Body composition is highly important in regard to overall physical fitness. Typically described as the relative proportion of all the tissues that make up the body, body composition is frequently quantified as body fat percentage (BF%), fat-free mass (FFM), and lean soft tissue (LST). Body composition has been shown to strongly relate to the overall health and fitness levels of athletes (6,7). Specific to females, extremely low body weight, BF%, and FFM are risk factors for the female athlete triad (18). Therefore, practitioners routinely evaluate these parameters when monitoring the outcomes of conditioning programs and tracking changes across a sport season (4,20,28). This information is also important to registered dietitians who work with athletes to estimate total daily energy expenditure and develop personalized dietary interventions based on specific nutritional requirements (25).
Dual-energy x-ray absorptiometry (DXA) has been used in research and laboratory settings as a criterion method to estimate body composition, as it has the ability to evaluate BF%, FFM, bone mineral density (BMD), and regional LST (2,27). Because DXA scans are expensive, involve exposure to radiation, and are primarily found only in exercise physiology laboratories and clinical settings, it is inconvenient for usage within large athletic populations. Therefore, sport practitioners often rely on convenient field measures of body composition, such as bioelectrical impedance analysis (BIA).
Bioelectrical impedance analysis measures the body's resistance and reactance (i.e., impedance) using a painless electrical current that is passed between points of contact, such as the hands and feet. Lean tissue supplies the least resistance to the current because of the electrolytes in body water and the high water content in lean tissue (24). The speed of the current is converted to estimate BF% and FFM using proprietary equations determined by the manufacturer. Traditional BIA methods that pass a single low-frequency (i.e., 50 Hz) current between only 2 poles (e.g., hand-to-hand, hand-to-foot, foot-to-foot) have not been shown to provide comparable body composition measures compared with DXA in female athletes (8,15).
Advanced multifrequency BIA devices pass a large range of frequencies through intra- and extracellular spaces. In addition, multifrequency BIA has the ability to measure impedance separately across 5 different cylinders within the human body. This allows for the analysis of LST throughout the body and within various segments, such as the arms, legs, and trunk. The disadvantage of BIA relates to the dependence upon hydration status, requirement of proper skin preparation, and necessity of precise placement of electrodes. However, because of the technical simplicity, the prevalence of the BIA method is growing within athletic conditioning and sport nutrition programs (16). However, there are no studies to date that have compared multifrequency BIA and DXA in female athletes. The purpose of this investigation was to determine the agreement between multifrequency BIA and DXA for measuring BF%, FFM, and total body and segmental LST in collegiate female athletes. Because of the advanced technology, it is reasonable to hypothesize that the results would show that the 2 devices would provide comparable body composition measures.
Experimental Approach to the Problem
Total and segmental body composition variables were measured in a group of female athletes (n = 45) through DXA and a multifrequency BIA device. The body composition values that were evaluated were as follows: BF%, FFM, and LST (i.e., FFM excluding bone) of the arms (ARMSLST), the legs (LEGSLST), the trunk (TRUNKLST), and the total body (TOTALLST). All measurements were taken on the same day for each subject.
Forty-five female collegiate athletes (age = 21.2 ± 2.0 years, height = 166.1 ± 7.1 cm, weight = 62.6 ± 9.9 kg) from the National Association for Intercollegiate Athletics participated in the study. Each participant provided written informed consent, which was approved by the University's Institutional Review Board for Human Participants. The participants were recruited from the University's soccer (n = 24), basketball (n = 10), cross-country (n = 7), and tennis (n = 4) teams. All participants were not pregnant and free from cardiopulmonary, metabolic, and orthopedic disorders. Data collection for each subject occurred during the morning hours as close as possible to awakening from sleep (i.e., from 7:00 AM to 9:00 AM). Each participant was required to report to the laboratory after an overnight fast, although the consumption of a moderate amount of water (i.e., 12 oz) was allowed. Hydration status was not analyzed in the study. Furthermore, when the participants were scheduled for the day of testing, they were told to avoid consuming stimulants (e.g., caffeine) or depressants (e.g., alcohol) and refrain from strenuous exercise for 24 hours before the data collection. All of the participants verbally agreed to the testing conditions.
