Comparison of Several Prediction Equations Using Skinfold Thickness for Estimating Percentage Body Fat in 12-15-year-old Indonesian Children

Background and Objective: Gold standard measures of percentage body fat (%BF) have more obstacles in epidemiological studies. Skinfold prediction equations are among the viable methods widely used to estimate %BF. This study aimed to compare several skinfold equations estimating %BF against the %BF obtained from bioimpedance analysis (BIA) in Indonesian children aged 12-15 years. Materials and Methods: We invited 610 children (290 boys, 320 girls) aged 12-15 years from junior high schools in Yogyakarta Province. Weight, height and skinfold thickness at the triceps, biceps, subscapular and suprailiac skinfolds were taken according to the protocols of the International Society for the Advancement of Kinanthropometry (ISAK). The percentage of body fat was measured with BIA and estimated using several available skinfold equations. Statistical analyses using paired tests, linear regressions and Bland and Altman plots were performed. Results: The skinfold equations of Durnin and Rahaman and Wickramasinghe significantly overestimated %BF (p<0.001) in boys and girls, while the equation of Watanabe et al. underestimated the %BF (p<0.001). The equation of Slaughter et al. gave the lowest bias but it slightly overestimated %BF in boys (-2.02±5.22% difference; p<0.001) and underestimated %BF in girls (0.78±4.48% difference; p<0.001). All equations showed high correlation with %BF obtained from BIA (r = 0.85-0.87; p<0.001) sand acceptable range of limits of agreement (LOA; 2.6-5.1%), with the lowest being from Slaughter et al. Conclusion: All selected equations were highly correlated with %BF obtained from BIA and within the acceptable LOA. The skinfold equation of Slaughter et al. showed the most applicability in predicting %BF in our sample population.