Résumé
A study was conducted to determine the accuracy of body composition prediction equations using the bioelectrical impedance (BI) method and skinfold thickness (ST) method for predicting longitudinal alterations in the body composition of schoolchildren. Seventeen healthy junior high school boys, aged 12 to 13 yr, participated in the study. Body density (Db) was determined by underwater weighing (UW) . Impedance was measured using a portable four-terminal impedance plethysmograph (800 μA, 50 kHz ; Selco, SIF-891) . Db values by the BI method were estimated from the equations developed for schoolchildren by Watanabe et al. (1993) and Kim et al. (1993) . Db by the ST method was estimated from the equations developed for schoolchildren by Nagamine et al. (1974) and Watanabe et al. (1993), using the sum of skinfold thickness at the triceps and subscapular area. Skinfold thickness was measured on the right side of the body with an Eiken-type skinfold caliper. The changes in Db and body fat measured by UW were small every year. The fat free mass (FFM) measured by UW increased significantly every year. The body composition (average Db, body fat and FFM) values estimated by the BI and ST methods showed no significant differences against average UW values. However, these average values estimated using the equation of Nagamine et al. (1974) were significantly different from the average UW values. The absolute amount of change in FFM (Δ FFM) estimated from BI and ST methods were correlated significantly with absolute amount of change in FFM (Δ FFM) determined by UW. In particular, the accuracy of body composition prediction equations for the BI method (r<SUP>2</SUP>= 0.81 for Watanabe et al., r<SUP>2</SUP> = 0.77 for Kim et al.) was higher than that of body composition prediction equations for the ST method (r<SUP>2</SUP>=0.41 for Nagamine et al., r<SUP>2</SUP>= 0.55 for Watanabe et al) . The findings of this study suggest that the body composition prediction equations based on the BI method are useful for valid assessment of longitudinal alterations in the body composition of schoolchildren.
Résumé
The tetrapolar bioelectrical impedance (BI) method has been proposed as a convenient, valid approach for estimating the body composition of normal healthy adults. However, the validity of the BI method has not yet been confirmed for Japanese junior high school boys and girls. The purpose of this study was to develop convenient and useful equations for predicting the body composition in junior high school boys and girls by the BI method. The subjects were 297 healthy boys and girls, aged 12.15 years, all of whom were Japanese. Impedance was measured using a tetrapolar bioelectrical impedance plethysmograph (800 pA, 50 kHz SIF-891) manufactured by Selco. Multiple regression analysis was used to derive prediction equations for Db that were specifically applicable to boys and girls. The effective prediction equations for Db were as follows : 1) Db=1.1860-0.1282 (Wt·Z) /Ht<SUP>2</SUP>, and 2) Db=1.1402-0.0706 (Wt·Z) /Ht<SUP>2</SUP>-0.0007· (abdomen) for boys. 1) Db=1.1337-0.0778 (Wt·Z) /Ht<SUP>2</SUP>, and 2) Db=1.1124-0.0498 (Wt·Z) /Ht<SUP>2</SUP>-0.0006· (subscapular) for girls, where Db=body density (g/ml), Wt=weight (kg), Z =impedance (ohms), Ht=height (cm) . Db estimated by each respective equation was highly correlated with body density measured by underwater weighing (UW-Db) : 1) r=0.881, SEE=0.00868/ml, 2) r=0.902, SEE=0.00788/nil for boys and 1) r= 0.741, SEE=0.0101 g/ml, 2) r=0.775, SEE =0.0095g/ml for girls. Furthermore, in a cross-validation analysis of prediction equations for Db, another sample consisting of 40 boys and 66 girls was used. Db estimated from each respective equation was correlated highly with UW-Db : 1) r=0.856, 2) r=0.887 for boys and 1) r=0.837, 2) r=0.860 for girls. There were no significant differences between the mean Db obtained by the BI method and that by the criterion method. We suggest that the prediction equations proposed in this study are useful for valid assessment of body composition of Japanese junior high school boys and girls aged 12 through 15 years.
