Introduction: Skinfold (SKF) equations exist to predict percent body fat (%BF) in athletes; however, none have been derived from multicomponent model reference measures.
Purpose: To develop and cross-validate a %BF prediction equation based on SKF in athletes using a four-component model as the reference measure.
Methods: Subjects were 132 collegiate athletes (20.7 +/- 2.0 yr; 78 males: 28 black, 50 white; 54 females: 10 black, 44 white). Four-component model estimates of %BF (%BF4C) included measures of total body water from deuterium dilution, bone mineral by dual- energy x-ray absorptiometry (DXA), and body density by densitometry using underwater weighing. SKF measures included subscapular, triceps, chest, midaxillary, suprailiac, abdominal, and thigh sites (7SKF). A prediction equation was developed on 102 athletes using 7SKF, race, and gender as predictor variables. Cross-validation was performed on a representative holdout sample of 30 athletes.
Results: The equation cross-validated well (slope and intercept both not different (P > 0.05) from the line of identity (LOI); r(YY') = 0.85, total error (TE) = 3.76%BF) and was better than the existing athlete SKF equations (intercept and slope both different from LOI (P < 0.01); r(YY') = 0.76, TE = 4.51%BF). Notably, a prediction equation developed using 3SKF sites (abdomen, thigh, and triceps) produced a similar accuracy (intercept and slope both not different from LOI (P > 0.05); r(YY') = 0.85, TE = 3.66%BF).
Conclusions: The new 7SKF equation improved on SKF equations developed using densitometry. The final equation based on the whole sample was %BF' = 10.566 + 0.12077*(7SKF) - 8.057*(gender) - 2.545*(race). Moreover, a 3SKF equation was comparable in accuracy to the 7SKF equation: BF' = 8.997 + 0.24658*(3SKF) - 6.343*(gender) - 1.998*(race).