Secondary Logo

Evaluation of the BAI using ADP in African American Females

Tyo, Brian M.; Mangum, Michael

Translational Journal of the American College of Sports Medicine: March 1, 2019 - Volume 4 - Issue 5 - p 28–33
doi: 10.1249/TJX.0000000000000080
Original Investigation
Free
SDC

Objectives The aim of this study was to determine the agreement between body adiposity index (BAI) and body fat percentage using air displacement plethysmography (ADP) in African American females.

Methods Seventy-two African American females (28.9 ± 10.2 yr) participated in the study. Pearson’s correlation coefficients were calculated to determine the relationships among waist circumference, hip circumference (HC), BAI, body mass index (BMI), and %BFADP. Bland–Altman plots were generated to analyze agreement between BAI and %BFADP.

Results BAI and BMI were highly correlated with each other and %BFADP. BAI was more accurate in African American females within the normal BMI category but underestimated more in overweight and obese categories. However, the number of false-negative results when evaluating obesity was the same for BAI and BMI within this sample, which suggests their value as a clinical tool may be similar. HC was correlated with %BFADP. However, waist circumference correlated stronger than HC in the obese group.

Conclusion The strong correlations of BAI and BMI to body fat percent were similar to previous studies in other races/ethnicities and methods. BAI underestimates body fat percent more with increasing levels of adiposity possibly because of the accumulation of fat in areas not captured by an HC measure.

Columbus State University, Columbus, GA

Address for correspondence: Brian M. Tyo, Ph.D., Columbus State University, 4225 University Avenue, Columbus, GA 31907-5645 (E-mail: tyo_brian@columbusstate.edu).

Back to Top | Article Outline

INTRODUCTION

The body adiposity index (BAI) was first introduced by Bergman and colleagues in 2011 (1) as a potential method to improve on limitations of body mass index (BMI). BMI has limitations in that it does not differentiate between fat mass and fat-free mass (FFM), does not account for variance in the density of FFM, and does not easily allow for a conversion to body fat percent. Such limitations lead BMI to be less valid and applicable in populations with greater amounts of FFM or higher/lower tissue densities. This may limit real-world applications and acceptance among various groups. Even with these limitations, researchers commonly use BMI to describe obesity-related trends because of the ease of measurement and wide acceptance. For instance, data from the National Health and Nutrition Examination Survey indicate that the prevalence of obesity, as measured by BMI, is >37% (2) in the United States. Researchers could benefit from the development of a method that uses simple anthropometrics to estimate percent body fat, thus the emergence of BAI.

BAI was developed based on simple anthropometrics (hip circumference [HC] and height) from cohorts of Mexican American adults and validated using dual-energy x-ray absorptiometry (DXA) as the criterion measure (1). BAI was later evaluated for African American adults. Since the publication (1), BAI has been studied among different populations (3–8) and compared with other methods including skinfolds (4,6,9), DXA (5,10), bioelectric impedance analysis (BIA) (7), and air displacement plethysmography (ADP) (3,11). What emerges from these studies is that estimates of body fat percent can vary substantially for individuals among the different measures (12,13). Having body fat percent accurately estimated using BAI could lead to more real-world application and practice that may be easier for the lay population to understand and accept.

African American females tend to have a high prevalence of obesity as measured by BMI (2) and abdominal adiposity as measured by waist circumference (WC) (14–16). It has also been demonstrated that density (kg·L−1), notably muscle and bone density, of FFM tends to be greater in African Americans compared with Caucasians (17–19), which can affect the assessment of body fat percent and the interpretation of BMI.

Because of the disproportionately high prevalence of excessive adiposity being reported among African American females, significantly higher density of FFM among African American females, and the need to establish concurrent validity it is important to compare BAI with other methods of estimating body fat percent and adiposity. Doing so will determine the extent of the efficacy and application of BAI in real-world, population-based studies. Few studies have compared BAI, ADP, and BMI (3,8) and to our knowledge none have done so among African American females. Therefore, the purpose of this study was to evaluate and determine the extent of the relationship among BAI, percent body fat as measured by ADP (%BFADP), and BMI among African American females. In addition, the purpose was to evaluate the relationships of simple anthropometrics with %BFADP to help determine the utility of the BAI. This study is important because it extends our understanding of the strengths and weaknesses of BAI, irrespective of the perceived validity of the technique.

