Cross-validation of prediction equations for estimating body composition in ballet dancers

In sports related to low body weight, such as classical ballet, the assessment of body composition is important for monitoring performance and health status. This study aimed to cross-validate anthropometry and bioelectrical impedance (BIA) predictive equations for estimating body composition of non-professional classical ballet dancers, using dual-energy-X-ray absorptiometry (DXA) as a reference method.

Thirty-seven female non-professional classical dancers (median age of 19 years), at intermediate/advanced level, were evaluated in a cross-sectional study. Body composition was evaluated by DXA, anthropometry and tetrapolar BIA. Twenty different predictive equations of anthropometry (n = 8) and BIA (n = 12) were used to estimate Body Fat (BF) and Fat-Free Mass (FFM), testing their validity against DXA using the Bland-Altman statistics.

For BF estimated by anthropometry equations, just one equation showed agreement with DXA (r = 0.852, p < 0.0005; p = 0.600 for one sample T-test). According to the Bland-Altman analysis, this equation also showed validity, with the absence of proportional bias. Regarding the predictive BIA equations tested, none were valid for our study group.

Only one of the anthropometric equations, the one proposed by Durnin and Womerley (1974), but none of the BIA equations analyzed, was valid for the evaluation of body composition of the studied classical dancers. Our results emphasize the importance of previous cross-validation of existing equations or the development of specific equations for body composition assessment in specific populations.