Share this article on:

Comparison of methods for assessing body composition changes during weight loss


Medicine & Science in Sports & Exercise: March 2002 - Volume 34 - Issue 3 - p 497-502
APPLIED SCIENCES: Physical Fitness and Performance

WEYERS, A. M., S. A. MAZZETTI, D. M. LOVE, A. L. GÓMEZ, W. J. KRAEMER, and J. S. VOLEK. Comparison of methods for assessing body composition changes during weight loss. Med. Sci. Sports Exerc., Vol. 34, No. 3, pp. 497–502, 2002.

Purpose Four cross-sectional studies have reported that percent body fat (%BF) measured by dual-energy x-ray absorptiometry (DXA) is significantly higher compared with values obtained with air displacement plethysmography (ADP) using the Bod Pod® in normal-weight individuals. This study was performed to confirm these findings in an overweight population and to assess whether DXA and ADP detected similar changes in body composition after moderate weight loss.

Methods Twelve women (42 ± 8 yr) and 10 men (40 ± 11 yr) had their %BF, fat mass (FM), and fat-free mass (FFM) measured using DXA and ADP before and after an 8-wk weight-loss program involving moderate energy restriction and exercise.

Results Body weight decreased significantly in women (−4.3 ± 3.4 kg) and men (−4.7 ± 3.1 kg). There were significant method (ADP vs DXA) and time (pre and post) effects but no method by time or gender interactions. Methods were significantly different in estimating %BF, FM, and FFM with ADP estimates of %BF and FM being lower and estimates of FFM higher than corresponding DXA values (P = 0.000). There were significant correlations accounting for a high degree of the shared variance between DXA and ADP (r = 0.98 to 0.99) for %BF, FM, and FFM and lower correlations for the changes in %BF (r = 0.66), FM (r = 0.86), and FFM (r = 0.34). In response to weight loss, the mean changes in %BF, FM, and FFM were not significantly different between methods (P > 0.05).

Conclusion Both DXA and ADP measure changes in body composition after small to moderate weight loss to the same extent and with similar sensitivity.

Human Performance Laboratory, Department of Kinesiology, University of Connecticut, Storrs, CT

Submitted for publication October 2000.

Accepted for publication June 2001.

Reliable and sensitive measurement of body composition is critical to accurately interpret the findings from research studies that involve weight-loss or weight-gain interventions. Comparisons between common body composition techniques generally yield high correlations for percent body fat (%BF) but may differ in the absolute values obtained. For example, four recent cross-sectional studies have each demonstrated that %BF values estimated from dual-energy x-ray absorptiometry (DXA) and air displacement plethysmography (ADP) using the Bod Pod® are highly correlated but DXA overestimates %BF compared WITH ADP by 2.0–2.9% in normal-weight individuals (2,10,11,17). Although these systematic differences in %BF between DXA and ADP have been observed in normal-weight volunteers, no research has been done comparing how these two methods compare in overweight persons and whether DXA and ADP detect similar changes in body composition that occur with moderate weight loss.

During the past decade, the prevalence of overweight and obesity-related diseases has increased dramatically. This trend has contributed to an increasing emphasis placed on the accurate assessment of body composition, especially for evaluation of weight loss interventions in research settings (i.e., dietary and nutritional supplement manipulations, exercise training programs, etc.). Hydrostatic weighing has been considered the “gold standard” for measuring body composition (15); however, it is not without limitations. Hydrostatic weighing is a two-component method that estimates fat mass (FM) and fat-free mass (FFM) from total body density and thus does not account for potential differences in the density of FFM. Hydrostatic weighing requires subjects to exhale maximally while submerged under water and thus is inconvenient and uncomfortable for many individuals. Consequently, DXA and ADP have emerged as common techniques to assess body composition.

