Predictive Validity of the Body Adiposity Index in Costa Rican Students

Objective: To verify the validity of the body adiposity index (BAI) in a sample of Costa Rican students. Methods: Volunteers were 93 females (mean age = 18.6 +/- 2.4 years) and 106 males (mean age = 19.2 +/- 2.8 years). Dual-energy X-ray absorptiometry (DXA) was used as the "gold standard" to determine body fat percentage (BF%). Pearson's correlation coefficient and paired samples t-test studied the association and mean differences between BAI and DXA BF%. Concordance between BAI and DXA BF% was determined by the Lin's concordance correlation coefficient and the Bland-Altman agreement analysis. Results: Significant correlations between BAI and DXA BF% were found for females (r = 0.74) and males (r = 0.53) (P < 0.001). Differences between methods were found for females (BAI = 29.3 +/- 4.1% vs. DXA = 36.5 +/- 7.9%) and males (BAI = 24.8 +/- 3.7% vs. DXA = 21.9 +/- 8.6%; P<0.001). Concordance was poor in females and males. Bland-Altman plots showed BAI underestimating and overestimating BF% in relation to the "gold standard" in females and males, respectively. Conclusions: BAI presented low agreement with BF% measured by DXA; therefore, BAI is not recommended for BF% prediction in this Central American sample studied. (C) VC 2015 Wiley Periodicals, Inc.