries, where there was no association

between fast-food consumption and BMI. We have also found that up to 25% of children worldwide consume fast-food frequently or very frequently, and this increases to over 50% in the adolescent age group. Children Our results in children are consistent with those of Shan et al22 who found that ‘Western fast-food’ and ‘snack food’ were associated with overweight and obesity among children aged 2–18 years in Beijing. However, in New Zealand, Duncan et al20 found that low levels of physical activity, skipping breakfast and insufficient sleep on weekdays were associated with increased adiposity in 5–11-year-olds, but that fast-food consumption was not. The magnitude of the association between fast-food consumption and BMI in children was small (mean BMI values 0.15 and 0.22 kg/m2 higher in frequent and very frequent consumers compared to infrequent consumers), and the clinical significance of this is uncertain. Given the long-term consequences of overweight and obesity in childhood, even a small change in mean BMI in a population could be

of major public health significance. An observational study such as this cannot attribute causality. The association we have found between fast-food consumption and BMI in children could be due to other specific dietary factors that have been shown to be associated with fast-food consumption, such as higher fat intake, greater consumption of sugary drinks, fewer fruits and non-starchy vegetables.11 Alternatively, fast-food consumption may be a marker of other factors that influence BMI such as parental BMI, individual socioeconomic circumstances

or patterns of activity and inactivity. Adolescents In our analysis, self-reported frequent fast-food consumption was associated with a lower BMI in adolescents, with stronger association in higher GNI countries. As far as we are aware, no other studies have demonstrated a lower BMI with higher fast-food consumption in adolescence and our findings appear to be in stark contrast to other research findings GSK-3 from high-GNI countries. One study showed that increasing fast-food consumption in American adolescents was associated with increased weight gain from adolescence to adulthood.12 Another study found that American adolescents who consumed greater quantities of fried food away from home were heavier,23 and, in a further American study, increasing frequency of eating quick-service food was associated with increasing z-BMI in female adolescents.