Development and validation of a new anthropometric equation to predict fat mass percentage in a heterogeneous Caucasian population

Abstract

Objective: (1) To develop a new regression equation for estimating fat mass percentage (%FM) from anthropometric measurements in a heterogeneous Caucasian population and (2) to compare it with the Durnin and Womersley equation, which is one of the most used anthropometric equations for FM assessment. Design: Body mass, stature and four skinfolds (biceps, triceps, subscapular and supracrestal) were assessed by an accredited anthropometrist, according to the International Society for Advancement in Kinanthropometry. Participants completed a dual-energy X-ray absorptiometry (DXA) whole-body scan to determine their %FM. A new anthropometric equation to estimate %FM was developed using multiple forward regression analyses with DXA as the reference method. Tests for the accuracy of the different equations included mean differences, coefficient of determination, SE of the estimate (SEE), concordance correlation coefficient (CCC) and Bland-Altman plots. Setting: Spain. Participants: Two hundred and eighteen healthy Caucasian participants aged 18-65 years participated in this cross-sectional study. Results: Our proposed equation explained 89·9 % of the variance in the DXA-derived %FM, with a low random error (SEE = 3·00 %), a very strong agreement (CCC = 0·93), no fixed or proportional bias and a relatively low individual variability (5·84 %). However, the Durnin and Womersley equations obtained a fixed bias of -3·65 % when compared with DXA and a greater individual variability (6·74 %). Conclusions: The proposed equation can accurately estimate %FM in a heterogeneous Caucasian population with a wide age range (18-65 years). Additionally, the Durnin and Womersley equation was inadequate when applied to our participants.