The relation between body mass index and body fat percentage in Brazilian adolescents: assessment of variability, linearity, and categorisation.

Reliable but accessible measures to capture body composition are highly important as the world is in an era of obesity-increase. The most used measure, the body mass index (BMI), nevertheless, has been judged as non-reliable to estimate body fat percentage (BF%). The present manuscript assessed the criticisms of BMI as a predictor of BF% and the BMI-based categories of nutritional status. 4164 children/adolescents from 9 to 18 years of age from Porto Velho, Brazil, had their anthropometric and skinfold measures taken. Controlling for socioeconomic status, school (private, public), sex and age, we compared proposed models/variables in the literature relating BMI and BF%. We evaluated the functions and the residual data to understand the variability of BF% estimate per BMI and evaluated three possible categorisations from BMI to predict BF% nutritional status. The function utilising (linear) BMI was the best to predict BF% (R2 = 0.70) with a variability of only 6.49% around the function. Nevertheless, no categorisation of nutritional status was reliable to predict the nutritional status of individuals. BMI is reliable to estimate BF%. Nevertheless, new normative values must be proposed; the original categorisation fails to capture the nutritional status of children/adolescents from this region.