ADP using the Bod Pod (Life Measurement Instruments, Concord, CA) is a relatively new technique that provides an estimate of %BF from body density without being submerged under water. This two-component technique involves sitting quietly in an air chamber while the volume of air displaced is measured. This method is noninvasive, comfortable, and quick. Dual-energy x-ray absorptiometry is an alternative method to assess body composition that provides an estimate of bone mineral content and soft tissue (fat and lean tissue). This three-component method provides a noninvasive, comfortable, and quick estimate of body composition that is independent of body density. Studies examining both DXA and ADP have revealed a high degree of reliability (i.e., similar results are obtained on repeated trials), and both methods are valid in reference to hydrostatic weighing (i.e., there is a strong agreement between methods) (1,3,9,10,12,21,22).

The purpose of this investigation was to compare estimates of %BF, FM, and FFM between DXA and ADP in overweight men and women. Another purpose was to evaluate and compare the sensitivity of these two body composition techniques for detecting small changes in body composition in response to an 8-wk weight-loss program. We hypothesized DXA would overestimate %BF relative to ADP in both genders and that both methods would detect similar changes in body composition after weight loss.

Back to Top | Article Outline


Experimental design.

Body composition was assessed using DXA and ADP before and after an 8-wk weight-loss program that included nutritional, exercise, and behavior modification components in overweight men and women.

Back to Top | Article Outline


Twelve overweight women (age, 42.4 ± 7.7 yr) and 10 overweight men (39.7 ± 11.0 yr) participated in the study. All subjects were required to have a body mass index greater than 25 kg·m−2 (men 30.0 ± 3.4; women 30.8 ± 7.0 kg·m−2). Initially, men had significantly greater body mass, fat-free mass, and waist to hip ratios and a significantly lower percent body fat compared with women. Based on initial testing, percent body fat ranged from 20 to 40% in men and from 35 to 57% in women. All subjects were healthy and demonstrated no endocrine, orthopedic, or any other pathological disorders, except for being overweight. Before the study, about one-half of the subjects were sedentary, several walked 2–3 times per week, and a few were involved in recreational type exercise (biking, tennis, basketball) and/or weight training 2–3 times per week. Prior approval by the Institutional Review Board for Use of Human Subjects was obtained for the investigation. Each subject had the risks of the experiment explained to them and signed an informed consent document.

Back to Top | Article Outline

Weight-loss intervention.

The weight-loss program followed general guidelines for reducing dietary fat and increasing exercise as well as instruction on behavior modification and nutritional supplements. Our objective was to create a 5- to 10-kg weight loss in each subject by moderate caloric restriction and increased energy expenditure over the 8 wk. Subjects consumed a self-selected diet comprised of commercially available food products in a free-living environment. To facilitate weight loss, subjects were required to attend a weekly group-format nutrition-education meeting led by a registered dietitian. Body weight was recorded and charted at each weekly meeting to ensure a steady rate of weight loss (0.5–1.0 kg·wk−1) over the 8 wk.

Back to Top | Article Outline

Experimental testing procedures.

After a 12-h overnight fast, subjects reported to the laboratory between 7:00 and 9:00 a.m. to have their body composition measured using DXA and ADP. Subjects were not permitted to drink or exercise for 3 h before testing, and each subject voided their bladder 1 h before testing. Body mass was determined on a calibrated clinical scale with subjects wearing a bathing suit. The DXA testing was always performed first, and ADP testing was performed within 10 min of DXA testing. Subjects wore an identical Lycra swimsuit or Lycra biking shorts and removed all metal objects and jewelry for the DXA and ADP testing.

Back to Top | Article Outline


DXA measurements were made with a total body scanner (ProdigyTM, Lunar Corporation, Madison, WI) that uses a constant potential x-ray source of 76 kVp and a cerium filter that produces dual-energy peaks of 38 and 62 keV. The soft tissue mass (fat and lean tissue) is measured pixel-by-pixel as a beam of photons penetrates the subject’s body (4,5,14). Subjects remained motionless in the supine position for approximately 6 min while the scanning arm of the DXA passed over their body from head to toe in parallel 1-cm strips. Percent body fat from the DXA testing was subsequently calculated as fat tissue mass divided by the total soft tissue mass plus the estimated BMC. Fat-free mass was calculated as lean tissue plus BMC. All analyses were performed by the same technician using computer algorithms (software version 1.20.020). Quality assurance was assessed by analyzing a phantom spine provided by the company, and daily calibrations were performed before all scans using a calibration block provided by the manufacturer. The coefficient of variation for percent fat from 12 repeated scans on a group of men and women ranging from 11.2 to 34.3% in our laboratory was 1.0%.