Back to Top | Article Outline

METHODS

Population

Participants were recruited from the surrounding community through dissemination of flyers and interpersonal communication. Potential participants were excluded from the study if they reported being pregnant, having an internal defibrillator, pacemaker, or joint replacement. Before data collection, each participant provided informed consent to participate in this study. The experimental protocol and informed consent were approved by the university institutional review board. The sample size was based on previous reports comparing BAI with ADP (3,11). Kalra et al. (3) included 80 Asian adults (34 females) in the analysis, and Rossato et al. (11) included 37 subjects (19 females) in a study on people with Down’s syndrome. The current study includes 72 African American women.

Back to Top | Article Outline

Study Protocol

Participants were asked to abstain from eating and exercise 2 h before the appointment, which is consistent with previous studies (3,11), and to wear tight-fitting Lycra-style clothing (1-piece or 2-piece). Participants were encouraged to hydrate (with water) normally, which was not restricted before measurement. In addition, participants were encouraged to void before anthropometrics. Shoes and socks were removed, and anthropometrics were recorded. Height was measured using a wall-mounted stadiometer (cm). WC and HC were measured using a tape with a tension (spring) handle (Creative Health Products, Inc., Plymouth, MI). WC was measured with the tape directly on the skin after a normal expiration at the narrowest portion of the torso, between the xiphoid process and the umbilicus. HC was measured at the greatest circumference of the buttocks. Measurements were performed in duplicate and repeated if values differed by more than 5 mm. The two closest values were averaged to obtain the final measure. Waist-to-hip ratio (WHR) was calculated by dividing WC by HC. BMI was calculated by dividing mass (kg) by height squared (m2).

Body composition was measured using ADP (Bod Pod® Cosmed, Inc., Concord, CA) for all participants while wearing “Lycra” style clothing and swim cap. Participants removed glasses and jewelry before measurement. Calibration and testing was completed according to manufacturer’s requirements. Body composition was calculated using an equation specifically designed for African American females (19). Body composition was also calculated using the BAI as described by Bergman and colleagues (1), where BAI (%BF) = [HC / (height1.5)] − 18.

Back to Top | Article Outline

Statistical Analysis

Mean and SD values were reported for continuous variables. One-way ANOVA was used to determine differences for anthropometrics among BMI categories. Bonferonni post hoc analyses were used to make pairwise comparisons where appropriate. A paired-samples t-test was used to determine differences between BAI and ADP. Lin’s concordance correlation coefficient was used to determine agreement between BAI and ADP. Bland–Altman plots were used to determine the limits of agreement between BAI and ADP. Pearson correlations were calculated to determine the extent of the linear relationships among BAI, BMI, ADP, WC, HC, WHR, height, and weight. Two stepwise linear regression analyses were performed. The first was to determine whether BAI and/or BMI predict %BFADP. The second was to determine whether anthropometrics (WC, HC, WHR, height, and weight) were significant predictors of %BFADP, singularly or in concert.

Participants were classified into BMI categories: normal weight (18.5–24.9 kg·m−2), overweight (25–29.9 kg·m−2), and obese (≥30 kg·m−2) for further statistical analysis. Lin’s concordance correlation coefficient was calculated to determine agreement between BAI and ADP within each of the BMI categories. Bland–Altman plots were also calculated to determine the limits of agreement between BAI and ADP within each of the BMI classifications. Lastly, two stepwise linear regression analyses were performed. The first was to determine whether BAI and/or BMI predict %BFADP within each BMI category. The second was to determine whether anthropometrics (WC, HC, WHR, height, and weight) were significant predictors of %BFADP, singularly or in concert within each BMI category.

Analyses were carried out using the Statistical Package for the Social Sciences for Windows (Version 21.0; SPSS Inc., Chicago, IL). The alpha level was set at 0.05 for individual tests.

Back to Top | Article Outline

RESULTS

Seventy-two participants who identified themselves as African American females participated in the study. Mean and SD values of the descriptive characteristics are presented in Table 1 and categorized by BMI. The study population ranged in age from 18 to 54 yr, and BMI ranged from 17.8 to 57.8 kg·m−2. Three participants had a BMI <18.5 kg·m−2. All groups were significantly different for weight, BMI, percent fat (ADP and BAI), WC, and HC (P < 0.05).

