Accuracy and concordance of anthropometry for measuring regional fat distribution in adults aged 20-55 years

Objectives: This study aimed to assess the accuracy and concordance of anthropometrically derived prediction equations for the estimation of regional fat mass (FM) distribution. Methods: Sixty‐two white males and 50 females with a large range of age (20−55 years) and BMI (16.6−33.4 kg/m2) were included. Whole body dual energy X‐ray absorptiometry (DXA) scans were acquired and anthropometric prediction equations for regional FM were developed and cross‐validated. On the basis of the total sample two anthropometrically derived indices of FM distribution $\Big({{{\rm FM}_{{\rm trunk}} } \over {{\rm FM}_{{\rm limbs}} }}$ ratio and ${{{\rm \%FM}_{{\rm trunk}} } \over {{\rm \%FM}_{{\rm legs}} }} \ {\rm ratio}\Big)$ were compared with their DXA analogues. Results: In both sexes multiple linear regression models predicted the regional DXA fat masses with good accuracy (P < 0.001). In men mean bias (limits of agreement) were: −6.8 g (−535,364) for FMarms, 65 g (−1921,2052) for FMtrunk, −21 g (−1374,1332) for FMlegs, −0.2% (−5.0,4.7) for %FMtrunk and −0.5% (−6.8,5.8) for %FMlegs. In women mean difference (limits of agreement) were: −86 g (−463,450) for FMarms, 30 g (−1784,1844) for FMtrunk, −278 g (−1782,1227) for FMlegs, 0.4% (−5.5,6.3) for %FMtrunk, and 0.3% (−8.3,8.9) for %FMlegs. No systematic (constant and proportional) differences between methods for the determination of FM distribution ratios were found, suggesting method interchangeability. The concordance for subject classification based on t‐scores according to the National Health and Nutrition Examination Survey (NHANES) was significant (P < 0.001), with substantial agreement for ${{{\rm FM}_{{\rm trunk}} } \over {{\rm FM}_{{\rm limbs}} }}$ ratio (κw = 0.80) and ${{{\rm \%FM}_{{\rm trunk}} } \over {{\rm \%FM}_{{\rm legs}} }}$ ratio (κw = 0.75). Conclusion: Anthropometric variables offer promise to the development of simple, noninvasive, and inexpensive screening to identify individuals with abnormal FM distribution. The anthropometrically derived indices of FM distribution demonstrate sufficient accuracy for clinical use. Am. J. Hum. Biol., 2013. © 2012 Wiley Periodicals, Inc.