Improved body composition assessment using biceps skinfold and physical activity score in premenarcheal girls: a DXA-based validation study.

Objective: Age-specific body mass index (BMI) is commonly employed as an index of adiposity for pediatric clinical and research purposes. However, BMI fails to discriminate between fat and lean mass, making it an imperfect monitor for obesity. We hypothesized that simple anthropometry and organized non-aquatic physical activity assessment (PA) would provide superior explanatory value for pediatric body composition outcomes.

Research methods and procedures: In a cross-sectional analysis, whole body DXA assessed body composition in 120 pre-menarcheal girls. Questionnaires were used to record and generate annual means for PA. Age, Tanner breast self-stage, height, weight, BMI, skinfold thicknesses, girths and PA were examined as potential predictors of body composition outcomes, using backward stepwise multiple linear regression. A parsimonious regression model was developed in 75% and cross-validated in 25% of the total sample; models were rerun with the total sample.

Results: Subject means were as follows: age = 10.4±1.2 y; lean soft tissue (LST) = 24.4±4.2 kg; fat mass (FM) = 8.1±4.1 kg; BMI = 17.6±2.5 kg/m²; PA = 6.8±5.0 h/wk; Tanner breast self-stage ranged from 1 to 3. BMI for age Z scores ranged from -2 to 2.1. Age and BMI alone yielded adjusted model r²=0.44 to 0.78. The final model, including age, height, weight, biceps skinfold and PA, yielded adjusted r²=0.61 to 0.92, P <0.001. Prediction of LST and FM increased from r²=0.64 and 0.76 to r²=0.92 and 0.91, respectively.

Discussion: Compared to BMI and age alone, models including biceps skinfold, PA, height, weight and age had superior explanatory value for clinically-relevant body composition outcomes, and are feasible for clinical use.