Introduction

For the past 30 y the most widely used method to evaluate body composition has been hydrodensitom-etry (UWW). Although UWW is considered the ‘gold standard’, it is time consuming and in certain populations (elderly, children, obese, disabled) impractical. Thus, an alternative method that addresses these limitations is needed.

Recently, a whole body air-displacement plethysmograph (BOD POD) has been validated against UWW as a measure of whole body density with differences between techniques ranging from 0.3% to 1.1% fat.1,2,3 Plethysmography determines body volume based upon the pressure/volume relationship. Boyle's law explains this relationship at isothermic conditions. PV=k where k is the proportionality constant.4 Poisson's law explains the pressure/volume relationship if temperature is not constant (adiabatic): PVY=k where Y is the ratio of the specific heat of the gas at constant pressure to that at constant volume.5,6

The distinction between these two air conditions (isothermal and adiabatic) is of great importance to the measurement of body volume in a plethysmograph because isothermic air is 40% more easily compressed than adiabatic air, consequently leading to an over prediction in body density (Db)7. The design of the BOD POD has attempted to minimize this problem by using a circulating air system, sinusoidal perturbation, Fourier coefficients, and harmonic analysis.7

Two contributors to isothermal conditions within the testing chamber are the subject's body surface area and air in the lung.6,7 Surface Area Artifact (SAA) is calculated automatically by the BOD POD software utilizing the Dubois formula and corrects for changes in chamber temperature and gas composition that is attributable to the subject's body surface area.7,8 The isothermal air in the lung is normalized to adiabatic conditions by dividing the measured thoracic gas volume by 1.4.7 However, a potential source of isothermal air that is not accounted for in the program software is isothermic air trapped within the fiber of the subjects clothing.

Although McCrory et al 3 validated the BOD POD in females, the subjects were relatively young with truncated %fats. Biaggi et al 1 and Nunez et al 2 validated the BOD POD in females with UWW, however the former study had a small sample size (n=24) and the latter study did not control for the type of clothing worn while in the BOD POD. To our knowledge no investigators have studied the potential confounding effect of trapped isothermic air in clothing in different types of swimsuits (one piece and two piece) and clothing schemes (hospital gown) and its effect on body composition estimates made by the BOD POD. This is problematic since standardized clothing is not widely used. Some studies allowed subjects to wear a hospital gown or shorts, while other studies allowed subjects to wear a one piece or two piece swimsuit. Thus, it is important to determine the effect that different types of clothing and different types of swimsuits have on body composition. Therefore, the purpose of this study was three fold: first, to validate the BOD POD against UWW in a group of women varying widely in age and Db; second, to explore the effect of isothermic air trapped in cloth-ing by comparing different swimsuits and clothing schemes; third, to determine whether calibrating the BOD POD with a hospital gown during the two point calibration could improve the Db estimate when wearing a hospital gown.

Methods

Subjects

Sixty-seven female subjects were selected from ongoing studies in the Nutrition Sciences Department at the University of Alabama at Birmingham. The subjects represented a wide range of densities (1.081–0.986 g/cm3) which corresponds to a body fatness of 6.5%–51.1% and ages 18–55 y. As described in detail in a subsequent section, the pro-cedures involved each subject having Db analyzed using UWW and BOD POD while wearing a one-piece swimsuit (OP). Additionally, in order to determine the effects of clothing on BOD POD estimates of Db, a subset of twenty-five subjects had their Db determined in the BOD POD while wearing a two-piece swimsuit (TP), a hospital gown (HG), and a hospital gown during calibration (GC). The descriptive data for both groups are contained in Tables 1 and 2. The study was approved by the University of Alabama at Birmingham (UAB) Institutional Review Board for human use, and written informed consent was obtained from each volunteer prior to testing.

Table 1 Descriptive data for all subjects (n=67)
Full size table
Table 2 Descriptive data for subset (n=25)
Full size table

Experimental design

Subjects were asked to refrain from eating 4 h prior to testing. Upon arrival to the body composition lab, subjects were asked to void their bladder. Height was measured to the nearest centimeter and weight to the nearest 0.1 kg. All BOD POD measurements were performed prior to UWW in order to avoid the possible influence of elevated body core temperature caused by UWW.

Instrumentation

BOD POD.