Body Composition Procedures
Total and segmental body composition was estimated with the InBody 720 (Biospace Co., Seoul, Korea). Before each measurement, the participants' palms and soles were wiped with an electrolyte tissue. Then, the participants stood on the InBody 720 scale with their soles in contact with the foot electrodes and body weight was measured. Sex, age, and height (which was derived with a wall-mounted stadiometer [SECA 220; Seca, Ltd., Hamburg, Germany]) were manually entered into the instrument by the investigator. Then, the participant grasped the handles with the palm, fingers, and thumb of each hand making contact with the hand electrodes. The body composition analysis was initiated while the participants remained as motionless as possible. The 8-electrode InBody 720 system measured body composition across the entire body and 5 segments (arms, legs, and trunk) by passing multiple frequencies at 5, 50, 250, and 500 kHz from the 8-polar contact points. The scanning time for the InBody 720 was approximately 2 minutes per subject. Test-retest procedures were performed on a separate group of active women (n = 20), which demonstrated that the InBody 720 device provided good reliability for BF% (interclass correlation coefficient [ICC] = 0.99, SEM = 0.16), FFM (0.99, SEM = 0.09), ARMSLST (ICC = 0.99, SEM = 0.02), LEGSLST (ICC = 1.00, SEM = 0.02), TRUNKLST (ICC = 1.00, SEM = 0.02), and TOTALLST (ICC = 1.00, SEM = 0.04), which agreed with a previous study that reported the reliability of the same BIA model in a group of nonathletic men and women (2).
A General Electric Lunar Prodigy DXA (Software version 10.50.086; GE Lunar, Corp., Madison, WI, USA) was used as the criterion measure for total and segmental body composition. The DXA was calibrated before each scan according to the manufacturer's instructions using a standard calibration block. Participants were instructed to remove all metal objects and assume a supine position on the scanning table. During the scan, the participants were required to remain motionless with their arms extended by their sides and the palms in neutral positions. Velcro straps were used to secure the knees and ankles. The scanning time was approximately 7 minutes per participant. All scans were performed by at least 1 of the 3 trained DXA technicians. To minimize the participants' exposure to radiation, test-retest procedures were not performed. However, the DXA used in this study was previously compared with another GE Lunar Prodigy with a different group of subjects (n = 10), and a strong correlations was revealed (r = 0.99).
All data were analyzed with SPSS/PASW version 18.0 (Somers, NY, USA). Mean and SD values were determined for each body composition measure, which were compared between the 2 devices with paired sample t-tests. A Bonferroni-adjusted p value was applied to reduce the chances of obtaining a type I error when multiple pairwise tests were performed. This procedure involved dividing the p value by the number of comparisons that were made (i.e., 0.05/6 = 0.0084). Therefore, the adjusted alpha level for significance of the mean comparisons was determined as p < 0.0084. Cohen's d statistic
determined the effect size of the differences in body composition values (11). Hopkin's scale for determining the magnitude of the effect size was used where 0–0.2 = trivial, 0.2–0.6 = small, 0.6–1.2 = moderate, 1.2–2.0 = large, >2.0 = very large (11). The constant error (CE) was determined as the differences between the 2 devices (CE = InBody 720 − DXA for each body composition parameter). Regression procedures were used to determine the correlation coefficient (r), shared variance (R2), and standard error of estimate (SEE) of each MFBIA measure compared with the DXA. Total error (TE) was determined as
. The method of Bland-Altman was used to identify the 95% limits of agreement between the InBody 720 and DXA body composition values (3). Significant trends in the Bland and Altman plots were determined using an alpha of 0.05.
Results comparing DXA and InBody 720 for BF% and FFM are depicted in Table 1. Compared with the DXA, the InBody 720 provided significantly lower estimates for BF% (p < 0.001) and significantly higher estimates for FFM (p < 0.001), and the Cohen's d statistic showed large and moderate effect sizes, respectively. The correlation coefficients were strong and significant (p < 0.001) between the DXA and InBody 720 for BF% and FFM and the SEE and TE were lowest for FFM. Figures 1 and 2 depict Bland-Altman plots for BF% and FFM, respectively. The 95% confidence intervals (CE ± 1.96 SD of residual scores [InBody 720 − DXA]) for BF% ranged from 2.29% above to −8.95% below the CE of −3.33% (Figure 1) and for FFM ranged from 5.85 kg above to −1.61 kg below the CE of 2.21 kg (Figure 2). The trend between the difference and mean of the 2 devices for BF% (r = 0.10, p = 0.24) and FFM (r = −0.07, p = 0.17) were not significant (Table 1).