Back to Top | Article Outline


During testing of body composition using the Bod Pod, each subject wore a swim cap in addition to the identical clothing as worn during the DXA testing. Two trials were performed on each subject. Subjects sat quietly in the chamber of the Bod Pod while their raw body volume was measured consecutively until two values within 150 mL were obtained. If greater than three raw body volumes were necessary, then two to three additional body volume measurements were obtained after recalibrating the Bod Pod. Thoracic gas volume was estimated by having subjects perform the panting maneuver (6,12). Briefly, the subject breathed normally for three breathing cycles through a tube connected to the internal system while wearing a nose-clip. At the midpoint of an exhalation, the airway tube was momentarily occluded, and the subject was signaled by the technician (same for all subjects at all testing sessions) to perform three small puffs of air into the tube while maintaining a tight seal around the end of the tube. Once a figure of merit of <1 was met (a measure of compliance to the breathing protocol), the measured thoracic gas volume was used to calculate a corrected body volume (corrected body volume = raw body volume − thoracic gas volume). In the event that a figure of merit of <1 was not met, an estimated thoracic gas volume was used for subsequent calculations (13). Body density was calculated as body mass divided by the corrected body volume. Body density was used to calculate %BF (18). Fat mass was calculated as %BF multiplied by total body mass obtained on the digital scale. Fat-free mass was calculated as body mass minus FM. Before each trial, the Bod Pod was calibrated using a 50.341-L cylinder. Also the Bod Pod scale, which is used to weigh each subject before the trial, was calibrated each morning by using two standard 10-kg weights. The coefficient of variation for percent fat from 44 paired trials in this study was 1.5%.

Back to Top | Article Outline

Statistical analyses.

Data were analyzed using a repeated measures ANOVA for two methods (DXA and ADP) by time (pre and post) with gender as a between subjects factor (Version 10.0.5, SPSS Inc., Chicago, IL). Pearson’s product-moment correlation coefficients were calculated to assess relationships between %BF, FM, and FFM obtained using the DXA and ADP. Significance in this study was set at P ≤ 0.05.

Back to Top | Article Outline


Body composition results are presented in Table 1. The weight-loss program was successful as indicated by the significant and similar decrease in body mass in both women and men. Although highly correlated (r = 0.998), body weights obtained from the scale were significantly higher than the corresponding weights obtained from DXA for women before weight loss but not for men. Total body BMC estimated from DXA was not significantly different before and after weight loss in women (2.81 ± 0.36 to 2.83 ± 0.33 kg) or men (3.48 ± 0.34 to 3.48 ± 0.35 kg). Similarly, total body bone mineral density was not significantly different before and after weight loss in women (1.206 ± 0.076 to 1.216 ± 0.74 g·cm−2) or men (1.257 ± 0.088 to 1.256 ± 0.10 g·cm−2).

Table 1

Table 1

The ANOVA table for percent body fat is presented in Table 2 (in terms of significance the ANOVA tables for FM and FFM were similar). There were significant method and time effects for %BF, FM, and FFM but no method by time or gender interactions, indicating that DXA and ADP measure body composition differently but they detect similar changes in response to weight loss. Compared with DXA, estimates for %BF and FM obtained from ADP were significantly lower and estimates of FFM were significantly higher compared with corresponding values obtained from DXA. These relationships were evident before and after weight loss (Fig. 1). Percent body fat from ADP was higher than DXA in 73% and 86% of the subjects before and after weight loss, respectively.

Table 2

Table 2



There were no significant differences in changes in %BF, FM, and FFM in response to weight loss between methods. Correlations for changes in body composition after weight loss between DXA and ADP were significant for %BF (r = 0.66) and FM (r = 0.86) but not for FFM (r = 0.34). Figure 2 shows the change in %BF pre to post for both ADP and DXA across the range of initial %BF levels. This plot shows no clear pattern in the change in %BF from ADP and DXA in individuals at low and high initial %BF levels. Figure 3 shows the 95% confidence limits for %BF estimated by DXA and ADP pre to post weight loss, which further supports the finding that ADP and DXA detect similar changes in %BF in response to weight loss.