Whole body air-displacement was evaluated with the BOD POD version 1.69 (Body Composition System; Life Measurement Instruments, Concord, CA). The BOD POD is a single ‘egg’ shaped unit consisting of two chambers; a testing chamber where the subject sits and a reference chamber where the breathing circuit, pressure transducers, and electronics are stored.7 The testing procedure involves several steps. First, calibration was conducted prior to subject entry into the BOD POD. Calibration involves the computation of the ratio of the pressure amplitudes (reference chamber and testing chamber) for an empty chamber and a known volume (≈50-L). The BOD POD software calculates a regression equation between the testing chamber volume and the ratio of the pressure amplitudes.7 Essentially, the relationship is linear for any testing chamber volume and the ratio of the pressure amplitudes.6,7 After the calibration was completed and the procedures fully explained to the subject the relevant clothing scheme, noseclip and swimcap (worn to minimize isothermal air trapped within the hair) were donned. The subject entered the BOD POD for two trials of approximately 45 s each. During this stage, the subject's raw body volume (Vbraw) was determined with the testing chamber door being opened between trials. If both volumes were within 150 ml then the two trials were averaged. However, if the trials were not within 150 ml a third trial was performed and the two trials that were the closest were averaged. The last step involved the measurement of the thoracic gas volume (VTG). This stage involves the subject to sit quietly in the BOD POD and breathe through a disposable tube and filter that was connected to the reference chamber in the rear of the BOD POD. After four or five normal breaths the airway was occluded during mid-exhalation and the subject was instructed to make two quick light pants. For VTG to be considered successful, four criteria had to be met. First, merit had to be <1. The merit is a theoretical value that demonstrates subject compliance during the measurement of the VTG. A merit of less than one indicates perfect agreement between chamber pressure and the airway pressure in the tube, with a merit >1 demonstrating poor compliance (usually due to leaking of air around the mouth).7 Second, air way pressure had to be below 35 cm H2O. Third, tidal volume had to be between 0.40–0.70 liters. Finally, the measured VTG had to be within 0.70 liters of the predicted VTG.

The Db from the BOD POD was calculated as follows: Db=M/(Vbraw+0.40VTG−SAA) where SAA and 0.40VTG are used to correct for the isothermic conditions within the chamber and M is the mass of the subject.

The repeat measures between consecutive days for Db derived from the BOD POD in eight healthy females (all wearing an OP) has an intraclass correlation of r=0.98 and a SEE of 0.004 g/cm3 in our laboratory.

Hydrostatic weighing.

Total body density was measured by UWW, with simultaneous measurement of residual lung volume by using the closed-circuit oxygen dilution technique.9 Underwater weight was measured to the nearest 50 g in a stainless steel tank in which the subject, while wearing a one-piece swimsuit was suspended from a LCL 20 Shear Beam Load Cell calibrated from 0 to 10,000 g (Omega, Stanford, CT). After one practice trial, underwater weight and residual lung volume were measured simultaneously 5 times. The average of multiple trial densities within 0.001 g/cm3 were used from the underwater weight. Body fat was calculated from whole body density (g/cm3) with the Lohman equation.10

The repeat measures between days for Db derived from UWW in eight healthy females (all wearing an OP) has an intraclass correlation of r=0.99 and a SEE of 0.002 g/cm3 in our laboratory, thus demonstrating excellent agreement between consecutive days.

Statistics

To determine the accuracy of Db estimates by the BOD POD linear regression analysis was used with Db measured by the BOD POD as the independent variable and Db measured by the criterion measure (UWW) being the dependent variable. Analyses were conducted for each clothing scheme. The accuracy of the BOD POD in determining Db was assessed by comparing the regression slope and intercept. If the regression slope was not significantly different from one and the intercept not significantly different from zero, the estimates of Db derived from the BOD POD were considered acceptable. Paired t-tests were used to compare group means of the BOD POD with UWW. To account for multiple t-tests and an increase in type I error, the Bonferroni statistical procedure was employed. The precision of the BOD POD was determined by the r2 and the SEE. Lastly, potential bias in Db estimates by the BOD POD was examined using residual plots. This analysis examines the discrepancy between techniques across the range of densities. Statistical significance was set at P<0.05.