TABLE 1

TABLE 1

Estimates of percent body fat were significantly greater for ADP (35.5 ± 9.8%) than BAI (32.8 ± 6.4%) in this sample (t71 = 4.5, P < 0.001). BAI underestimated body fat percent by a mean of 2.62% when compared with ADP. Lin’s concordance analysis revealed a C_b coefficient of 0.873 between ADP and BAI. Bland–Altman plots of BAI and ADP are presented in Fig. 1. BAI was correlated to both ADP (r = 0.894, P < 0.001) and BMI (r = 0.913, P < 0.001), and BMI was correlated to ADP (r = 0.893, P < 0.001). Correlations related to body fat percent (%BFADP) are presented in Table 2. WC, HC, and weight were highly correlated with %BFADP. Stepwise regression analysis revealed that BAI was a better predictor of %BFADP explaining 79.9% of the variance (R2 = 0.799, SEE = 4.4, β = 0.894, P < 0.001). However, BMI explained an additional 3.4% of the variance (R2 = 0.834, SEE = 4.0, P < 0.001) with β = 0.472 for BAI and β = 0.461 for BMI. Of the anthropometric variables, HC significantly contributed to the stepwise regression model (R2 = 0.781, SEE = 4.6, β = 0.640, P < 0.01). In addition, WHR and weight significantly contributed to the model (R2 = 0.849, SEE = 3.9, P < 0.001) with β = 1.364, β = 0.323, and β = −0.584 for HC, WHR, and weight, respectively.

Figure 1

Figure 1

TABLE 2

TABLE 2

Once classified into BMI categories (normal, overweight, and obese), Lin’s C_b coefficient of agreement was 0.864, 0.576, and 0.580 for the normal, overweight, and obese groups, respectively. Bland–Altman plots by BMI category (normal, overweight, and obese) are presented in Fig. 2. It can be seen that BAI overestimated body fat percent by 0.5% in the normal weight group and underestimated body fat percent by 3.3% and 5.5% for overweight and obese groups, respectively. %BFADP remained highly correlated to HC in the normal weight group, whereas WC was highly correlated in the obese group. Stepwise regression analysis revealed that BMI was a better predictor of %BFADP for each BMI category explaining 58.0% of the variance (R2 = 0.580, SEE = 4.4, β = 0.761, P < 0.001) for normal weight, 24.9% of the variance (R2 = 0.249, SEE = 3.6, β = 0.499, P < 0.05) for overweight, and 69.0% of the variance (R2 = 0.690, SEE = 3.00, β = 0.838, P < 0.001) for the obese group. Stepwise regression analysis revealed that HC was a better predictor of %BFADP for normal weight (R2 = 0.684, SEE = 3.8, β = 0.827, P < 0.001) and overweight (R2 = 0.231, SEE = 3.7, β = 0.481, P < 0.05). However, WC was a better predictor of %BFADP in the obese group (R2 = 0.660, SEE = 3.1, β = 0.820, P < 0.001).

Figure 2

Figure 2

Back to Top | Article Outline

DISCUSSION

This is the first study to evaluate the real-world utility of BAI in an African American female cohort using ADP. BAI correlated well with %BFADP, which is consistent with previous work in south Asian adults (3) and clinically severe obese females (8). BAI underestimated body fat percent by 2.6% with significant negative bias. This indicates that BAI overestimated %BFADP in females with lower body fat percent levels while underestimating more in females with higher fat levels. Mean difference of BAI and %BFADP was the smallest in the normal BMI group and progressively underestimated more in the overweight and obese groups, which is also consistent with previous work using ADP (3,8) and DXA (1,20,21). Vinknes et al. (20) reported that BAI overestimated BF% in the normal weight group, and multiple studies found that BAI had the highest correlation with BF% for values from 20% to 30% (1,21,22), which was also found in the current study. The current study is also consistent with Elisha et al. (23) that reported a mean difference of ~6.8% in a group of obese females. In addition, Geliebter et al. (8) found that BAI underestimated %BFADP by ~5% in clinically severe obese females. These findings may be affected by the way the obese group stores excess fat compared with the normal and overweight groups. For example, the BAI uses HC to estimate %BF. Therefore, if females tend to store disproportionately more fat mass in other locations other than the hips, the BIA measure may tend to underestimate %BF.