Results of the total and segmental LST values are depicted in Table 1. There were no significant differences between the 2 devices for ARMSLST (p = 0.371), TRUNKLST (p = 0.567), LEGSLST (p = 0.049), and TOTALLST (p = 0.520). The correlation coefficients were strong and significant (p < 0.001) between the DXA and InBody 720 for all LST values. Figure 3 depicts Bland-Altman plots for the TOTALLST values. The 95% confidence intervals for TOTALLST ranged from 4.02 kg above to −4.44 kg below the CE of −0.21 kg. The trends between the difference and mean of the 2 devices for ARMSLST (r = −0.08, p = 0.32), TRUNKLST (r = −0.06, p = 0.53), LEGSLST (r = 0.11, p = 0.25), and TOTALLST (r = 0.06, p = 0.36) were not significant (Table 1).
This investigation sought to determine the agreement between multifrequency BIA and DXA for measuring BF%, FFM, and total body and segmental LST in collegiate female athletes. The findings showed that the InBody 720 underestimated BF% by 3.33% and overestimated FFM by 2.12 kg compared with DXA. The strong correlations, small SEE and TE, and tight limits of agreement suggest that the InBody 720 may consistently provide lower BF% and higher FFM values than DXA in college-age female athletes. However, when comparing LST values between the 2 devices, there were no significant mean differences between DXA and the InBody 720 for ARMSLST, TRUNKLST, LEGSLST, and TOTALLST. In addition, compared with DXA, each LST value of the InBody 720 showed strong correlations, small SEE and TE, and tight limits of agreement. Therefore, the original hypothesis was partially accepted as the findings indicated comparable measures between DXA and InBody 720 for the LST variables but not for BF% and FFM.
The few studies that have compared InBody 720 BIA and DXA have primarily included nonathletic groups and have yielded conflicting results, with some studies showing no significant mean differences and overall good agreement between the 2 devices (5,14), whereas others reported opposite findings (2,27). This discrepancy may be related to differences in the body mass and fat status of the studied samples. For example, Shafer et al. (26) indicated that when compared with DXA, the InBody 320 (a similar but less advanced version of the device used in this study) underestimated BF% and overestimated FFM in normal weight subjects but overestimated BF% and underestimated FFM in obese subjects (26). Others have provided similar results, showing a tendency for the InBody 720 to provide lower and higher BF% values in leaner and overfat subjects, respectively, when compared with DXA (5,14,27). Proportional bias has also been demonstrated with the agreement of FFM values, with the InBody 720 providing higher values in leaner subjects but lower values in obese subjects when compared with DXA (14,27). Therefore, a small range of total body mass and BF% may exist in women, in which the InBody 720 provides the best agreement with DXA (26). This possibility could explain why the InBody 720 provided significantly lower BF% and higher FFM values compared with DXA in the current heterogeneous sample of active female athletes with normal body mass index (BMI) levels.
The InBody prediction equations are not available to the public and cannot be manipulated by the user. Adjusting the regression equations for specific populations (e.g., female athletes) may allow for more accurate body composition measures compared with the equation that is preset by the InBody manufacturer. Recently, Aandstad et al. (1) measured BF% using DXA, the InBody 720, and a single frequency hand-to-foot BIA device that predicted BF% with 10 different equations (adjusted by the investigators) in a group of first-year military cadets. The characteristics of the female sample (n = 26) were similar to this study (i.e., age = 21.0 ± 4.0 years, BMI = 24 ± 2.5 kg·m2, and DXA BF% = 25.6 ± 4.7%) (1). Compared with DXA, the InBody 720 provided significantly lower values of 1.9 with 95% limits of agreement of 5.2% (1), which is comparable to the current findings. However, of the various BF% predict equations analyzed with the single frequency BIA device, 4 provided no significant mean differences compared with DXA and demonstrated tighter limits agreement (95% limits of agreement ranging from 4.0 to 4.6%) compared with the InBody 720 analyzed in the investigation by Aandstad et al. (1) and in the current investigation. Therefore, it is reasonable to consider that having the ability to modify the preset equation of the InBody 720 may increase its accuracy for predicting BF% within specific populations. Additional research is needed to investigate this hypothesis.
An explanation of the significant differences in FFM between DXA and InBody 720 may also be because of how each account for bone content. Dual-energy x-ray absorptiometry is considered the “gold standard” for BMD and bone mineral content measurements (9,13). However, the InBody 720 predicts bone mass with an equation that is based on DXA normative values established from the general population (27). Because BMD and bone mineral content vary considerably in a female athletic population (29), a constant general value to predict BMD and bone mineral content may not be appropriate. This postulation is supported by the findings related to LST for which bone is not included. When the LST values were compared between the 2 devices in this study, excellent agreement across all parameters was observed. Therefore, when excluding all of the bone in the body, DXA and the InBody 720 seem to have excellent agreement for measuring total and segmental LST (i.e., FFM excluding BMD) in the current sample of female athletes.
