The ability to accurately estimate body composition is important to practitioners. Bioelectrical Impedance Analysis (BIA), Skinfold Measurements (SM), and Anthropometric Girth Measurements (AGM) are commonly used as field based assessments. These assessments are based on predictive equations that assume fat distributions would be similar for different populations. However, it has been reported that individuals with Down Syndrome (DS) have different body proportions than reference groups used for the predictive equation development. Therefore, the purpose of this study was to cross-validate prediction equations from two different SMs, AGM and BIA for individuals with DS. Participants were 14 individuals with DS and 9 without DS. Skinfold was measured from 7 sites (chest, tricep, midaxillary, abdomen, suprailiac, subscapular, and thigh). Two independent raters measured each site twice and used an average score. The body composition was estimated from two equations: Jackson and Pollock (1978) and Lohman (1981). AGM was estimated using Kelley and Rimmer (1981) equation, which measured circumference of the abdomen, forearm, and buttocks. BAI was measured using the Valhalla body composition analyzer. The criterion measure of body composition is estimated from air-displacement plethysmography using the Bod-PodŽ. The dependent variables were percentage of body fat estimated from each measure and differences between the criterion measurement and the estimated values from the prediction equations. Validity was estimated using Pearson correlation. A MANOVA was employed to examine the accuracy of prediction equations. Correlations between the criterion measurement and the results from prediction equations for the control group ranged from .41 to .94. The results from the SMs and AGM were acceptable ranges .81 to .94, but BMI appears to be troublesome, r=.41. Correlations between the criterion and four prediction equations for individuals with DS ranged from .11 to .54. None of the measures appear to demonstrate an appropriate level of validity evidence. A MANOVA indicated significant differences in accuracy of measurement between control and experimental groups on Jackson and Pollock's and Lohman's prediction equations, F(1,21)=23.14, p<.01 and F(1,21)=10.67, p<.01, respectably. For the control group, Jackson and Pollock's and Lohman's equation had an average of 5.08% and 5.38% of errors, respectively. However, the equation for individuals with DS had an average of 14.90% and 13.20% of errors. Based on the results of this study, a new prediction equation for estimating body composition for individuals with DS is needed.
This research was supported by Northwest Health Foundation (GP no. 2001-238).