Validity of Impedance Predictions at Various Levels of Fatness

Results of whole body electrical resistance (RES) measure- ments have been proposed as estimates of total body water and of fat-free mass. The validity of RES to predict percent body fat (%BF) was evaluated in a sample of 403 male (Mean values: age: 32.1 yr, stature: 178.6 cm, weight: 87.5 kg, %BF: 21.7%), and 135 female (Mean values: age: 27.1 yr, stature: 164.6 cm, weight: 62.6 kg, %BF: 26.2%) military personnel. There was general over-prediction of individuals having lower %BF values and under-prediction of individuals having higher %BF values using equations supplied by the manufacturer of the RES measurement device. This problem of non-generalizability was not alleviated by 1) re-determination of regres­sion constants using the variables contained in the manufacturer’s equations on this particular sample; 2) incorporation of anthropometric variables in models involving RES and stature; and 3) weighting of the cases to provide equal power at all percent body fat values. Subcutaneous adipose tissue mass was estimated from skinfold thickness and body surface area. The difference between this subcutaneous adipose tissue mass and total fat mass predicted from hydrodensitometry (residual fat) was compared with accepted values for “essential” fat for men and women. In this sample, over-prediction of low %BF individuals occurred at approximately the %BF value at which predicted residual fat becomes less than accepted “essential” fat values. This finding suggests that problems of non-generalizability of equations contain­ing RES values may be associated with violation of the assumptions of the fixed-density, two-compartment model used for conversion of body density values to percent body fat values, which are the criterion measure for most equation development, rather than with the use of RES as a predictor