A primary limitation of our study was the lack of a hydration status measurement. However, the participants followed a strict testing protocol: for example, no food consumption after midnight prior, consume approximately a cup and a half of water before testing, no strenuous activity, caffeine, or alcohol 24 hours before data collection, and come to the laboratory as close to awakening from sleep as possible. Additionally, acute changes in hydration status (hyperhydrated, dehydrated, or euhydrated) should affect both DXA and the InBody 720 as both methods assume a constant hydration of approximately 73% for FFM (16,27). This may be a comparable source of error for both devices for FFM and BF% (21,23), especially because FFM hydration levels deviate considerably in active individuals (12,16,19). Therefore, the current results may be independent of hydration status. Nevertheless, the focus of this study was to compare both devices in the same individuals in the same hydrated state, and because the measurements were taken the same day with the same hydration status, a direct comparison can be made without hydration influencing the relationship between techniques. Still, due to the aforementioned issues with hydration for both devices, it has been suggested that multicompartment models that include measures of total body water are the preferred method of body composition validation studies in athletes (16). However, the current investigation sought to compare 2 methods and not validate 1 method to a criterion. Therefore, the lack of measuring acute hydration status and FFM hydration of the subjects should be considered a minor study limitation.
One study that compared the InBody 720 to a 4-compartment model found that it underestimated BF% by almost 3% and provided a SEE of 4.85% in a group of women from a general population (10). When DXA was compared with a 5-compartment model, Moon et al. (17) showed that it overestimated BF% by 3.71% and provided a wide range in individual error (i.e., ±6.3%) in a group of female athletes. Therefore, the FFM and BF% errors associated with DXA may be similar to the errors observed from the InBody 720 device when compared with criterion multicompartment model that accounts for body water. Future investigations should crossvalidate the InBody 720 and DXA methods in collegiate female athletes using the multicompartmental approach that includes a measure of total body water, while accounting for hydration status. Additional research is also warranted to determine the effects of acute changes in hydration status for both methods.
Dual-energy x-ray absorptiometry is a commonly used laboratory measure for evaluating the body composition. However, this method is inconvenient for monitoring athletes in field settings. The InBody 720 is a multifrequency BIA that has potential to serve as an alternative to DXA technology because of its ability to estimate BF%, FFM, and LST throughout the body and within various segments (arms, legs, and trunk). Although this study found significant differences in BF% and FFM, the InBody 720 and DXA seem to provide excellent agreement for measuring LST suggesting that the InBody 720 can be used as an alternative to DXA in athletic women. Because of strength and conditioning training, athletes may possess large variations in soft tissue (muscle) within the arms, legs, and trunk compared with a general nonathletic population (22). Therefore, the ease of use for the InBody 720 over the DXA and the findings of this investigation suggest an advantage for using the InBody 720 in female athletes to estimate total and segmental LST in female athletes and could serve as a useful tool to rapidly track changes during diet and exercise intervention. However, future research is needed to determine the relationship between DXA and the InBody 720 and their ability to track changes compared with a criterion method in both individuals and groups of athletes.
1. Aandstad A, Holtberget K, Hageberg R, Holme I, Anderssen AA. Validity and reliability of bioelectrical impedance analysis and skinfold thickness in predicting body fat in military personnel. Mil Med 179: 208–217, 2014.
2. Anderson LJ, Erceg DN, Schroeder E. Utility of multifrequency bioelectrical impedance compared with dual-energy x-ray absorptiometry for assessment of total and regional body composition varies between men and women
. Nutr Res 32: 479–485, 2012.
3. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet 327: 307–310, 1986.
4. Carling C, Orhant E. Variation in body composition in professional soccer players: Interseasonal and intraseasonal changes and the effects of exposure time and player position. J Strength Cond Res 24: 1332–1339, 2010.
5. Dolezal BA, Lau MJ, Abrazado M, Storer TW, Cooper CB. Validity of two commercial grade bioelectrical impedance analyzers for measurement of body fat percentage. J Exerc Physiol Online 16: 74–83, 2013.
6. Esco MR, Olson MS, Williford HN. Relationship of push-ups and sit-ups tests to selected anthropometric variables and performance results: A multiple regression study. J Strength Cond Res 22: 1862–1868, 2008.
