TO THE EDITOR: Bergman et al. in article titled “A Better Index of Body Adiposity (1)” introduce the body adiposity index (BAI) as a measure of body adiposity being superior to the commonly used BMI. The BAI correlated more strongly with percentage body fat (PBF) determined by dual-energy X-ray absorptiometry (DXA). Although the authors need to be commended for their attempt to find an easily calculated alternative to the BMI, we would like to point out some difficulties.
The correlation between body height and PBF in the study of Bergman et al. (1) may largely be driven by sex differences in both, height and PBF. This raises the question whether one could use sex instead of height besides hip circumference to calculate the BAI. It furthermore makes a comparison to BMI somewhat unfair: BMI is largely independent of height while BAI attempts to reflect both, relative body fatness and differences in body composition between males and females. Although in the total population BMI may not be highly correlated with PBF from DXA, the correlation was quite high considering women and men separately, a finding supported by NHANES data (2). BMI together with sex can be used to predict PBF from densitometry quite accurately (r2 ∼0.67) (3). Whether the BAI is superior here still needs to be demonstrated.
The particularly strong correlation between hip circumference and PBF from DXA compared to other measures in the study of Bergman et al. (1) may in part be due to the over-representation of women. Unfortunately, Bergman et al. (1) do not report whether BAI predicts PBF among males and females similarly well. Figures 3 and 4 indicate that the relation between BAI and PBF from DXA may be nonlinear with different slopes for men and women and that BAI may overestimate PBF at the lower end of body fatness (thus, more frequently among males). Furthermore, while hip circumference might be more informative compared to waist circumference to predict PBF among women, the correlation between waist circumference and PBF is considerably stronger among men (2). Thus, it seems possible that waist circumference among men and hip circumference among women outperform BAI.
Taken together, the added value of the BAI beyond BMI seems that it captures the well-established difference in body composition between males and females (4). However, using hip and waist circumferences as gender-specific correlates of body fatness could be more accurate and easier options for this task. No matter how PBF is estimated from easily obtained anthropometric measurements—based on single circumferences, ratios like the BAI, or more comprehensive prediction models—the superiority of PBF over BMI to predict obesity related markers or endpoints has not clearly been demonstrated so far (5). Furthermore, future studies are needed to determine cutoffs of PBF in relation to type 2 diabetes, cardiovascular morbidity, and mortality. Such cutoffs will likely need to be sex-specific given the large sexual dimorphism related to body composition (4).