The assessment of body fat percentage (BF%) in athletic populations is important in numerous circumstances, such as determining the outcomes of strength and conditioning programs and evaluating overall physical fitness and health status (1,8,9). Determination of BF% estimates involves techniques primarily found in exercise physiology laboratories, such as hydrostatic weighing and dual-energy x-ray absorptiometry (DXA). Although highly accurate, these methods are often too costly, time consuming, and not readily available to practitioners. Fortunately, other techniques exist that predict BF% in field settings, such as skinfold techniques and bioelectrical impedance analysis. However, applied prediction methods are at risk of providing inaccurate individual estimations and are often impractical because of issues with intra- and interrater reliability, technician error, and the inability to evaluate a large group in a short time (1,7,10).
Body mass index (BMI) is a simplistic ratio of body weight to height and has been the primary determinate of overweight and obesity classifications in public health settings (17,24). It is a reliable measure, and the simplistic calculation provides a convenient tool for quickly evaluating a considerable number of subjects. Although it is highly related to clinical markers of chronic disease, it is not a valid measure of BF% because of its inability to distinguish between fat and nonfat tissues (23). Thus, BMI often misclassifies athletes as being overweight, as these individuals typically possess greater muscle masses and lower levels of BF% compared with nonathletes at the same relative body weight (13,19,20).
Recently, the body adiposity index (BAI) was created as a clinical alternative to BMI, with all the associated benefits, i.e., reliability and rapid calculation (3). The purported advantage of BAI over BMI is that the former predicts BF% via a simple equation of hip circumference to height (i.e., BF% = [hip circumference/height1.5] − 18), while body weight assessment and correction are unnecessary (3). The BAI method was developed from large samples of Mexican Americans (n = 1,733, 61% women) and African Americans (n = 223, 56.5% women) (3). Studies determining the accuracy of this new method in athletic populations are warranted because of the possibility of it serving as an inexpensive field predictor of BF%. The potential accuracy of this method would be especially important in a female athletic group, as women tend to distribute body fat in the hip region, a variable accounted for by the BAI. Limited investigation has shown strong correlations between BAI and DXA-derived BF% in apparently healthy and postmenopausal obese women (3,6). However, there are no studies exploring the accuracy of the BAI specifically in female athletes. This is an important area of research because negative body image and eating disorders are prevalent in this group (18). Both these conditions could potentially be exacerbated by an inaccurate estimation of BF%. Therefore, the purpose of this study was to cross-validate the BAI for predicting BF% in a group of collegiate female athletes by using DXA as the criterion variable. It was hypothesized that the BAI method would not be an accurate predictor of BF% in this cohort.
Experimental Approach to the Problem
This study was conducted to determine if the BAI could provide a reasonable estimate of BF% in a group of college-aged female athletes (n = 30). The independent variable in this study was the BAI-derived BF% values, which was compared with the independent criterion variable of DXA-derived BF%. All measurements were taken on the same day for each subject.
Thirty female athletes (age 20.0 ± 1.3 years; height 166.6 ± 6.6 cm; weight 66.5 ± 8.8 kg; BMI 22.9 ± 2.6 kg·m−2) from the National Association for Intercollegiate Athletics participated in the study and provided written informed consent as approved by the Institutional Review Board for Human Subjects. The athletes were recruited from the university's soccer (n = 14), tennis (n = 7), and basketball (n = 9) teams. Each subject completed health history questionnaires. Those subjects who were apparently healthy; free from cardiopulmonary, metabolic, and orthopedic disorders; and not pregnant were included in the data collection process.
Height was measured (to the nearest 0.1 cm) with a wall-mounted stadiometer (SECA 220; Seca, Ltd, Hamburg, Germany) with the subjects standing erect, without shoes. Body weight was measured (to the nearest 0.1 kg) with a calibrated digital weighing scale (Tanita BWB-800A; Tanita, Corp., Tokyo, Japan) in light clothing without shoes. Body mass index was calculated as weight in kilograms divided by height in meters squared and rounded to the nearest 0.1 kg·m−2. Hip circumference was measured over nonrestrictive lightweight shorts horizontally at the maximal extension of the gluteus maximus with the subjects standing erect and legs slightly apart (1,3). The mean of 3 measurements was recorded. Predicted BF% was calculated by the BAI method via the following equation: BAI-BF% = [hip circumference/height1.5] − 18.
