Abstract
BACKGROUND: Most authorities define childhood overweight/obesity as a BMI exceeding the same high centile cutoff at all ages, but it seems unlikely that true obesity prevalence (excess body fat) is constant throughout childhood. OBJECTIVES: We investigated how fat mass (FM) and lean mass (LM), adjusted for height, relate to BMI and each other across childhood, using a uniquely large database of body composition measures, estimated using gold standard methods. METHODS: Cross-sectional and cohort data were collated from representative samples of healthy children aged 6 wk to 20 y and children attending obesity clinics aged 7-16 y. Body composition was measured by deuterium dilution up to age 4 y, and by either deuterium or the criterion 4-component model from 4 to 20 y. FM and LM were expressed, respectively, as fat mass index (FMI; FM/height2) and lean mass index (LMI; LM/height2). RESULTS: There were 2367 measurements of weight, height, and body composition from 1953 individuals. Before age 6 y, the variability in FMI, LMI, and BMI was much less than after; FMI was low (mainly <8 kg/m2) and FMI and LMI were weakly negatively correlated. From mid-childhood, upper limits for both BMI and FMI rose, but FMI in children with BMI <91st centile still rarely exceeded 8. With increasing age, the correlation of FMI with LMI rose to 0.5-0.7, driven mainly by children with a high FMI also having a high LMI. CONCLUSIONS: Raised fat levels are much less common at younger than older ages, and young children with a high BMI centile have lower FMI than older children with the same BMI centile. Current BMI centile cutoffs thus overdiagnose obesity in younger groups. More stringent cutoffs are required for children aged <6 y, matching the WHO recommendation for 0-5 y.