7. Esco MR, Olson MS, Williford HN. The relationship between selected body composition variables and muscular endurance in women
. Res Q Exerc Sport 81: 272–277, 2010.
8. Esco MR, Olson MS, Williford HN, Lizana S, Russell A. 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.
9. Genant HK, Cooper C, Poor G, Reid I, Ehrlich G, Kanis J, Khaltaev N. Interim report and recommendations of the World Health Organization task-force for osteoporosis. Osteoporos Int 10: 259–264, 1999.
10. Gibson AL, Holmes JC, Desautels RL, Edmonds LB, Nuudi L. Ability of new octapolar bioimpedance spectroscopy analyzers to predict 4-component-model percentage body fat in Hispanic, black, and white adults. Am J Clin Nutr 87: 332–338, 2008.
11. Hopkins WG, Marshall SW, Batterham AM, Hanin J. Progressive statistics for studies in sports medicine and exercise science. Med Sci Sports Exerc 41: 3–12, 2009.
12. Hutcheson L, Hutcheson L, Berg KE, Prentice E. Body impedance analysis and body water loss. Res Q Exerc Sport 59: 359–362, 1988.
13. Kanis JA, Glüer CC. An update on the diagnosis and assessment of osteoporosis with densitometry. Osteoporos Int 11: 192–202, 2000.
14. Ling CHY, de Craen AJM, Slagboom PE, Gunn DA, Stokkel MPM, Westendorp RGJ, Maier AB. Accuracy of direct segmental multi-frequency bioimpedance analysis in the assessment of total body and segmental body composition in middle-aged adult population. Clin Nutr 30: 610–615, 2011.
15. Loenneke JP, Wilson JM, Wray ME, Barnes JT, Kearney ML, Pujol TJ. The estimation of the fat free mass index in athletes. Asian J Sports Med 3: 200–203, 2012.
16. Moon JR. Body composition in athletes and sports nutrition: An examination of the bioimpedance analysis technique. Eur J Clin Nutr 67: 54–59, 2013.
17. Moon JR, Eckerson JM, Tobkin SE, Smith AE, Lockwood CM, Walter AA, 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.
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. O'Brien C, Young AJ, Sawka MN. Bioelectrical impedance to estimate changes in hydration status. Int J Sports Med 23: 361–366, 2002.
20. Ostojic SM. Seasonal alterations in body composition and sprint performance of elite soccer players. J Exerc Physiol 6: 24–27, 2003.
21. Pietrobelli A, Wang Z, Formica C, Heymsfield SB. Dual-energy x-ray absorptiometry: Fat estimation errors due to variation in soft tissue hydration. Am J Physiol 274: E808–E816, 1998.
22. Prior BM, Modlesky CM, Evans EM, Sloniger MA, Saunders MJ, Lewis RD, Cureton KJ. Muscularity and the density of the fat-free mass
in athletes. J Appl Physiol (1985) 90: 1523–1531, 2001.
23. Saunders MJ, Blevins JE, Broeder CE. Effects of hydration changes on bioelectrical impedance in endurance trained individuals. Med Sci Sports Exerc 30: 885–892, 1998.
24. Scharfetter H, Schlager T, Stollberger R, Felsberger R, Hutten H, Hinghofer-Szalkay H. Assessing abdominal fatness with local bioimpedance analysis: Basics and experimental findings. Int J Obes Relat Metab Disord 25: 502–511, 2001.
25. Segal KR. Use of bioelectrical impedance analysis measurements as an evaluation for participating in sports. Am J Clin Nutr 64: 469S–471S, 1996.
26. Shafer KJ, Siders WA, Johnson LK, Lukaski HC. Validity of segmental multiple-frequency bioelectrical impedance analysis to estimate body composition of adults across a range of body mass indexes. Nutrition 25: 25–32, 2009.
27. Sillanpaa E, Cheng S, Hakkinen K, Finni T, Walker S, Pesola A, Ahtiainen J, Stenroth L, Selanne H, Sipila S. Body composition in 18- to 88-year-old adults—Comparison of multifrequency bioimpedance and dual energy x-ray absorptiometry. Obesity 22: 101–109, 2014.
28. Silvestre R, Kraemer WJ, West C, Judelson DA, Spiering BA, Vingren JL, Maresh CM. Body composition and physical performance during a National Collegiate Athletic Association Division I men's soccer season. J Strength Cond Res 20: 962–970, 2006.
29. 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, 2005.