Criterion BF% was determined by a total body DXA scan (GE Lunar Prodigy, Software version 10.50.086; GE Lunar, Corp., Madison, WI, USA). Before each scan, the DXA was calibrated according to the manufacturer’s instructions using the standard calibration block. Subjects were required to remove all metal objects. The subjects were instructed to assume a supine position with their arms by their sides and palms in a neutral position. During the scan, each subject remained motionless with their knees and ankles held together with Velcro straps.
All data were analyzed with SPSS/PASW Statistics version 18.0 (Somers, NY, USA). The constant error (CE) was determined as the difference between the criterion and predicted BF% values (CE = criterion [DXA] − predicted [BAI] BF%). A paired sample t-test was used to determine if the mean difference was significant. A regression procedure was used to ascertain the correlation coefficient (r), shared variance (R2) and standard error of estimate (SEE) of the BAI compared with the DXA. The following equation was used to establish the total error (TE) of the prediction method:
(14,15). The Bland-Altman method was used to identify the 95% limits of agreement between the criterion and predicted values (4). A priori statistical significance was set at p ≤ 0.05.
All subjects who agreed to participate in the study completed the data collection process. The mean (±SD) predicted BF% by the BAI was 27.1 ± 3.4, whereas the mean (±SD) criterion BF% by the DXA was 26.7 ± 5.9. Paired t-test showed that the mean BF% values were not significantly different (CE = 0.49 ± 5.9%). The regression procedure showed that there was not a significant relationship between the BAI and criterion (r = 0.28, R2 = 0.08, p = 0.14, Figure 1) and also revealed an SEE = 5.78%. In addition, the TE was 5.84%. Bland-Altman plot of the difference in BF% between the BAI and DXA is represented in Figure 2. The limits of agreement (95% confidence intervals) between the DXA and BAI ranged between −10.2 and 11.8%. In addition, the trend between the difference and mean of the 2 methods was statistically significant (r = −0.52, p < 0.01), indicating a greater overestimation of BF% as body fat decreased. An additional regression procedure showed a significant relationship between BMI and DXA (r = 0.49, R2 = 0.24, p < 0.05).
The BAI was introduced in 2011 as a new method of estimating BF% (3). It was primarily developed as a possible replacement to BMI for classifying individual adiposity (3). The advantage of this method is the simplistic calculation, requiring only the variables of hip circumference and height. It was validated in groups of Mexican Americans and African Americans (3), and additional scientific data are extremely limited. A current (May 22, 2012) PubMed search of “Body Adiposity Index” reveals only 12 scientific studies that have been published or are “epub ahead of print” that explore this method in various groups, primarily from general or clinical populations. However, the popularity of this simplistic measure is growing in mainstream. Current (5/22/2012) Google search of the same key phrase (i.e., Body Adiposity Index) results in more than 70,000 sites. Further scientific study is warranted especially among various athletic groups before the BAI could be recommended as a simple inexpensive field parameter of BF%. This study was the first to cross-validate the BAI in a cohort of collegiate female athletes.
Although the mean difference between BAI and DXA was not different with the sample of female athletes, it became less accurate when applied to each individual. There was no significant correlation between the 2 BF% techniques, and the BAI only explained 8% of the variation that occurred in DXA. In addition, the SEE of 5.78% and TE of 5.84% were quite large in comparison with other referenced field predictors of BF% (1). Most skinfold techniques and equations have SEE and TE values ranging from 2.3 to 2.7% when cross-validated in collegiate female athletes (15).