Back to Top | Article Outline


This study compared changes in body composition after weight loss assessed with DXA and ADP. Our primary finding was that DXA and ADP measured similar absolute changes in %BF, FM, and FFM after weight loss in women and men. However, DXA estimates of %BF were slightly but statistically higher than corresponding ADP values. Despite small but significant differences between measurements from the DXA and ADP, %BF values before and after weight loss were highly correlated between the DXA and ADP across a relatively wide range of body fat levels (20–57%) in middle-aged women and men. As a result of these differences in %BF, DXA estimates of FM were significantly higher and estimates of FFM significantly lower both before and after weight loss when compared with corresponding ADP values.

In this study, DXA overestimated %BF and FM whereas it underestimated FFM compared with corresponding ADP values. The difference in %BF of 1.8–2.7% between DXA and ADP in this study is remarkably similar to data in four previous cross-sectional studies that reported mean differences between DXA and ADP ranging from 2.0 to 2.9% (3,11,12,18). This consistent and systematic overestimation of %BF by ADP relative to DXA does not appear to be influenced by gender, age, or degree of adiposity (2,10,11,17). Compared with DXA, higher %BF values were obtained using ADP in relatively lean (12–15%BF) football players (2), normal-weight middle-aged white men (17), normal-weight middle-aged men and women (10), and boys and girls 10–18 yr of age (11). Our data confirm these differences in %BF between DXA and ADP in an overweight group of men and women before and after moderate weight loss. The differences between ADP and DXA in this study were between approximately −7.4% and +3.2% with 95% confidence, which is remarkably similar to the results in a similar study comparing ADP and DXA that reported differences between −10.4% and +4.4% with 95% confidence (10). Also in agreement with our data, there were no relationships between these differences and the initial level of body fat (10).

Several explanations have been put forward to explain these systematic differences in %BF between DXA and ADP. Incomplete scanning of the entire body could account for errors in the DXA estimates of %BF (7). This is an unlikely explanation in this study because all subjects were completely within the scanning area of the DXA. Inaccuracies in the measurement of thoracic gas volume measurements could also account for differences in %BF by using ADP. Using an estimated thoracic gas volume (as opposed to thoracic gas volume measured on the Bod Pod) may slightly enhance the agreement between %BF obtained from ADP and DXA (2); however, the effect is small. There were 88 individual ADP trials in our study, 78% of which used a measured and 22% used an estimated value for the thoracic gas volume. McCrory et al. (13) compared estimated and measured thoracic gas volumes and observed no significant difference between the methods. A mean difference of 54 mL corresponded to a small difference of only 0.2% in %BF. In support of the findings of McCrory et al.(13), the difference between measured versus estimated thoracic gas volumes resulted in an average %BF difference of only 0.07% in this study. Thus, the use of measured and estimated thoracic gas volumes accounted for only minimal error in %BF measurements using ADP. Variances in the density of FFM may also explain some of the discrepancies between ADP and DXA, with ADP presumably being affected to a greater extent than DXA by potential variances in water, protein, and mineral that comprise FFM.

Data from several studies have shown that moderate weight loss can result in an underestimation of BMC and bone density from DXA, which could potentially explain the discrepancy in %BF between DXA and ADP (19,20). In contrast, we did not observe any change in total body BMC or bone density in men or women. Potential reasons to explain why we did not observe changes in bone mass include the short-term nature of the intervention (8 wk), the incorporation of regular exercise that may have counteracted a loss in bone mass, and/or the moderate weight loss (approximately −4.5 kg). Furthermore, if the discrepancy in %BF between ADP and DXA were due to real or artifactual bone changes resulting from weight loss, this still does not explain the significant difference between these methods before weight loss in our study or that of other cross-sectional studies (2,10,11,17). The differences between ADP and DXA could be due to some of the assumptions underlying the physical concepts of DXA, which are discussed in detail by Pietrobelli et al. (14). For example, DXA can only resolve the fractional masses of two components. Pixels containing a mixture of bone, lean soft tissue, and fat tissue are difficult to assess and require the use of complex algorithms that involve using a %BF value estimated from neighboring pixels. The accuracy of this assumption in overweight individuals versus normal weight individuals is unknown.