Previous work has indicated that BAI is a superior indicator of %BF when compared with BMI (1,22,24), whereas others have refuted such findings (20,21,25,26). In this sample, BAI correlated well with %BFADP and was nearly identical with the correlation of BMI with %BFADP. Furthermore, the current study found that BAI was significantly correlated to %BFADP for the normal weight and obese groups, but not the overweight group. The reasons for this insignificant value are unclear and warrant further investigation because of the health disparity of overweight/obesity in this population. Interestingly, the correlation of BMI with %BFADP was significant within all BMI categories, with the weakest correlation also in the overweight group. BAI was a better predictor of %BFADP for the entire cohort; however, BMI was a better predictor of %BFADP for the individual BMI categories. The benefits of accurate specific body fat percent measures as opposed to appropriate aggregate classification remain unclear, and utility may be dependent on the research question or clinical application. It is interesting to note that both BMI and BAI had the same number of false negatives (n = 13) and false positives (n = 0) when using 35% fat as the criteria for “obesity.” As such, BMI and BAI would perform similarly aggregating individuals into an “obese” category. Some researchers and practitioners may consider the calculation for BMI to be a simpler measure (height and weight) and a more universally accepted standard when compared with BAI (height and WC).

In the sample aggregate (N = 72), both WC and HC are powerful univariate predictors of ADP, BMI, and BAI. However, when the sample was broken into three levels of adiposity (normal, overweight, and obese), the coefficients became less stable. Specifically, waist and HC became mostly poor predictors in the overweight group while maintaining their strength in the normal and obese groups. Explanations for this paradox are not forthcoming from these data. However, several plausible explanations exist.

Many would turn to the small sample size of the individual cells (n = 22, 20, and 27) and argue that the lack of statistical power explains the varying correlation coefficients. Although possible, the pattern of coefficients across groups remains scientifically intriguing. The presence of steatopygia in the African American population, although common, is certainly not universal. Consequently, the three groups defined by this study may well be anthropometrically distinct so that selection pressures are acting on the grouping. If so, then the relationship between HC and adiposity indices may not be a linear function across all levels of adiposity.

As indicated in previous studies (1,20,21), HC correlated well with %BF, possibly due to sexual dimorphism (27). The current study extends these findings, indicating that the correlation of HC and %BFADP was significant within all BMI categories but was lowest in the overweight group. The reason for the difference is unclear. However, this could be due to variability in fat distribution for a given weight status. WC correlated well with %BFADP and remained significant in the normal weight and obese groups. The highest correlation between WC and %BFADP was seen in the obese groups, which may be indicative of accumulating a greater amount of fat in the area. In support of this point, Elisha et al. (23) reported significant correlations between change in WC and visceral fat with change in %BF in a group of obese females participating in a weight loss study. Unfortunately, Elisha et al. did not provide a change in HC measure to provide further support. The current study demonstrates that HC predicts %BFADP best in the normal weight group. HC is still the main anthropometric predictor in the overweight group, although the strength of the relationship is diminished. Finally, in the obese group, WC becomes the most predictive anthropometric variable. These findings may be indicative of adiposity accumulating in other areas of the body not captured by HC with increasing adiposity levels.

This study does have limitations. For instance, only African American females were included in the study. Findings from this study should not be extended to men, other racial/ethnic groups, or populations with various metabolic morbidities that may affect adiposity or distribution of adiposity. However, this study is strengthened by evaluating BAI using ADP and focusing on a population with reports of disproportionately high levels of adiposity.

Back to Top | Article Outline

CONCLUSION

BAI does not appear to substantially improve on BMI when evaluating adiposity. BAI tends to underestimate %BF when compared with %BFADP. In addition, BAI seems to be most accurate in African American females that are in the normal BMI group. Lastly, BAI may underestimate %BF more in overweight and obese groups because of individuals accumulating more fat in areas that would not be captured by an HC measure. Further research is needed to elucidate where/how various populations gain (or lose) weight and how this may affect the utility of the BAI and its application to clinical decision-making. In addition, when BMI is utilized in research studies, we encourage the assessment of HC to allow for further comparisons with BAI. Doing so will help further elucidate the utility (or lack of utility) of the BAI. The authors have identified questions in this study that warrant further investigation. For example, understanding the degree in which steatopygia may frustrate attempts to classify African American females when conducting anthropometric research, especially when viewing African American females as a homogenous group.

