Journal Logo

BASIC SCIENCES: Original Investigations

Effects of clothing on accuracy and reliability of air displacement plethysmography


Author Information
Medicine and Science in Sports and Exercise: February 2002 - Volume 34 - Issue 2 - p 282-285
  • Free


The BOD POD body composition system (Life Measurement Instrument, Concord, CA) uses air displacement plethysmography (ADP) for the estimation of percent body fat (%fat). This method offers advantages over hydrostatic weighing (HW) such that it requires less technical expertise to administer, is quick to perform, and may be accommodating to a wider range of individuals. Some research provides evidence supporting the accuracy and reliability of estimated %fat for ADP compared with HW while wearing the recommended minimal clothing (5,6,15), whereas others (1,2,7) have discovered ADP overestimates body density and thus underestimates %fat. The principles of physics that underlie the design of ADP make it highly sensitive during body volume measurements. It has been suggested that the attire worn during ADP will effect its accuracy and may effect reliability of %fat estimations due to the distinction between gas under isothermal or adiabatic conditions (3). Under isothermal conditions, the volume of compressed air is decreased in proportion to the rise in pressure (Boyle’s law). On the other hand, temperature is not constant as air volume changes under adiabatic conditions (Poisson’s law), making it more difficult to compress. In other words, materials such as cloth will be more easily compressed and show a negative volume when measured by ADP because of isothermal conditions within the material (3).

ADP measures the volume of a body within its chamber. Instruments in the rear (reference) chamber of the plethysmograph create small volume changes by moving a diaphragm dividing both the front (test) and rear chambers, which are equal but opposite in sign. This will create reciprocating pressure changes in both chambers. Providing the formula VT/VR = PR/PT, where T = test chamber and R = reference chamber, it can be observed that PR/PT becomes a linear function of VT due to VR being fixed. Therefore, the volume of a body within the test chamber (VT) is determined by the linear relationship of the pressure ratio between the two compartments (PR/PT) (3).

Because the isothermal nature of clothing makes it more compressible, it will cause a greater pressure change to occur in the front chamber of the BOD POD during body volume measurements (3). The effect is a smaller pressure ratio and a decreased body volume measurement for a given individual. Placing a smaller volume in the density formula will produce a greater body density and create a reduced %fat estimation for a given individual. It is for this reason that it is recommended to wear minimal clothing during ADP measurement.

Since the development of the BOD POD its use has become increasingly popular in the university setting and with professional athletic teams. There may be individuals in the clinical setting, however, who may be uncomfortable wearing a swimsuit during testing because of a negative body image. It may therefore become necessary to allow a subject to wear clothing other than the recommended attire. On the other hand, because ADP is highly sensitive to clothing, it was theorized that even a swimsuit could effect measures of body volume and that a nude condition might provide more accurate data. Although it has been suggested that clothing will effect %fat by ADP (3), no research has quantified the effects of different clothing on the accuracy or reliability of %fat measurements using ADP. Therefore, this investigation sought to 1) determine the impact of wearing clothing (a hospital gown) and a nude condition on the accuracy of %fat estimations by ADP and 2) to assess whether effects on the reliability of %fat are reflected by these different conditions compared with the recommended swimsuit condition and the gold standard, hydrostatic weighing (HW).



Fifteen adult volunteers (10 female, five male) participated in the study. Mean (± SD) age was 25.4 ± 9.2 yr, mean body weight was 67.8 ± 12.2 kg, mean height was 168.9 ± 9.8 cm, and mean residual lung volume (RV) was 1.46 ± 0.037 L. The range of %fat by HW was 17.7–33.9% for female subjects and 10.6–33.9% for male subjects. Experimental procedures were approved by the Medical Center Institutional Review Board and were in compliance with the American College of Sports Medicine policies for use of human subjects. All subjects signed a written informed consent before participation in the study. Subjects were asked to avoid any major changes in their diet and/or exercise regimens while participating in this study; however, no formal logs were maintained. All subjects stated no changes had occurred for the duration of the investigation.

Study design.

