Validation of field estimates of body composition against the 4-compartment Gold Standard with homogeneous Japanese males

Abstract

Field body composition models are derived for specific populations using regression analysis. In order to generalize these field models to different populations, it is necessary to validate the field models against appropriate validated reference measures. The purpose of this study was to use a group of homogenous Japanese males to validate a number of field estimates, as well as two-compartment (2C), and three compartment (3C) laboratory models against the “gold standard”, four-compartment (4C) reference model. Skinfold measurements, underwater weighing (UWW), Dual-Energy X-ray Absorptiometry (DXA), and total body water (deuterium dilution) data were collected on 20 ethnically Japanese males between the ages of 18 and 45. Skinfold measurements (Jackson & Pollock 3- and 7-site, Nagamine, Kagawa, Minematsu) were used to estimate body fat percentage (%BF). Models were compared using Spearman’s rho, one sample Wilcoxon Signed Rank Test, and Bland-Altman plots. Estimated %BF for the participants ranged from 16.1±7.1% to 23.7±8.0% with the 4C reference at 18.5±8.0%. All 2C and 3C models were significantly correlated with the 4C model (Ρ=0.707 to 0.997, p<0.01 for all), but there were significant differences between the 4C reference and the Jackson & Pollock 3-site, Kagawa, DXA, and Siri 3C models (p<0.03 for all). Results of this study indicated that the Jackson & Pollock 7-site skinfold equation was the most valid field model, and the Siri 3C model was the most valid the laboratory model when compared to the Lohman 4C Gold Standard reference model for Japanese males.