Body Fat Percentage Prediction with Circumferences: A Revisit with a Chinese Female Sample

This study aimed to update and validate regression models for estimating body fat percentage (BF%) in Chinese women using anthropometric measures. A total of 630 women aged 20–79 yr. underwent dual-energy X-ray absorptiometry (DXA) and standardized anthropometric assessments. Development of experimental models and cross-validation of the published equations were done on the entire sample. The final prediction models developed from this study are Total BF% = 5.76 + 0.666×BMI + 0.180×Waist (RMSE = 3.47 ± 0.36); and Android BF% = -13.3 + 0.711×BMI + 0.475×Waist (RMSE = 5.81 ± 0.64). Although models with higher complexity improved prediction accuracy, simpler models using waist circumference, BMI, or WtHR were nearly as effective. Among published equations, the Rush et al. (1997) model from a Western sample surprisingly outperformed equations derived from Asian samples. These findings support the use of circumference-based models as accessible, low-cost alternatives to DXA for estimating BF% in Chinese and potentially multiethnic female populations.