Subjects reported to the laboratory on three consecutive days, followed by a fourth session scheduled approximately 3 wk (range 19–24 d) from the third testing day. The fourth session was scheduled 3 wk later in an effort to determine the ability of ADP to accurately and reliably measure any changes in %fat that might occur over time and to compare the three conditions. Due to the requirement of dry conditions (6), it was necessary to perform testing with ADP before HW for the four testing days. Each testing session was comprised of estimating %fat using ADP wearing a swimsuit (ADPSS), wearing a hospital gown (ADPHG), and in the nude (ADPN), followed by measurement via HW. Simply for comfort of the subject, the three ADP conditions were performed in the order listed above, therefore allowing the individual to become accustomed to wearing progressively less clothing. All measurements were performed under consistent testing conditions, including time of day and temperature of the ADP laboratory. Subjects were tested by same gender technicians for all ADP measurements.


Procedures for ADP are explained elsewhere (3,6). Briefly, while wearing the attire for the appropriate trial and a swim cap, each subject was weighed on a calibrated digital scale to the nearest 20 g. Next, the subject was seated within the ADP chamber for two measures of body volume. If these two body volume values agreed within 150 mL, the data was accepted, averaged, and used for calculation. If these measurements were not within 150 mL, then a third body volume was measured. If two of the three measures did not agree within 150 mL, the system was recalibrated and the test was repeated (3).

Next, the measurement of thoracic gas volume occurred. Dempster and Atkins (3) provide a detailed explanation of this procedure. For this investigation, measurement of thoracic gas volume was performed only once, occurring on the first ADP measurement on the first testing day. Data from our laboratory revealed that measures of thoracic lung volume obtained up to 3 months apart were highly reliable (intraclass reliability r = 0.96). From the data collected for weight, body volume, and thoracic lung volume, computer software determined body density and then estimated %fat by using the Siri equation (12). Because the system was recalibrated between all trials for all subjects, the chance of an order effect between conditions is possible but unlikely.

Hydrostatic weighing.

An indoor pool was used to obtain underwater weight. A scale suspended over the water was attached to a chair submersed underwater. Subjects entered the pool and removed air bubbles from their swimsuit and body hair. All participants were instructed to maximally exhale while submersing completely underwater. Each trial was terminated when it was evident no more air was being expired and the two investigators present agreed upon a weight within 20 g. All subjects performed a minimum of six trials and until three measurements agreed within 20 g. The three heaviest weights were then averaged to calculate body density from the following formula: body density = Ma/[Ma − Mw/Dw] − Vr, where Ma is mass in air, Mw is mass underwater, Dw is the temperature-corrected water density, and Vr is residual volume. Percent fat was estimated using the Siri (12) formula.

Residual lung volume.

A SensorMedics metabolic cart (SensorMedics, Yorba Linda, CA) was used to determine residual volume through the nitrogen dilution method (14). Day-to-day reliability in our laboratory is high (intraclass reliability r = 0.99), with a CV of 6.3%. This resulted in a mean day-to-day difference of 0.052 mL, which produces only a 0.35% difference in body fat. Residual lung volumes tend to be slightly less when measured under water compared with on land, but the effect on computed body fat is small to insignificant (9,10). Therefore, residual lung volume can be measured in air before underwater weighing without a loss of accuracy in body composition compared with simultaneous measurement of residual volume with underwater weighing (4).

Statistical analysis.

Statistics were determined using SPSS version 8.0. Reliability for each condition (i.e., ADPN trial 1 vs trial 2 vs trial 3 vs trial 4) was determined using a single-factor ANOVA with repeated measures. When significant differences were observed, a Tukey’s post hoc analysis was performed. Reliability of measurements was analyzed further by determining the intraclass correlation coefficient for each condition for the four testing sessions. Because no significant differences were found for any of the conditions over the four days, the data for all four days were averaged for each subject to examine accuracy. The accuracy of ADP estimates of %fat were compared with HW also using an ANOVA with repeated measures and Tukey’s post hoc analysis. The alpha level was set at P ≤ 0.05. Results are expressed as mean ± SD.



Data for each condition over the four trials are presented in Table 1. The ANOVA indicated no differences in %fat measurements for any of the ADP conditions or HW across the four trials. Evidence of a strong relationship for body density and %fat was also revealed by the correlation coefficients, which are presented in Table 2. The technical error, as determined by Mueller and Martorell (8), was less than 1.0%fat for the swimsuit and nude conditions; however, the technical error while wearing the hospital gown was found to be nearly 3 times greater (Table 2).

Table 1
Table 1:
Percent body fat estimates.
Table 2
Table 2:
Reliability of %fat and density estimations for ADP attire conditions.


