Development of a Body Fat Prediction Equation for Young Adults

Abstract

Given the rising prevalence of obesity, particularly in young adults, it is important to develop valid field‐based methods that can used to estimate body fatness. Body mass index (BMI; in kg/m2) is currently used to categorize obesity, despite the fact that BMI is only moderately correlated with percent body fat. The purpose of this study was to develop a body fat prediction equation specifically for a college‐aged population. 270 Caucasian (n=161) and African American (n=109) students aged 18‐24 volunteered to participate. Height and weight were measured and BMI values calculated. Total percent body fat (%BF) was estimated using dual‐energy X‐ray absorptiometry (DXA). Stepwise multiple linear regression examined significant predictor variables of %BF, including gender, race, age, and BMI. The predicted residual error sum of squares (PRESS) statistic cross‐validated the model. The reduced model explained 80% of the variance in total percent body fat (%BF = 1.53[BMI] ‐ 16.47[gender] ‐ 1.29[race] ‐ 4.5; note: 0 =female, 1=male; 0=Caucasian, 1=African American). Additionally, the root mean squared error (RMSE) for the model (4.2 %BF) was equivalent to the cross‐validation statistic. These results suggest that body adiposity can be predicted with greater accuracy in a college‐aged population when gender and race are included in addition to BMI.Grant Funding Source: N/A