Results

The results of the first regression equation which tests the validity of the BOD POD are shown in Figure 1, upper panel. As shown in Figure 1, upper panel the regression between Db by UWW and BOD POD did not significantly deviate from the line of identity. Paired t-test results revealed UWW Db (1.030 g/cm3) was significantly higher (P<0.01) than BOD POD Db (1.028 g/cm3) however, the difference was small and well within the error of measurement. This represents a 1% fat difference between BOD POD and UWW. The BOD POD estimation of Db explained 94% of the variance by UWW and the estimate had a SEE of 0.005 g/cm3. The residual plot presented in Figure 1, lower panel revealed no bias in the BOD POD estimate of Db across the wide spectrum of densities.

Figure 1
figure 1

Upper panel is the regression of Db by UWW against the Db by BOD POD in the total sample of 67 subjects. The dotted line is the line of identity (regression slope=1 and regression intercept=0). The slope and intercept were not significantly different from one and zero respectively. The lower panel is the residual plot. The dashed line represents the mean difference between Db BOD POD−Db UWW and the solid line is the regression line. No bias between the techniques was observed, as indicated by a non-significant P-value.

Full size image

The next series of regression equations explore the effect of different swimsuits on the assessment of Db utilizing the BOD POD. The regression for Db by UWW vs Db by BOD POD in the sub-set group, while subjects wore a OP is shown in Figure 2, upper panel. The regression slope and intercept were not significantly different from one and zero respectively. Paired t-test results revealed that the OP Db was significantly lower (P<0.01) than the UWW Db (0.004 g/cm3). This is a difference of 1.9% fat between methods. The OP estimation of Db explained 86% of the variance of Db measured by UWW, while the estimate had a SEE of 0.005 g/cm3. The residual plot indicated no systematic differences across the range of different densities (Figure 2, lower panel).

Figure 2
figure 2

Upper panel is the regression of Db by UWW against the Db by BOD POD in the sub-set group of subjects wearing a OP. The dotted line is the line of identity (regression slope=1 and regression intercept=0). The slope and intercept were not significantly different from one and zero respectively. The lower panel is the residual plot. The dashed line represents the mean difference between Db TP−Db UWW and the solid line is the regression line. No bias between the techniques was observed, as indicated by a non-significant P-value.

Full size image

The regression for Db by UWW vs Db by BOD POD in the sub-set group while subjects wore a TP is shown in Figure 3, upper panel. The regression slope and intercept was not significantly different from one and zero respectively. Paired t-test results revealed that the TP Db was significantly lower (P<0.01) than the UWW Db (0.004 g/cm3). This is a difference of 1.9% fat between methods. The TP estimation of Db explained 86% of the variance of Db measured by UWW. The estimate had a SEE of 0.005 g/cm3. The residual plot indicated no systematic differences across the range of different densities (Figure 3, lower panel).

Figure 3
figure 3

Upper panel is the regression of Db by UWW against the Db by BOD POD in the sub-set group of subjects wearing a TP. The dotted line is the line of identity (regression slope=1 and regression intercept=0). The slope and intercept were not significantly different from one and zero respectively. The lower panel is the residual plot. The dashed line represents the mean difference between Db TP−Db UWW and the solid line is the regression line. No bias between the techniques was observed, as indicated by a non-significant P-value.

Full size image

Additionally, the OP Db (1.0405 g/cm3) and TP Db (1.0404 g/cm3) in the sub-set group was analyzed. The mean difference in densities was 0.0001 g/cm3 which translates to a difference of 0.05% fat. Paired t-test results revealed that OP Db and TP Db were not statistically different. The amount of shared variance between the OP and TP was 88%.

The next analysis was conducted to determine the impact of isothermic air on the estimate of Db measured by the BOD POD. This was done with the subject wearing a hospital gown while being measured in the BOD POD. The regression for Db by UWW vs Db by HG is shown in Figure 4, upper panel. The regression slope and intercept significantly deviated from the line of identity, indicating poor agreement between the two methods. In addition, paired t-test results revealed that the HG Db (1.056 g/cm3) was significantly higher (P<0.01) than UWW Db (1.044 g/cm3). The discrepancy in densities represents a difference of 5.5% fat. The HG estimation of Db explained 84% of the variance by UWW and the SEE was 0.006 g/cm3. The residual plot indicated that the HG systematically under predicted Db at the higher densities and over predicted Db at the lower densities (Figure 4, lower panel).

Figure 4
figure 4

Upper panel is the regression of Db by UWW against the Db by BOD POD in the sub-set group of subjects wearing a HG. The dotted line is the line of identity (regression slope=1 and regression intercept=0). The slope and intercept were significantly different from one and zero respectively. The lower panel is the residual plot. The dashed line represents the mean difference between Db HG−Db UWW and the solid line is the regression line. Bias between the techniques was observed, as indicated by a significant P-value.

