Introduction: body mass index (BMI) is used worldwide to track excess weight; however, it has limitations in predicting body fat percentage (BF%). Triponderal mass index (TMI) has been studied as an alternative indicator to predict BF%.
Objective: to compare BMI and TMI as predictors of BF% and develop TMI cutoff points for screening high adiposity in Brazilian children and adolescentes.
Methods: a cross-sectional and multicenter study conducted with 226 individuals aged 5 to 17 years from two municipalities in the Southeast and South regions of Brazil. BF% was assessed by bioimpedancemetry. The association between BMI and TMI with BF% was assessed using generalized additive models. ROC curve analyzes were performed to verify the accuracy of BMI and TMI in detecting high adiposity. The areas under the curve (AUC) of BMI and TMI were compared using non-parametric analysis. The TMI cutoff points were obtained using Youden's J index.
Results: bioimpedancemetry detected high BF% in 54.1 % (95 % confidence interval [CI], 44.8-63.2) of boys and 63.5 % (95 % CI, 54.5-71.9) of girls. TMI was able to predict BF% better than BMI in males, presenting a higher R-square (0.737 versus 0.646, respectively). The TMI presented AUC significantly greater than BMI to indicate high adiposity in the entire population (p = 0.007) and in females (p = 0.014). TMI cutoff points for different age groups and sex were presented.
Conclusion: TMI proved to be a better predictor of excess body fat than BMI in Brazilian children and adolescents.