COMPARISON OF CIRCUMFERENCE-BASED AND SKINFOLD-BASED BODY FAT ESTIMATION EQUATIONS

Abstract

1572 The purpose of this study was to compare body fat estimation equations based on circumference measurements (CIRC) with equations based on skinfold measurements (SF). Both were then compared with equations that combined circumferences and skinfolds (COM). Equations were developed on data from 320 Navy personnel (161 males; 159 females) and cross-validated on data from 60 males and 54 females. Equations were developed separately for males and females using stepwise linear regression with a four-compartment (fat mass, total body water, bone mineral content, and residual lean mass) body composition model as the criterion. The CIRC equation for males included the log10's of the abdomen (AB) and forearm (FA) circumferences. The log10sum of the mid-axillary (MA), medial calf (MC), and supraspinale skinfolds comprised the SF equation. The COM equation contained the log10's of the AB and FA circumferences and the log10 sum of the MA, MC, and iliac crest (IC) skinfolds. R values were 0.921, 0.885, and 0.940 for the CIRC, SF, and COM equations, respectively. Cross-validation R values were 0.865, 0.871, and 0.892. For the females, the CIRC equation included the log10's of the AB, thigh (TH), and shoulder (SH) circumferences. The SF equation included the log10 sum of the triceps, IC, TH, subscapular(SC), AB, MA, and MC skinfolds. The COM equation included the log10 sum of the TH, IC, SC, and MC skinfolds, and the log10's of the AB, SH, and hip circumferences. R values were 0.884, 0.882, and 0.927 for CIRC, SF, and COM, respectively. Cross-validation R values were 0.887, 0.914, and 0.937. Statistical comparisons of R values revealed no significant differences between R values for CIRC and SF equations or between CIRC and COM equations for males. The R for COM was significantly higher than that for SF (p = 0.008). For females, there were no significant differences between CIRC and SF R values, but the R value for COM was significantly higher than both CIRC (p = 0.04) and SF (p = 0.03) R values. In conclusion, the best equation for males or females included both circumferences and skinfolds. However, the equations based on circumferences were as good as the skinfold equations at estimating body fat using a four-compartment body composition criterion.