The author thanks all participants who contributed to this manuscript. He also thanks his friends and colleagues, Dr. Clayton Nicks, Dr. Kate Early, and Rebecca Spataro-Kearns, for their encouragement, expertise, and professionalism.

The results of this study do not constitute endorsement by the American College of Sports Medicine. The authors declare no conflicts of interest and no sources of external funding.

Back to Top | Article Outline

REFERENCES

1. Bergman RN, Stefanovski D, Buchanan TA, et al. A better index of body adiposity. Obesity. 2011;19(5):1083–9. doi: 10.1038/oby.2011.38. PubMed PMID: 21372804; PubMed Central PMCID: PMC3275633.
2. Flegal KM, Kruszon-Moran D, Carroll MD, Fryar CD, Ogden CL. Trends in obesity among adults in the United States, 2005 to 2014. JAMA. 2016;315(21):2284–91. doi: 10.1001/jama.2016.6458. PubMed PMID: 27272580.
3. Kalra S, Mercuri M, Anand SS. Measures of body fat in South Asian adults. Nutr Diabetes. 2013;3:e69. doi: 10.1038/nutd.2013.10. PubMed PMID: 23712281; PubMed Central PMCID: PMC3671745.
4. Gupta S, Kapoor S. Body adiposity index: its relevance and validity in assessing body fatness of adults. ISRN Obes. 2014;2014:243294. doi: 10.1155/2014/243294. PubMed PMID: 24587942; PubMed Central PMCID: PMC3920613.
5. Esco MR. The accuracy of the body adiposity index for predicting body fat percentage in collegiate female athletes. J Strength Cond Res. 2013;27(6):1679–83. doi: 10.1519/JSC.0b013e3182712714. PubMed PMID: 22990566.
6. Wickel EE. Evaluating the utility of the body adiposity index in adolescent boys and girls. J Sci Med Sport. 2014;17(4):434–8. doi: 10.1016/j.jsams.2013.06.002. PubMed PMID: 23921073.
7. Lopez AA, Cespedes ML, Vicente T, et al. Body adiposity index utilization in a Spanish Mediterranean population: comparison with the body mass index. PloS One. 2012;7(4):e35281. doi: 10.1371/journal.pone.0035281. PubMed PMID: 22496915; PubMed Central PMCID: PMC3322155.
8. Geliebter A, Atalayer D, Flancbaum L, Gibson CD. Comparison of body adiposity index (BAI) and BMI with estimations of % body fat in clinically severe obese women. Obesity. 2013;21(3):493–8. doi: 10.1002/oby.20264. PubMed PMID: 23592658; PubMed Central PMCID: PMC3470730.
9. Freedman DS, Blanck HM, Dietz WH, DasMahapatra P, Srinivasan SR, Berenson GS. Is the body adiposity index (hip circumference/height(1.5)) more strongly related to skinfold thicknesses and risk factor levels than is BMI? The Bogalusa Heart Study. Br J Nutr. 2013;109(2):338–45. doi: 10.1017/S0007114512000979. PubMed PMID: 22716994; PubMed Central PMCID: PMC4427245.
10. Freedman DS, Thornton JC, Pi-Sunyer FX, et al. The body adiposity index (hip circumference / height(1.5)) is not a more accurate measure of adiposity than is BMI, waist circumference, or hip circumference. Obesity. 2012;20(12):2438–44. doi: 10.1038/oby.2012.81. PubMed PMID: 22484365; PubMed Central PMCID: PMC3477292.
11. Rossato M, Dellagrana RA, de Souza Bezerra E, et al. Comparison of body adiposity index (BAI) and air displacement plethysmograph with estimations of % body fat in adults with Down's syndrome. Eur J Clin Nutr. 2017;71(11):1341–4. doi: 10.1038/ejcn.2017.18. PubMed PMID: 28294169.
12. Biaggi RR, Vollman MW, Nies MA, et al. Comparison of air-displacement plethysmography with hydrostatic weighing and bioelectrical impedance analysis for the assessment of body composition in healthy adults. Am J Clin Nutr. 1999;69(5):898–903. PubMed PMID: 10232628.