From the scarcity of available research, 2 studies have shown that the BAI correlated more strongly with DXA-determined BF% compared with BMI in nonathletic groups of women (3,6). Another study showed that the BAI and BMI correlated similarly with DXA in apparently healthy subjects (2). In athletes, BMI has not been shown as a valid predictor of BF% because of a greater amount of lean muscle mass compared with nonathletes at the same body weight (13,19,20). However, the present study showed that there was actually a stronger relationship between BMI and DXA (r = 0.49, p < 0.05) than between BAI and DXA (r = 0.28, p > 0.05). This is supported by previous research showing lower SEE and TE than the current findings when predicting BF% in athletic women with equations that use the variables of weight and height (11,15). For instance, previous investigation from the current laboratory reported SEE values ranging from 4.53 to 4.82% and TE values ranging from 4.67 to 5.41% when cross-validating BMI-based equations that predict BF% in college-aged female athletes (11). Additionally, the height- and weight-based equation of Fornetti et al. (12) produced acceptable estimates of BF% (SEE = 2.97% and TE = 3.38%) when compared with multicompartment modeling in collegiate female athletes in another study (15). It should be noted that DXA was chosen as the criterion variable in this study because previous researches suggest that it is appropriate for body composition research in female subjects because of the variation of bone mineral density (16,22). However, the use of this laboratory technique as the criterion could be considered a limitation of the study. Previous investigation has shown that the DXA provided mean BF% values that were 3.70% higher compared with a 5-compartment model in female athletes (14).
Another important finding of the study was the Bland-Altman plot showing large limits of agreement when comparing the BAI with the DXA. These results suggest that the BAI could underestimate BF% by as much as 10.2% or overestimate BF% as much as 11.8%. In addition, and possibly the most notable finding, the association between the difference and the mean was significant and strongly negative. This suggests that the BAI underestimated BF% for the athletes with higher body fatness but overestimated the leaner athletes. To illustrate this finding, Figure 3 represents 2 subjects from the present study’s sample with differing criterion DXA-BF% values (represented by the solid black bars) but similar BAI values (represented by the gray bars). Subject A had a DXA-derived BF% of 30.5 and a BAI-estimated BF% of 25.0. Subject B had a DXA-derived BF% of 15.0 and a BAI-estimated BF% of 25.8. The BAI underestimated BF% by 5.5 in subject A, who had higher body fatness. However, the BAI overestimated BF% by 10.8 in subject B, the leaner athlete.
This finding is probably because of individual differences in the cross-sectional area of the lean skeletal muscles that reside in the hip region (e.g., gluteus maximus). These muscles are highly activated with lower-body resistance exercises (5,21), which athletes commonly perform. Thus, the leaner athletes in the study may have had greater hip circumferences because of larger hip musculatures even though BF% was low. In this regard, the use of BAI for pre- and postassessment of an athlete would show an increased BF% after training when the only possible change that occurred was increased hip girth because of muscular hypertrophy. Athletic women are at a heightened risk of body image and eating disorders (18), which have the potential of being induced by an overestimation of BF%. In consideration of the findings, the BAI is not recommended for predicting individual BF% in college-aged female athletes.
Practitioners should be aware of the emerging BAI method. Predicting BF% with this technique requires measuring only hip circumference and height and the ability to perform the simple calculation. Because it is inexpensive and easy to determine, it has the potential to become widely used in applied sports settings. It is highly important to cross-validate newly developed testing methods before being used with assurance by personal trainers, strength and conditioning specialists, sports nutritionists, and coaches. The results of this study suggest that the BAI method has a wide range of individual error when predicting BF% in female athletes and provides a greater overestimation of BF% as body fat decreases. These are important findings, as inaccurate BF% estimates could place an underweight or overweight female athlete in an acceptable category or could potentially induce a body image or eating disorder. This study also suggests that the BAI has even greater limitations compared with BMI when used as a surrogate of BF% in female athletes. Because of the results of this investigation, practitioners should not use BAI for predicting individual BF% for female athletes.
1. American College of Sports Medicine. ACSM's Guidelines for Exercise Testing and Prescription (8th ed.). Philadelphia, PA: Wolters Kluwer, 2010.
2. 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 24: 828–830, 2011.
3. Bergman RN, Stefanovski D, Buchanan TA, Sumner AE, Reynolds JC, Sebring NG, Xiang AH, Watanabe RM. A better index of body adiposity. Obesity 19: 1083–1089, 2011.
4. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 1: 307–310, 1986.
5. Da Silva EM, Brentano MA, Cadore EL, De Almeida AP, Kruel LF. Analysis of muscle activation during different leg press exercises at submaximum effort levels. J Strength Cond 22: 1059–1065, 2008.
