Objective: The purpose of this study was to evaluate whether the recently developed body adiposity index (BAI) in Mexican American and African American women could be validated in postmenopausal American white women and to determine if it is a better obesity classification measure than body mass index (BMI) is in the latter population.
Methods: Total body percentage adiposity (%adiposity) measured by dual-energy x-ray absorptiometry (DXA) was compared with total body %adiposity predicted by BAI in 187 overweight/obese postmenopausal white women (mean ± SD %adiposity, 45.9 ± 4.9% and 38.3 ± 6.2% for DXA and BAI, respectively). SPSS 19.0 and Medcalc 11.6.1.0 were used to conduct Pearson's correlations (r), paired t test, receiver operating curve, and Lin's concordance coefficient (ρc) and to create Bland and Altman's limit-of-agreement plot.
Results: Pearson's correlation analysis revealed a strong association between DXA %adiposity and BAI (r = 0.78, P < 0.001), DXA %adiposity and BMI (r = 0.75, P < 0.001), and BMI and BAI (r = 0.90, P < 0.001). Bias correction factor was 0.51 between DXA %adiposity and BAI. Paired t test showed a significant mean difference between measurements (P < 0.0001), and the plot showed that BAI underestimated DXA %adiposity by 7.56%. Concordance coefficient (ρc = 0.39; 95% CI, 0.33-0.46) revealed a poor agreement strength. There was no difference between the area under the curve statistic for BAI and BMI.
Conclusion: Based on our results, BAI has limitations for use in a clinical setting in overweight/obese postmenopausal white women but may be practical for research applications and eventually developed into an easy method to estimate overweight/obesity in other settings.