AUTHOR=Ribeiro da Costa José Ronaldo , Costa Roberto Fernandes da , Goncalves Ciro Alexandre Mercês , de Oliveira Borges Michelle Vasconcelos , Almeida-Neto Paulo Francisco De , De Assis Gilmara Gomes , Cabral Breno Guilherme De Araujo Tinoco , Dantas Paulo Moreira Silva TITLE=The Body Adiposity Index is not applicable to the Brazilian adult population JOURNAL=Frontiers in Nutrition VOLUME=9 YEAR=2022 URL=https://www.frontiersin.org/journals/nutrition/articles/10.3389/fnut.2022.888507 DOI=10.3389/fnut.2022.888507 ISSN=2296-861X ABSTRACT=Background

Obesity is a serious disease that burdens public health systems around the world. It is a risk factor for the development of several non-communicable chronic diseases that are related to the amount and distribution of body fat. Body composition assessment using simple and low-cost techniques can help in the early detection of excess fat, allowing for the prevention and treatment of both obesity and associated diseases. Thus, identifying and proposing valid anthropometric indices for this purpose can be a great ally of health programs.

Objective

To verify the validity of the Body Adiposity Index (BAI) in relation to Dual Energy X-Ray Absorptiometry (DXA) for estimating body fat percentage in Brazilian adults, as well as to propose a new mathematical model to estimate the fat-free mass of this population.

Methods

In a cross-sectional study, 424 subjects (of which 220 were women), aged between 20 and 59 years, were evaluated by BAI and DXA, then randomly divided into two groups stratified by sex: the development group (n = 283) and the cross-validation group (n = 141). Statistical analyses to test the validity of BAI as a predictor of fat mass, in addition to proposing a new mathematical model for estimating fat-free mass, using DXA as a reference method. The analysis included paired t-test, stepwise multiple regression, coefficient of concordance correlation, and Bland-Altman plots.

Results

The BAI validity analysis showed a low correlation coefficient of agreement [CCC = 0.626; ρ (precision) = 0.795; Cb(accuracy) = 0.787]; in addition, the mean difference in the Bland-Altman plot was different from zero in the cross-validation group (p < 0.01) and limits of agreement (LOA) ranged between−8.0 and 14.4 kg, indicating a poor agreement between the BAI and the reference method. The new mathematical model for estimating FFM showed a high correlation coefficient of agreement (CCC = 0.952; ρ = 0.953; Cb = 0.999), in addition to acceptable LOA in the Bland-Altman plot (-6.7 and 6.7).

Conclusion

In the studied sample, the BAI showed low validity for estimating body fat, while the new proposed model was found to be a good option to assess the body composition of Brazilian adults.