6. 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 52: 145–151, 2013.
7. Esco MR, Olson MS, Williford H. The accuracy of hand-to-hand bioelectrical impedance analysis in predicting body composition in college-age female athletes. J Strength Cond Res 25: 1040–1045, 2011.
8. Esco MR, Olson MS, Williford H. Muscular fitness in young women
and its association to body mass, waist circumference and abdominal skinfold thickness. Res Q Exerc Sport 81: 272–277, 2010.
9. Esco MR, Olson MS, Williford H. Relationship of push-ups and sit-ups to selected anthropometric variables and performance results: A multiple regression study. J Strength Cond Res 22: 1862–1868, 2008.
10. Esco MR, Olson MS, Williford HN, Lizana S, Schaefer GR. The accuracy of various skinfold equations to predict body fat
percentage in college-age female athletes. Gazzetta Italiana Medica 170: 11–18, 2011.
11. Esco MR, Williford H, Russell AR. Cross-validation of BMI-based equations for predicting percent body fat
in female collegiate athletes. J Exerc Physiol 14: 43–52, 2011.
12. Fornetti WC, Pivarnik JM, Foley JM, Fiechtner JJ. Reliability and validity of body composition measures in female athletes. J Appl Physiol 87: 1114–1122, 1999.
13. Kraemer WJ, Torine JC, Silvestre R, French DN, Ratamess NA, Spiering BA, Hatfield DL, Vingren JL, Volek JS. Body size and composition of national football league players. J Strength Cond Res 19: 485–489, 2005.
14. Moon JR, Eckerson JM, Tobkin SE, Smith AE, Lockwood CM, Walter AA, Cramer JT, Beck TW, Stout JR. Estimating body fat
in NCAA Division I female athletes: A five-compartment model validation of laboratory methods. Eur J Appl Physiol 105: 119–130, 2009.
15. Moon JR, Tobkin SE, Smith AE, Lockwood CM, Walter AA, Cramer JT, Beck TW, Stout JR. Anthropometric estimations of percent body fat
in NCAA Division I female athletes: A 4-compartment model validation. J Strength Cond Res 23: 1068–1076, 2009.
16. Mudd LM, Fornetti W, Pivarnik JM. Bone mineral density in collegiate female athletes: Comparisons among sports. J Athl Train 42: 403–408, 2007.
17. National Institutes of Health and National Heart Lung and Blood Institute. Clinical guidelines on the identification, evaluation, and treatment of overweight and obesity in adults: The evidence report. Obes Res 6: 51S–209S, 1998.
18. Nattiv A, Loucks AB, Manore MM, Sanborn CF, Sundgot-Borgen J, Warren MP. American College of Sports Medicine position stand: The female athlete triad. Med Sci Sports Exerc 39: 1867–1882, 2007.
19. Nevill AM, Stewart AD, Olds T, Holder R. Relationship between adiposity and body size reveals limitations of BMI. Am J Phys Anthropol 129: 151–156, 2006.
20. Ode JJ, Pivarnik JM, Reeves MJ, Knous JL. Body mass index as a predictor of percent fat in college athletes and nonathletes. Med Sci Sports Exerc 39: 403–409, 2007.
21. Paoli A, Marcolin G, Nicola P. The effect of stance width on the electromyographical activity of eight superficial thigh muscles during back squat with different bar loads. J Strength and Cond Res 23: 246–250, 2009.
22. Warner ER, Fornetti WC, Jallo JJ, Pivarnik JM. A skinfold model to predict fat-free mass in female athletes. J Athl Train 39: 259–262, 2004.
23. Witt KA, Bush EA. College athletes with an elevated body mass index often have a high upper arm muscle area, but not elevated triceps and subscapular skinfolds. J Am Diet Assoc 105: 599–602, 2005.
24. World Health Organization. Obesity: Preventing and Managing the Global Epidemic. Report of a WHO Consultation on Obesity. Geneva, Switzerland: World Health Organization, 1998.
Keywords:Copyright © 2013 by the National Strength & Conditioning Association.
women; body fat; dual-energy x-ray absorptiometry