Trocki O, Baer D J, Castonguay T W
Department of Nutrition and Food Science, University of Maryland, College Park 20742, USA.
Physiol Behav. 1995 Apr;57(4):765-72. doi: 10.1016/0031-9384(94)00325-4.
Total body electrical conductance (TOBEC) has been recommended for serial measurements of body composition in animals and humans. This study examined the accuracy of the TOBEC technique in predicting body composition of a population of adult male rats that had undergone seven different treatments, including adrenalectomy and blocking of glucocorticoid receptors, in the study of the etiology of obesity. The predicted body composition values of the animals (n = 57, body weight 550 +/- 8 g) obtained by using the manufacturer's and Baer's equations were compared to the actual body composition obtained by direct carcass analysis. Both equations underestimated lean body mass and reciprocally overestimated body fat (manufacturer's 103 +/- 4 g, Baer's 55 +/- 3 g). A new prediction equation was developed based on the conductivity index and the actual lean body mass. This revised equation was able to accurately estimate the lean body mass of the animals used in the same experiment but over-estimated lean body mass of larger animals (n = 10, wt. 647 +/- 13 g). Conclusions based on multiple comparisons (Duncan's) of predicted and actual values resulted in different effects of treatments on body composition. To improve accuracy and reliability of the TOBEC technique, a prediction equation should be developed from the same population as the studied population, and experimental group sizes used for examining treatment effects should be relatively large.
总体电导率(TOBEC)已被推荐用于对动物和人类身体成分进行连续测量。本研究在肥胖病因学研究中,检验了TOBEC技术在预测经历了七种不同处理(包括肾上腺切除术和糖皮质激素受体阻断)的成年雄性大鼠群体身体成分方面的准确性。将使用制造商方程和贝尔方程获得的动物(n = 57,体重550 +/- 8 g)预测身体成分值与通过直接胴体分析获得的实际身体成分进行比较。两个方程均低估了瘦体重,相应地高估了体脂(制造商方程为103 +/- 4 g,贝尔方程为55 +/- 3 g)。基于电导率指数和实际瘦体重建立了一个新的预测方程。这个修订后的方程能够准确估计同一实验中所用动物的瘦体重,但高估了较大动物(n = 10,体重647 +/- 13 g)的瘦体重。基于预测值与实际值的多重比较(邓肯法)得出的结论显示,不同处理对身体成分有不同影响。为提高TOBEC技术的准确性和可靠性,应从与研究群体相同的群体中建立预测方程,并且用于检验处理效果的实验组规模应相对较大。