Development and validation of anthropometric equations to estimate body composition in adult women

OBJECTIVE: To develop anthropometric equations to predict body fat percentage (BF%). METHODS: In 151 women (aged 18-59) body weight, height, eight- skinfold thickness (STs), six- circumferences (CIs), and BF% by hydrodensitometry were measured. Subjects data were randomly divided in two groups, equation-building group (n= 106) and validation group (n= 45). The equation-building group was used to run linear regression models using anthropometric measurements as predictors to find the best prediction equations of the BF%. The validation group was used to compare the performance of the new equations with those of Durnin-Womersley, Jackson-Pollock and Ramirez-Torun. RESULTS: There were two preferred equations: Equation 1= 11.76 + (0.324 x tricipital ST) + (0.133 x calf ST) + (0.347 x abdomen CI) + (0.068 x age) - (0.135 x height) and Equation 2= 11.37 + (0.404 x tricipital ST) + (0.153 x axilar ST) + (0.264 x abdomen CI) + (0.069 x age) - (0.099 x height). There were no significant differences in BF% obtained by hydrodensitometry (31.5 ±5.3) and Equation 1 (31.0 ±4.0) and Equation 2 (31.2 ±4.0). The BF% estimated by Durning-Womersley (35.8 ±4.0), Jackson-Pollock (26.5 ±5.4) and Ramirez-Torun (32.6 ±4.8) differed from hydrodensitometry (p <0.05). The interclass correlation coefficient (ICC) was high between hydrodensitometry and Equation 1 (ICC= 0.77), Equation 2 (ICC= 0.76), and Ramirez-Torun equation (ICC= 0.75). The ICC was low between hydrodensitometry and Durnin-Womersley (ICC= 0.51) and Jackson-Pollock (ICC= 0.53) equations. CONCLUSION: The new Equations-1 and 2, performed better than the commonly used anthropometric equations to predict BF% in adult women.