Beyond BMI in Childhood and Adolescence: Limitations and Alternatives

Accurate assessment of body size in growing children is a central challenge in pediatrics, with implications for both clinical care and public health. The body mass index (BMI), though widely used, was developed for adults, where body weight scales approximately with the square of height. In adults, this scaling reflects stable skeletal proportions after growth, allowing BMI to function as a height-adjusted indicator. In children, however, the weight–height relationship shifts systematically with age, sex, and maturation. Thus, the conventional definition of a index – minimizing correlation with height – fails to remove growth-related bias. Even if the BMI exponent in weight/height is tuned to eliminate correlation, misclassification of thinness or obesity may still occur. We analyzed 7.86 million Japanese schoolchildren aged 5–17 years using an allometric framework and identified two distinct regimes. In the regime, observed after skeletal maturity (16 years in males, 14 in females), all weight-for-height centile curves share a common exponent (), producing BMI distributions invariant to height. In the regime, characteristic of younger children, exponents differ across centiles, causing BMI to overestimate obesity in taller children and underestimate it in shorter ones. To address this, we introduce – requiring not only independence of the mean or median, but invariance of the entire distribution across heights – as the criterion for valid adjustment. For multi-scaling ages, centile-specific methods or direct use of weight centiles by sex, age, and height may offer a more accurate and interpretable basis for pediatric assessment.