13. Maddalozzo GF, Cardinal BJ, Snow CA. Concurrent validity of the BOD POD and dual energy x-ray absorptiometry techniques for assessing body composition in young women. J Am Diet Assoc. 2002;102(11):1677–9. PubMed PMID: 12449296.
14. Ford ES, Li C, Zhao G, Tsai J. Trends in obesity and abdominal obesity among adults in the United States from 1999–2008. Int J Obes (Lond). 2011;35(5):736–43. doi: 10.1038/ijo.2010.186. PubMed PMID: 20820173.
15. Ladabaum U, Mannalithara A, Myer PA, Singh G. Obesity, abdominal obesity, physical activity, and caloric intake in U.S. Adults: 1988–2010. Am J Med. 2014;127(8):717–27. doi: 10.1016/j.amjmed.2014.02.026. PubMed PMID: 24631411.
16. Beydoun MA, Wang Y. Gender-ethnic disparity in BMI and waist circumference distribution shifts in US adults. Obesity. 2009;17(1):169–76. doi: 10.1038/oby.2008.492. PubMed PMID: 19107129; PubMed Central PMCID: PMC2610345.
17. Prior BM, Modlesky CM, Evans EM, et al. Muscularity and the density of the fat-free mass in athletes. J Appl Physiol. 2001;90(4):1523–31. PubMed PMID: 11247955.
18. Cote KD, Adams WC. Effect of bone density on body composition estimates in young adult black and white women. Med Sci Sports Exerc. 1993;25(2):290–6. PubMed PMID: 8450735.
19. Ortiz O, Russell M, Daley TL, et al. Differences in skeletal muscle and bone mineral mass between black and white females and their relevance to estimates of body composition. Am J Clin Nutr. 1992;55(1):8–13. PubMed PMID: 1728823.
20. Vinknes KJ, Elshorbagy AK, Drevon CA, et al. Evaluation of the body adiposity index in a Caucasian population: the Hordaland Health Study. Am J Epidemiol. 2013;177(6):586–92. doi: 10.1093/aje/kws271. PubMed PMID: 23444101.
21. Segheto W, Coelho FA, Cristina Guimaraes da Silva D, et al. Validity of body adiposity index in predicting body fat in Brazilians adults. Am J Hum Biol. 2017;29(1): doi: 10.1002/ajhb.22901. PubMed PMID: 27502080.
22. Chang H, Simonsick EM, Ferrucci L, Cooper JA. Validation study of the body adiposity index as a predictor of percent body fat in older individuals: findings from the BLSA. J Gerontol A Biol Sci Med Sci. 2014;69(9):1069–75. doi: 10.1093/gerona/glt165. PubMed PMID: 24158764; PubMed Central PMCID: PMC4158412.
23. Elisha B, Rabasa-Lhoret R, Messier V, Abdulnour J, Karelis AD. Relationship between the body adiposity index and cardiometabolic risk factors in obese postmenopausal women. Eur J Nutr. 2013;52(1):145–51. doi: 10.1007/s00394-011-0296-y. PubMed PMID: 22209967.
24. Dhaliwal SS, Welborn TA, Goh LG, Howat PA. Obesity as assessed by body adiposity index and multivariable cardiovascular disease risk. PloS One. 2014;9(4):e94560. doi: 10.1371/journal.pone.0094560. PubMed PMID: 24714547; PubMed Central PMCID: PMC3979843.
25. Barreira TV, Harrington DM, Staiano AE, Heymsfield SB, Katzmarzyk PT. Body adiposity index, body mass index, and body fat in white and black adults. JAMA. 2011;306(8):828–30. doi: 10.1001/jama.2011.1189. PubMed PMID: 21862743; PubMed Central PMCID: PMC3951848.
26. Zhao D, Li Y, Zheng L, Yu K. Brief communication: body mass index, body adiposity index, and percent body fat in Asians. Am J Phys Anthropol. 2013;152(2):294–9. doi: 10.1002/ajpa.22341. PubMed PMID: 23996556.
27. Schulze MB, Stefan N. The body adiposity index and the sexual dimorphism in body fat. Obesity. 2011;19(9):1729. doi: 10.1038/oby.2011.153. PubMed PMID: 21874027.
© 2019 American College of Sports Medicine