Anthropometric multicompartmental model to predict body composition In Brazilian girls

Abstract

Background: Anthropometric models remain appropriate alternatives to estimate body composition of peripubertal populations. However, these traditional models do not consider other body components that undergo major changes during peripubertal growth spurt, with restrictions to a multicompartimental approach as a quantitative growth. DXA has great potential to determine pediatric body composition in more than one component (3-C), but has limited use in field settings. Thus, the aim of this study was to propose and validate an anthropometric model for simultaneous estimation of lean soft tissue (LST), bone mineral content (BMC) and fat mass (FM) in healthy girls, from a multivariate approach of densitometric technique, as the criterion method. Methods: A sample of 84 Brazilian girls (7-17 years) was defined by chronological age and maturity offset. Whole total and regional DXA body scan were performed and, the components were defined (LST, BMC and FM) and considered as dependent variables. Twenty-one anthropometric measures were recorded as independent variables. From a multivariate regression, an anthropometric multicompartmental model was obtained. Results: It was possible to predict DXA body components with only four predictive measurements: body weight (BW); supra-iliac skinfold (SiSk); horizontal abdominal skinfold (HaSk) and contracted arm circumference (CaCi) with high coefficients of determination and low estimation errors (LST = 0.6662657 BW -0. 2157279 SiSk -0.2069373 HaSk + 0.3411678 CaCi -1.8504187; BMC = 0.0222185 BW -0.1001097 SiSk -0.0064539 HaSk -0.0084785 CaCi + 0.3733974 and FM = 0.3645630 BW + 0.1000325 SiSk -0.2888978 HaSk -0.4752146 CaCi + 2.8461916). The cross-validation was confirmed through the sum of squares of residuals (PRESS) method, presenting accurate coefficients (Q(PRESS)(2) from 0.81 to 0.93) and reduced error reliability (S-PRESS from 0.01 to 0.30). Conclusions: When sophisticated instruments are not available, this model provides valid estimates of multicompartmental body composition of girls in healthy Brazilian pediatric populations.