Despite significant differences observed in body composition data between various methods in cross-sectional studies, in the current study, measurements of the changes in body composition after weight loss did not differ significantly between DXA and ADP. Body composition measurements from both DXA and ADP have previously been shown to be reliable on repeated occasions (3,10,20,21). Because DXA and ADP are consistent, their measurements of small reductions in %BF (about 2%) during a short-term weight loss intervention in overweight adults are similar despite small differences in absolute %BF values between techniques. Because subjects only lost about 2% body fat in this study, it is still possible that larger reductions in percent fat could be detected differently by ADP and DXA. After a 10-wk low-carbohydrate weight-loss diet in women who were overweight (∼40% body fat), Ryde et al. (16) reported that DXA underestimated small changes (<3%) and overestimated large changes (∼6%) in %BF as compared with hydrostatic weighing. Conversely, Hendel et al. (7) reported similar measurements of the changes in %BF, FM, and FFM between DXA and total body potassium in obese women and men. Houtkouper et al. (8) recently reported that DXA was a more sensitive method for assessing small changes in body composition resulting from an exercise-training program compared with hydrostatic weighing. Our data indicate ADP or DXA detect similar changes in %BF, FM, and FFM in response to moderate weight loss.

The advantages and disadvantages of using ADP and DXA in a research or field setting should be considered. Both methods are relatively easy to perform and comfortable for the subject. Advantages of ADP over DXA include less initial investment and no exposure to radiation (although the radiation exposure for DXA is extremely low). An advantage of DXA is that absolute measurement of fat, lean, and bone mass is performed, whereas ADP simply measures a relative concentration that is used to estimate lean and fat tissue. Also, DXA provides information on BMC/density and regional body composition.

In summary, the findings of this study indicate that DXA and ADP detect similar changes in %BF, FM, and FFM after a small to moderate weight loss (4 to 5 kg) in overweight women and men. Our findings also support previous reports that DXA overestimates %BF and FM and underestimates FFM compared with ADP (2,10,11,17). There were high correlations between DXA and ADP for %BF, FM, and FFM. The correlations between DXA and ADP for changes in %BF, FM, and FFM were lower. The mean changes in %BF, FM, and FFM were similar between methods. Given both DXA and ADP are quick, convenient, relatively easy to use, and comfortable for the subject, our data lend support to the use of either DXA or ADP for measuring small to moderate changes in body composition in overweight adults over time.

We thank a dedicated group of subjects for making this study possible. This study was partially sponsored by Natural Alternatives International, San Marcos, CA.

Address for correspondence: Jeff S. Volek, Ph.D., R.D., Assistant Professor, Human Performance Laboratory, Department of Kinesiology, University of Connecticut, Storrs, CT 06269; Email: jvolek@uconnvm.

