Mueller W H, Marbella A, Harrist R B, Kaplowitz H J, Grunbaum J A, Labarthe D R
School of Public Health, University of Texas Health Science Centre, Houston.
Ann Hum Biol. 1989 Nov-Dec;16(6):495-506. doi: 10.1080/03014468900000642.
The ratios of circumferences (waist/hip, waist/thigh) have been proposed in lieu of skinfold measurements for studies of obesity and body fat distribution in adults. The skinfold method has been used successfully in children to study the growth and development of patterns of body fat distribution, but circumferences have not. We studied the relationship between these two methodologies as indicators of body fat and its anatomical distribution among 365 normal children aged 6-11 years, using canonical correlation analysis. With this method, weighted vectors of four body circumferences on the one hand and five skinfolds on the other are formed in such a way that the correlation between the two sets of variables is maximized. Weights (regression coefficients) are assigned each variable and their strength and sign help us to select the best combination of circumferences which describe a component of centralised obesity. A first canonical correlation was substantial in both boys and girls (0.84) and was independent of age. It appeared to relate to fatness level. A second canonical correlation was low (0.34 in boys, 0.35 in girls) (p less than 0.01). It too was age independent and in both sexes it reflected differences between fat on the trunk and on the lower extremity, and was thus a component of centralised fat distribution. The simple waist/thigh ratio correlated better with this canonical variable (0.67-0.88) than the more commonly used waist/hip ratio (0.45-0.79). The 'best' index of centralised fat in children is therefore, the waist/thigh circumference ratio, the same one that has been suggested for adults.
