Oppliger R A, Clark R R, Nielsen D H
Physical Therapy Graduate Program, The University of Iowa, Iowa City 52242-1997, USA.
J Orthop Sports Phys Ther. 2000 Sep;30(9):536-43. doi: 10.2519/jospt.2000.30.9.536.
Methodologic study to derive prediction equations for percent body fat (%BF).
To develop valid regression equations using NIR to assess body composition among high school wrestlers.
Clinicians need a portable, fast, and simple field method for assessing body composition among wrestlers. Near-infrared photospectrometry (NIR) meets these criteria, but its efficacy has been challenged.
Subjects were 150 high school wrestlers from 2 Midwestern states with mean +/- SD age of 16.3 +/- 1.1 yrs, weight of 69.5 +/- 11.7 kg, and height of 174.4 +/- 7.0 cm. Relative body fatness (%BF) determined from hydrostatic weighing was the criterion measure, and NIR optical density (OD) measurements at multiple sites, plus height, weight, and body mass index (BMI) were the predictor variables.
Four equations were developed with multiple R2s that varied from .530 to .693, root mean squared errors varied from 2.8% BF to 3.4% BF, and prediction errors varied from 2.9% BF to 3.1% BF. The best equation used OD measurements at the biceps, triceps, and thigh sites, BMI, and age. The root mean squared error and prediction error for all 4 equations were equal to or smaller than for a skinfold equation commonly used with wrestlers.
The results substantiate the validity of NIR for predicting % BF among high school wrestlers. Cross-validation of these equations is warranted.
用于推导体脂百分比（%BF）预测方程的方法学研究。
使用近红外光谱法（NIR）开发有效的回归方程，以评估高中摔跤运动员的身体成分。
临床医生需要一种便携、快速且简单的现场方法来评估摔跤运动员的身体成分。近红外光谱法（NIR）符合这些标准，但其有效性受到了质疑。
研究对象为来自美国中西部两个州的150名高中摔跤运动员，平均年龄±标准差为16.3±1.1岁，体重为69.5±11.7千克，身高为174.4±7.0厘米。通过水下称重法测定的相对体脂率（%BF）为标准测量指标，多个部位的近红外光密度（OD）测量值，加上身高、体重和体重指数（BMI）为预测变量。
开发了四个方程，其复相关系数（R²）在0.530至0.693之间，均方根误差在2.8%BF至3.4%BF之间，预测误差在2.9%BF至3.1%BF之间。最佳方程使用了二头肌、三头肌和大腿部位的OD测量值、BMI和年龄。所有四个方程的均方根误差和预测误差均等于或小于摔跤运动员常用的皮褶厚度方程。
结果证实了近红外光谱法（NIR）在预测高中摔跤运动员体脂百分比（%BF）方面的有效性。有必要对这些方程进行交叉验证。