Résumé
A study was conducted to investigate the validity of skinfold-based prediction equations for body density (Db, g/m<I>l</I>) developed by Nagamine et al. (1974), and to formulate convenient, useful equations for predicting Db by the skinfold thickness (ST) method in junior high school boys and girls. The subjects of the study were 269 healthy boys and girls, aged 12-15 years. The dependent variable, Db, was determined by underwater weighing (UW) . Independent variables included single skinfold thickness at three sites (triceps, subscapular and abdomen) and the sum of two skinfolds. Db by the ST method was estimated from the equations developed by Nagamine et al. (1974) for boys and girls, using the sum of skinfold thickness at the triceps and subscapular area. Skinfold thickness was measured on the right side of the body with an Eiken-type skinfold caliper. Db estimated by the ST method was correlated significantly with Db determined by UW (r=0.873 for boys and r=0.723 for girls) . However, average Db values estimated by the ST method were significantly lower than those deter-mined by UW (differences in Db values when predicted by the Nagamine equations: 0.0099 for boys and 0, 0114 for girls) . Therefore, we developed linear regression equations for predicting Db. The best-fitting prediction equation for Db was Db=1.0881-0.0010·X for boys, and Db=1.0715-0.0007·X for girls, where X is the sum of the triceps and subscapular skinfold thickness (mm) for boys and girls. Db estimated from the respective equation was correlated significantly with hydrodensitometrically determined Db (r=0.872, SEE=0.0089 for boys; r=0.722, SEE=0.0104 for girls) .<BR>Furthermore, in a cross-validation analysis of prediction equations for Db developed in the present study, Db estimated from the respective equation was correlated highly with hydrodensitometrically determined Db (r=0.887 for boys and r=0.740 for girls) . There were no significant differences between the Db values predicted by the ST method against hydrodensitometrically determined Db values (difference values: 0.0012 for boys and 0.0013 for girls) . The final phase of this study was to develop more stable equations, combining validation and cross-validation samples. On the basis of the final analyses, we recommend the equations Y=1.0875-0.0010X and Y=1.0716-0.0007X, with SEE of 0.0088g/m<I>l</I> for boys and 0.0105g/m<I>l</I> for girls, respectively. It is suggested that the prediction equations finally developed in the present study will be applicable to junior high school boys and girls.
Résumé
Several prediction equations for estimating body composition of Japanese men and women have recently been developed using a linear regression model with a combination of impedance and anthropometric measurements as independent variables. The purpose of this study was to determine the cross-validity of body density (Db) estimated from bioelectrical impedance (BI) and skinfold thickness (ST) methods in comparison with underwater weighing (UW) as a criterion reference method. Percentage body fat (%BF) was derived from Db according to the equation Brozek et al. Fifty-seven healthy Japanese women, aged 19 to 57 years, volunteered to participate in the study. Impedance was measured by use of a portable four-terminal impedance plethysmograph (Selco, SIF-891) . %BF derived from the BI method (r=0.860-0.875) was correlated with hydrodensitometrically determined %BF to a greater extent than %BF obtained using the ST method (r=0.7330.758) or ultrasound method (r=0.536-0.721) . Correlations of various anthropometric indices (r=0.655-0.691) with hydrodensitometrically determined %BF were even lower. It was noteworthy, however, that mean %BF derived from existing BI equations differed significantly from hydrodensitometrically determined mean %BF. Therefore, we attempted to develop a new equation that was applicable to Japanese adult women as follows: Db=1.1613-0.1038 (Wt⋅Z ) /Ht<SUP>2</SUP>, where Wt=weight in kg, Z=impedance in ohms, and Ht=height in cm. The prediction accuracy of this equation was r=0.866 or SEE=0.0077 g/ml. Cross-validation of this equation on a different sample (122 Japanese women, aged 18 to 59 years) revealed a correlation of r=0.869 in terms of %BF, SEE=3.2%, and no significant difference between estimated %BF and the criterion. We suggest that the BI method is one of the most convenient, valid means of assessing human body composition, and that the newly developed BI equation could be useful particularly when the subjects are Japanese adult women in their late teens to fifties.
Résumé
In previous assessments of body composition by whole body bioelectrical impedance (BI) analysis, electrodes have almost always been placed on the right side of the body. In fact, the most commonly used equations of Lukaski et al, were developed using BI measurements obtained on the right side of the body. However, in some individuals with traumatic injury or orthopedic problems, it would sometimes be necessary to measure BI on the right left side of the body. In the present study, we investigated the effects of electrode placement on BI and the derived percentage body fat. Subjects were 72 nontrained, healthy adult women : age ; 28.1±12.6 yr (1866), height ; 156.3±6.0cm, weight ; 50.5±7.7 kg, percentage body fat ; 24.4±5.2%. BI was measured for each subject in a supine position by use of a Selco SIF-881 plethysmograph (800 μA, 50 kHz) and ECG electrodes (Nikon Kohden) . The tetrapolar configuration was adopted in order to minimize contact impedance or skin-electrode interation. Eating and exercise were prohibited for at least 3 h prior to assessment. The effects of electrode placement were determined under four conditions: 1) the right arm and right leg (R side), 2) the right arm and left leg (R side-L side), 3) the left arm and right leg (L side-R side), 4) the left arm and left leg (L side) . Body density was predicted from the equation developed by Nagamine et al., and percentage body fat was derived from the body density according to Brozek et al. There were significant differences in BI values among the four conditions. Dominant side BI values were significantly lower than those on the non-dominant side. Percentage body fat values estimated under four different BI test conditions (i, e., R, R-L, L-R, and L) in terms of electrode placement were found to be highly correlated (r= 0.9420-0.956) with hydrodensitometrically determined percentage body fat. However, the mean percentage body fat on the dominant side of the body were significantly lower than that on the non-dominant side. We suggest that electrodes can be placed either on the dominant side or on the non-dominant side of the body for normal individuals, assuming that the lowest value from the four combinations of measurements can be used as the criterion value of BI. When the subjects are athletes, BI values obtained on the dominant side or a mean of the values measured on both sides should be adopted.