Because no differences were observed within any of the testing conditions, the mean value for the four trials was used to examine the accuracy of %fat estimations between methods. An ANOVA revealed a significant main effect between conditions (P < 0.007). Post hoc analysis determined %fat ADPHG was significantly lower than ADPSS, ADPN, and HW (P < 0.05), whereas no differences were found among the other three conditions (Fig. 1).

Percent body fat and body density measures are reported as means ± SD for ADP while wearing a hospital gown (ADPHG), wearing the recommended swimsuit (ADPSS), and nude condition (ADPN), and assessed by hydrostatic weighing (HW). * ADPHG significantly different from other three measures (P < 0.05).


Minimal clothing is suggested for measurements using ADP because of the different responses of gas under adiabatic and isothermal conditions. Clothing, due to its isothermal nature, will be more compressible during the volume perturbations, thus causing a decrease in the pressure ratio (VT/VR = PR/PT). A decrease in the pressure ratio will cause a subsequent decrease in the body volume measurement due to the linear relationship between the two. Substituting a smaller body volume (density = mass/volume) for a given subject will have the effect of creating a larger body density and smaller %fat estimation. Although Dempster and Atkins (3) reported that clothing will have an effect, to our knowledge, this is the first investigation to quantify the effects of clothing type and a nude condition on %fat estimates using ADP.

The primary, and expected, finding was the significant underestimation of %fat (approximately 8–9%fat) while wearing a hospital gown compared with the recommended swimsuit or nude conditions. The significance of these findings can be observed in a practical application of the data. An individual weighing 91 kg wearing the attire recommended by the manufacturer is measured at 32% fat, indicating he/she is carrying 29 kg of fat mass and 62 kg of lean mass. If the same individual were to wear a hospital gown, the %fat may now be estimated at 23–24%, resulting in approximately 22 kg and 68 kg of fat and lean mass, respectively. This creates the illusion of a more healthy body composition, when in reality this individual would be considered at a higher risk for the development of cardiovascular disease. Conversely, a lean individual that may only be 10%fat measured in a swimsuit, would be estimated to have 1–2% fat if allowed to wear a hospital gown. Therefore, research will have to determine whether other forms of clothing in between a hospital gown and swimsuit (i.e., gym shorts) could provide accurate and reliable measures.

A secondary, yet important, finding was that a nude condition did not provide more accurate or reliable results compared with the swimsuit condition. Because ADP is highly sensitive to clothing, it was hypothesized that even a small swimsuit may alter %fat estimations. This was not the case, however, because our data revealed only a 0.0036 ± 0.0015 g·mL−1 mean difference (calculated from Fig. 1) between the swimsuit and nude conditions for body density, which correlates to a nonsignificant (P = 0.94) difference of 1.5%fat.

Although not a focus of this investigation, the comparison of %fat ADPSS with HW lend some support to the growing body of literature on ADP. McCrory et al. (6) and Vescovi et al. (13) both reported no differences in %fat for samples varying from lean to overweight, with mean values of 25.4% and 22.0% fat, respectively. This agreement between studies seems to indicate ADP can strongly assess body composition when an individual or mean of a group falls within an average range of %fat. It is not surprising that the current data for ADPSS and HW, with mean values of 22.0% and 22.6%, respectively, are not significantly different. Collins et al. (1) and Vescovi et al. (13), however, determined the accuracy of measurements to be less precise when using groups of lean individuals. The range of %fat in this investigation was large (17.7–33.9% for female subjects and 10.6–33.9% for male subjects), but due to a small sample size, it was not possible in this investigation to assess subsets (e.g., lean and overweight).

This investigation found no effect within any attire condition on the reliability of %fat estimations by ADP. Our data demonstrated that there was no difference for any of the conditions over the four testing trials (Table 1). Others have reported strong test-retest data by using the recommended swimsuit attire (1,6,13); however, no other literature describes the reliability under different clothed conditions. The accuracy of ADP within a certain condition was examined further by the determination of technical error. In their investigations, Sardinha et al. (11) and Collins et al. (1) reported a technical error of measurement for %fat of 0.966% and 0.448%, respectively. The current study calculated two values for technical error. One using the trials performed on days 1 and 2, and the other from trials on days 1 and 4. Our results from days 1 and 2 for ADPSS (0.733%fat) and ADPN (0.815%fat) conditions fall between the range just mentioned; however, the technical error for ADPHG is nearly 3 times greater than the other two ADP conditions (Table 2). When day 4 was substituted for day 2 in the calculation of technical error, all the values increased (Table 2). This is due primarily to the lower, nonsignificant, estimation of %fat 3 wk after the third assessment. Because there appears to be similar reductions by all conditions, it is suggested to rely more heavily on the technical error determined from trials 1 and 2. It could be concluded from our results that repeated measures within a particular clothing condition of ADP, although not significantly different, may be adversely effected by the amount of clothing worn as shown by the large technical error of ADPHG.