Full size image

The final set of analysis explore the role of calibration and the possibility of correcting for the isothermic effect of trapped air in the clothing fibers on Db estimates. Theoretically, by putting the gown in the chamber during the two point calibration, the regression slope developed from the computer software would be changed, thus potentially improving the Db estimate. The regression of Db by UWW and GC significantly deviated from the line of identity (Figure 5, upper panel). Paired t-test results revealed the discrepancy in Db by the two techniques to be significant at the P<0.01 level. The GC under predicted Db by 0.007 g/cm3. This would correspond to a difference of 3.2% fat. The GC estimation of Db explained 84% of the variance by UWW and the estimate had a SEE of 0.006 g/cm3. The residual plot demonstrates that the GC overestimated Db at the higher percent fat levels and underestimated Db at the lower percent fat levels (Figure 5, lower panel).

Figure 5
figure 5

Upper panel is the regression of Db by UWW against the Db by BOD POD in the sub-set group of subjects wearing a hospital gown that was calibrated prior to subject entry (GC). The dotted line is the line of identity (regression slope=1 and regression intercept=0). The slope and intercept were significantly different from one and zero respectively. The lower panel is the residual plot. The dashed line represents the mean difference between Db GC−Db UWW and the solid line is the regression line. Bias between the techniques was observed, as indicated by a significant P-value.

Full size image

Discussion

The results from this study demonstrate a high level of agreement between Db estimates of the BOD POD and UWW in either an OP or TP swimsuit. In addition, no systematic variation in estimates of Db were observed in subjects varying widely in %fat estimates (6.5%–51.1%) and in age (18–55 y), suggesting that the accuracy of the BOD POD extends to a diverse range of body fatness and ages. These results as well as the results of Nunez et al,1 Biaggi et al,2 and McCrory et al,3 support the use of the BOD POD estimates of density as a substitute for UWW in adults.

From a practical standpoint these findings are significant because the BOD POD is quicker and less laborious than UWW. In addition, the BOD POD may be more desirable from the subject's point of view because it does not require submersion of the head in water. From a research perspective the BOD POD can be used in the place of UWW where subject compliance is low (children, elderly, disabled, obese) and time constraints prohibitive.

Since our laboratory has not tried to measure individuals weighing more than 145 kg we cannot speculate on the upper weight limit. However, the manufacturers claim the BOD POD can accommodate individuals weighing up to 165 kg.

Both the OP and TP swimsuits used in BOD POD assessments validated well with UWW suggesting that an OP or TP can be worn interchangeably. The swimsuits worn by the subjects were all tight fitting with little material. These data suggest that either an OP or TP swimsuit can be used during the BOD POD analysis as long as they are tight fitting.

This data also demonstrates the large impact isothermal air has on Db values in the BOD POD. When subjects entered the BOD POD in the HG, Db should have decreased because the volume of the subject was being artificially increased, but in actuality it increased. This would indicate that the trapped isothermic air in the clothing had a large impact on chamber air behavior relative to its small volume. This could have implications when measuring older and more overweight populations since such persons are more likely to wear ‘bulky or skirted’ swimsuits, thus introducing a significant amount of error in the BOD POD Db. It would be advised that all subjects wear only tight fitting swimsuits.

It was hypothesized that the GC would correct for the effect that isothermic air has on Db. Calibrating the BOD POD with the hospital gown prior to analysis prevented the marked overestimation in Db (0.012 g/cm3) or an underestimation in percent fat of 5.3% as seen in the HG, where GC was associated with a modest underestimation in density (0.007 g/cm3) or an overestimation of 3.2% fat. Unfortunately, GC demonstrated a regression slope significantly different from one (P<0.01), causing a large underestimation in percent fat for women greater than 25% fat.

In conclusion, this study supports the use of the BOD POD as a substitute for UWW for the determination of Db and the estimation of percent fat in women wearing tight fitting swimsuits. The high R2, low SEE, zero intercept, and a regression slope of one all indicate strong agreement between the two techniques. It also appears that the type of swimsuit worn (OP or TP) does not affect the measurement of Db. However, caution should be made in using the BOD POD for evaluating Db while subjects are clothed in anything other than a tight fitting swimsuit