OBJECTIVE: To develop the best prediction equation for estimating fat-free mass (FFM) from bioelectrical impedance and anthropometric variables in Japanese boys. SUBJECTS: 141 boys, aged 9-14 years, were recruited from elementary schools managed by Osaka City. METHODS: FFM was calculated from underwater weighting (UW) using a two-component model. Prediction equations were developed using multiple regression analyses. Adjusted R2 and standard errors of the estimate (SEE) ranged from 0.831 to 0.854 and from 1.25 to 1.80 kg in the validation group (n = 84), and ranged from 0.817 to 0.912 and from 1.27 to 1.86 kg in the cross-validation group (n = 57), respectively. The validation group and cross-validation group were combined in order to develop a more stable prediction equation with a greater sample size (n = 141). RESULTS: The variables ZI (height2/impedance), body mass (M) and abdominal skinfold thickness were found to be the best combination of predictors. Adjusted R2 and SEE for finally developed regression equations ranged from 0.949 to 0.975 and from 1.50 to 2.10 kg. CONCLUSION: As a result of all analyses and considering the practicability of the measurement, we recommend the equation FFM = 0.56ZI+0.20M+1.66, with an adjusted R2 of 0.971 and SEE of 1.59 kg, which can predict FFM with high precision in Japanese boys.
OBJECTIVE: To develop the best prediction equation for estimating fat-free mass (FFM) from bioelectrical impedance and anthropometric variables in Japanese boys. SUBJECTS: 141 boys, aged 9-14 years, were recruited from elementary schools managed by Osaka City. METHODS: FFM was calculated from underwater weighting (UW) using a two-component model. Prediction equations were developed using multiple regression analyses. Adjusted R2 and standard errors of the estimate (SEE) ranged from 0.831 to 0.854 and from 1.25 to 1.80 kg in the validation group (n = 84), and ranged from 0.817 to 0.912 and from 1.27 to 1.86 kg in the cross-validation group (n = 57), respectively. The validation group and cross-validation group were combined in order to develop a more stable prediction equation with a greater sample size (n = 141). RESULTS: The variables ZI (height2/impedance), body mass (M) and abdominal skinfold thickness were found to be the best combination of predictors. Adjusted R2 and SEE for finally developed regression equations ranged from 0.949 to 0.975 and from 1.50 to 2.10 kg. CONCLUSION: As a result of all analyses and considering the practicability of the measurement, we recommend the equation FFM = 0.56ZI+0.20M+1.66, with an adjusted R2 of 0.971 and SEE of 1.59 kg, which can predict FFM with high precision in Japanese boys.