The purpose of this study was to determine whether the reliability and/or accuracy of ADP estimations of %fat were effected by clothing types. Although mean estimations of %fat are not significantly altered with repeated measures, the large technical error of the hospital gown condition suggests reliability might be effected by excess clothing. Furthermore, the accuracy of %fat was significantly lowered (nearly 9%fat) by wearing a hospital gown, related to the isothermal nature of clothing. Although the measurement of body volume during ADP is highly sensitive to clothing, it does not appear that a nude condition provides more accurate or reliable estimation of %fat. Therefore, this investigation supports the wearing of minimal clothing (i.e., swimsuit) for %fat estimations.

We thank Heather A. Young for providing valuable statistical advice for this article.

Address for correspondence: Jason D. Vescovi, M.S., Exercise Science Programs, The George Washington University, 817 23rd Street, N.W., Washington, D.C. 20052; E-mail: [email protected]


1. Collins, M. A., M. L. Millard-Stafford, P. B. Sparling, et al. Evaluation of the BOD POD for assessing body fat in collegiate football players. Med. Sci. Sports Exerc. 31: 1350–1356, 1999.
2. Collins, M. A., M. L. Millard-Stafford, E. M. Evans, T. K. Snow, L. B. Rosskopf, and K. J. Cureton. Validation of air displacement plethysmography for estimating body fat in young adults. Med. Sci. Sports Exerc. 30: s146, 1998.
3. Dempster, P., and S. Atkins. A new air displacement method for the determination of human body composition. Med. Sci. Sports Exerc. 27: 1692–1697, 1995.
4. Hsieh, S., G. Kline, J. Porcari, and F. I. Katch. Measurement of residual volume sitting and lying in air and water (and during underwater weighing) and its effects on computed body density. Med. Sci. Sports Exerc. 17: 204, 1985.
5. Levenhagen, D. K., M. J. Borel, D. C. Welch, et al. A comparison of air displacement plethysmography with three other techniques to determine body fat in health adults. J. Parenter. Enteral Nutr. 23: 293–299, 1999.
6. McCrory, M. A., T. D. Gomez, E. M. Bernauer, and P. A. Mole. Evaluation of a new air displacement plethysmograph for measuring human body composition. Med. Sci. Sports Exerc. 27: 1686–1691, 1995.
7. Millard-Stafford, M. L., M. A. Collins, L. B. Rosskopf, T. K. Snow, and S. A. Webb. Air displacement plethysmography compared to multi-component models for estimating body composition in African American men. Med. Sci. Sports Exerc. 30: s145, 1998.
8. Mueller, W. H., and R. Martorell. Reliability and accuracy of measurement. In: Anthropometirc Standardization Reference Manual, T. G. Lohman, A. F. Roche, and R. Martorell (Eds.). Champaign, IL: Human Kinetics, 1988, pp. 83–86.
9. Ostrove, S. M., and P. Vaccaro. Effect of immersion on RV in young women: implications for measurement of body density. Int. J. Sports Med. 3: 220–223, 1982.
10. Robertson, C. H. Jr., C. M. Engle, and M. E. Bradley. Lung volumes in men immersed to the neck: dilution and plethysmographic techniques. J. Appl. Physiol. 44: 679–682, 1978.
11. Sardinha, L. B., T. G. Lohman, P. J. Teixeira, D. P. Guedes, and S. B. Going. 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: 786–793, 1998.
12. 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: NAS/NRC, 1961, pp. 223–224.
13. Vescovi, J. D., S. L. Zimmerman, W. C. Miller, L. Hildebrandt, R. L. Hammer, and B. Fernahll. Comparison of the BOD POD for estimating percent body fat in a heterogeneous group of adults. Eur. J. Appl. Physiol. 85: 326–332, 2001.
14. Wilmore, J. H. A simplified method for the determination of residual lung volumes. J. Appl. Physiol. 27: 96–100, 1969.
15. Yee, A., and M. Kern. Validation of the BOD POD: method for estimating percent body fat in an elderly population. Med. Sci. Sports Exerc. 30: s146, 1998.


© 2002 Lippincott Williams & Wilkins, Inc.