Accurate assessment of body composition in obese females.

Although several generalized regression equations exist for body composition assessment the accuracy of these equations for special populations is uncertain. This study examined the ability to predict body composition variables from girth measurements in obese women. Girth measurements were taken on 156 obese women at abdomen 1 and abdomen 2 (mean value of the two was used), buttocks, and right thigh. Stepwise multiple-regression analysis generated on 110 randomly assigned subjects yielded the following equations: density (g/cc) = -0.00020(mean abdomen) - 0.00032(wt) + 0.00039(ht) + 0.97783 (R = 0.76, SEE = +/- 0.00061); % fat = 0.11077(mean abdomen) - 0.17666(ht) + 0.14354(wt) + 51.03301 (R = 0.76, SEE = +/- 2.9); fat wt (kg) = 0.62039(wt) - 0.17844(ht) + 0.09495(mean abdomen) + 5.88874 (R = 0.96, SEE = +/- 2.9); and lean body wt (kg) = 0.37939(wt) + 0.17898(ht) - 0.09494(mean abdomen) - 6.00423 (R = 0.89, SEE = +/- 3.0). Data from the remaining 46 women were used for cross-validation. These equations were valid, with validity coefficients for fat weight and lean body weight of r = 0.92 (SE = +/- 3.3 kg) and r = 0.86 (SE = +/- 3.3 kg), respectively, with no significant mean differences between actual and predicted values. The use of girth measurements is a simple and practical method of estimating fat weight and lean body weight in obese women.