Concurrent validity of five prediction equations to evaluate fat percentage in a sports group expected to yield high performance from Medellín, Colombia

INTRODUCTION: No equations to predict the body composition of athletes from Medellín expected to have high performance have been constructed and, thus, decisions regarding their training and nutrition plans lack support. OBJECTIVE: To calculate the concurrent validity of five prediction equations for fat percentage in a group of athletes from Medellín, Colombia, expected to yield high performance. MATERIALS AND METHODS: We conducted a cross-sectional analysis to validate diagnostic tests using secondary-source data of athletes under the age of 18 who were part of the “Medellín Team”. The gold standard was dual-energy X-ray densitometry (DEXA). We analyzed the Slaughter, Durnin and Rahaman, Lohman, and Johnston prediction equations, as well as the five-component model. We used the intraclass correlation coefficient to assess the consistency of the methods and the Bland-Altman plot to calculate the average bias and agreement limits of each of the equations. RESULTS: We included 101 athletes (50,5% of them women). The median age was 14,8 years (IR: 13,0 - 16,0). The concurrent validity was “good/excellent” for the Johnston and the Durnin and Rahaman equations and the five-components model. The Lohman equation overestimated the fat percentage in 12,7 points. All of the equations showed broad agreement limits. CONCLUSIONS: The Durnin and Rahaman and the Johnston equations, as well as the five- component model, can be used to predict the FP in the study population as they showed a “good/excellent” concurrent validity and a low average bias. The equations analyzed have low accuracy, which hinders their use to diagnose the individual fat percentage within this population.