Back to Top | Article Outline


1. Clark, R., J. Kuta, and J. Sullivan. Prediction of percent body fat in adult males using dual-energy x-ray absorptiometry, skinfolds, and hydrostatic weighing. Med. Sci. Sports Exerc. 25: 528–535, 1993.
2. Collins, M., M. Millard-Stafford, P. Sparling, et al. Evaluation of the BOD POD® for assessing body fat in collegiate football players. Med. Sci. Sports Exerc. 31: 1350–1356, 1999.
3. Dempster, P., and S. Aitkens. A new air displacement method for the determination of human body composition. Med. Sci. Sports Exerc. 27: 1692–1697, 1995.
4. Gotfredsen, A., J. Jensen, J. Borg, and C. Christiansen. Measurements of lean body mass and total body fat using dual photon absorptiometry. Metabolism 35: 88–93, 1986.
5. Gotfredsen, A., J. Borg, C. Christiansen, and R. B. Mazess. Total body bone mineral in vivo by dual photon absorptiometry II, accuracy. Clin. Physiol. 4: 357–362, 1984.
6. Gundlach, B., and G. Visscher. The plethysmometric measurement of total body volume. Hum. Biol. 58: 783–799, 1986.
7. Hendel, H. W., A. Gotfredsen, T. Andersen, L. Hojgaard, and J. Hilsted. Body composition during weight loss in obese patients estimated by dual energy x-ray absorptiometry and by total body potassium. Int. J. Obes. 20: 1111–1119, 1996.
8. Houtkooper, L. B., S. B. Going, J. Sproul, R. M. Blew, and T. G. Loham. Comparison of methods for assessing body-composition changes over 1 y in postmenopausal women. Am. J. Clin. Nutr. 72: 401–406, 2000.
9. Johansson, A., A. Forslund, A. Sjodin, H. Mallmin, L. Hambraeus, and S. Ljunghall. Determination of body composition –a comparison of dual-energy x-ray absorptiometry and hydrodensitometry. Am. J. Clin. Nutr. 57: 323–6, 1993.
10. Levenhagen, D., M. Borel, D. Welch, et al. A comparison of air displacement plethysmography with three other techniques to determine body fat in healthy adults. J. Parenter. Enteral Nutr. 23: 293–299, 1999.
11. Lockner, D. W., V. H. Heyward, R. N. Baumgartner, and K. A. Jenkins. Comparison of air-displacement plethysmography, hydrodensitometry, and dual X-ray absorptiometry for assessing body composition of children 10 to 18 years of age. Ann. N. Y. Acad. Sci. 904: 72–78, 2000.
12. McCrory, M., T. Gomez, E. Bernauer, and P. Mole. Evaluation of a new air displacement plethysmograph for measuring human body composition. Med. Sci. Sports Exerc. 27: 1686–1691, 1995.
13. McCrory, M., P. Mole, K. Dewey, and E. Bernauer. Body composition by air-displacement plethysmography by using predicted and measured thoracic gas volumes. J. Appl. Physiol. 84: 1475–1479, 1998.
14. Pietrobelli, A., C. Formica, Z. Wang, and S. B. Heymsfield. Dual-energy X-ray absorptiometry body composition model: review of physical concepts. Am. J. Physiol. 271: E941–E951, 1996.
15. Roubenoff, R., J. Kehayias, B. Dawson-Hughes, and S. Heymsfield. Use of dual-energy x-ray absorptiometry in body-composition studies: not yet a “gold standard.” Am. J. Clin. Nutr. 58: 589–591, 1993.
16. Ryde, S. J. S., R. Eston, M. A. Laskey, C. J. Evans, and D. A. Hancock. Changes in body fat: measurements by neutron activation, densitometry and dual energy x-ray absorptiometry. Appl. Radiat. Isot. 49: 507–509, 1998.
17. Sardinha L., T. Lohman, P. Teixeira, D. Guedes, and S. Goings. Comparison of air displacement plethysmography with dual-energy X-ray absorptiometry and 3 field methods for estimating body composition in middle-aged men. Am. J. Clin. Nutr. 68: 786–93, 1998.
18. Siri, W. E. Body composition from fluid spaces and density: analysis of methods. In: Techniques for Measuring Body Composition, J. Brozek and A. Henschel (Eds.). Washington, DC: National Academy of Sciences, 1961, pp. 223–244.
19. Tothill, P., M. A. Laskey, C. I. Orphanidou, and M. van Wijk. Anomalies in dual energy X-ray absorptiometry measurements of total-body bone mineral during weight changes using Lunar, Hologic, and Norland instruments. Br. J. Radiol. 72: 661–669, 1999.
20. Van Loan, M. D., H. L. Johnson, and T. F. Barbieri. Effect of weight loss on bone mineral content and bone mineral density in obese women. Am. J. Clin. Nutr. 67: 734–738, 1998.
21. Wagner, D., and V. Heyward. Techniques of body composition assessment: a review of laboratory and field methods. Res. Q. Exerc. Sport. 70: 135–149, 1999.
22. Wagner, D. R., V. H. Heyward, and A. L. Gibson. Validation of air displacement plethysmography for assessing body composition. Med. Sci. Sports Exerc. 32: 1339–1344, 2000.


©2002The American College of Sports Medicine