72

Comparison of Air-Displacement Plethysmography, Hydrodensitometry, and Dual X-ray Absorptiometry for Assessing Body Composition of Children 10 to 18 Years of Age

D. W. LOCKNER,a V. H. HEYWARD,a R. N. BAUMGARTNER,c AND K. A. JENKINSd aCenter for Exercise and Applied Human Physiology and cClinical Nutrition Program, University of New Mexico, Albuquerque, New Mexico 87131, USAdDepartment of Human Performance, Leisure, and Sport, New Mexico Highlands University, Las Vegas, New Mexico

ABSTRACT: Body density (Db) of 54 boys and girls 10–18 years of age (13.9 ± 2.4years) was measured in an air-displacement plethysmograph, the BOD POD,and compared to Db determined by hydrodensitometry (HW). Both Db valueswere converted to percent body fat (%BF) using a two-component model con-version formula and compared to %BF determined by dual energy X-rayabsorptiometry (DXA). Body density estimated from the BOD POD (1.04657 ±0.01825 g/cc) was significantly higher than that estimated from HW (1.04032 ±0.01872 g/cc). The relative body fat calculated from the BOD POD (23.12 ± 8.39%BF) was highly correlated but, on average, 2.9% BF lower than %BF DXA.Average %BF estimates from HW and DXA were not significantly different. De-spite consistently underestimating the %BF of children, the strong relationshipbetween DXA and the BOD POD suggests that further investigation may im-prove the accuracy of the BOD POD for assessing body composition in children.

INTRODUCTION

Interest in assessing body composition in children is rising, due in part to the in-creasing incidence of childhood obesity. Because methods that are appropriate foradults may not be valid for children, we must carefully evaluate new techniques todetermine their accuracy and reliability for that population by comparing results toa reference method. For decades, hydrodensitometry (HW) has served as the refer-ence method to assess body density (Db) of adults and children. However, the errorassociated with HW may be large for children due to procedural difficulties in per-forming underwater weighing and assessing residual lung volume.1 Therefore, usingHW as a reference method for children may be inappropriate.

bAddress for correspondence: D. Lockner, Center for Exercise and Applied Human Physiol-ogy, Johnson Center, University of New Mexico, Albuquerque, NM 87131. Voice: 505-277-3160; fax: 505-277-4362.

dlockner@unm.edu

73LOCKNER et al.: BODY COMPOSITION OF CHILDREN

Dual energy X-ray absorptiometry (DXA) is widely used for body compositionassessment and is gaining support as a reference method.2 For children, DXA maybe even more appropriate than HW as a reference method because it requires littleactive participation by the child and makes no assumptions about the composition ofthe fat-free mass. Ellis3 suggests DXA as a criterion method for children when directchemical analysis is not possible.

Recent reports of the validity of air displacement plethysmography for determin-ing body composition in adults4,5 prompted this investigation of its usefulness forchildren. This device, the BOD POD, is readily accessible, rapid, has an easy test-ing protocol, and does not require submersion in water; thus, it is an attractive toolfor assessments of children. Lacking a singular reference standard, the purpose ofthis study was to compare the BOD POD to HW and DXA for assessing body com-position of children.

METHODS

Participants

Fifty-four healthy children (42 females, 12 males), aged 10–18 years, were re-cruited from personal contacts in a large metropolitan area and surrounding towns.The study was approved by the university’s Institutional Review Board, and writteninformed consent was given by participants and parents prior to data collection.

Hydrostatic Weighing

The children were weighed underwater at residual lung volume using a load cellsystem, which was calibrated before each test. As many trials were made as needed toobtain three underwater weights within ±50 grams. Residual lung volume (RV) wasmeasured while the children were seated, out of water, using the helium dilution tech-nique,6 again with as many trials as needed to obtain two RV values within 100 milli-liters. These data from the two trials were averaged and used to calculate body volume.

Air Displacement Plethysmography

Body volume was measured in the BOD POD (Life Measurement Instruments,Concord, CA), an air displacement plethysmograph previously described.7 TheBOD POD was calibrated before each test following the manufacturer’s recommen-dations. Participants wore a tight-fitting swimsuit or biking shorts, and a swim cap,and remained seated in the chamber during the measurement. The software-drivenautomated test included measurement of volume of the thoracic gas. For this part ofthe test, the children were prompted to puff gently against an occluded airway whilechanges in body volume were recorded. If a valid measure was not obtained afterthree trials, the volume of thoracic gas was predicted by the BOD POD software,which also calculated body density at the conclusion of each test.

Dual Energy X-ray Absorptiometry

The bone mineral content and percent body fat (%BF) of each person were mea-sured by dual energy X-ray absorptiometry (DXA) (Lunar DPX, Lunar Radiation

74 ANNALS NEW YORK ACADEMY OF SCIENCES

Corp., Madison, WI, Software Version 3.6Z). The densitometer was calibrated dailyaccording to the manufacturer’s recommendation, and all scans were performed bya licensed radiological technician. Subjects lay supine with arms and legs at theirsides during the 15- to 20-minute scan.

Converting Body Density Values to Percent Body Fat

For comparison to %BF obtained from DXA, %BF was calculated from the two-component Siri equation.8 All hydrostatic weighing, BOD POD, and DXA measure-ments were taken on the same day within three hours. The order of testing was ran-domized for each person.

Statistical Analysis

A paired t test was used to compare DbHW and DbBP. The validity coefficient (r),standard error of the estimate (SEE), and slope and intercept of the regression linewere calculated with linear regression. The average measured volume of thoracic gas(Vtg) was compared to the average predicted Vtg using a paired t test. Paired t testswere also used to compare %BF estimates from HW and BOD POD to %BF estimat-ed by DXA. The relationship between %BFDXA and %BF, determined by each of thetwo densitometric methods, was explored using linear regression.

An alpha value of .05 was set for testing significance. The Statistical Package forthe Social Sciences (SPSS, Version 7.5 for Windows) was used for all analyses.

RESULTS

The average age of the children was 13.9 ± 2.4 years (range = 10–18 years). Self-re-ported ethnicity was 59% Caucasian, 35% Hispanic, and 6% Native American. TABLE 1summarizes their other characteristics.

TABLE 1. Physical characteristics of the children (n = 54)

Variable Mean ± SD Range

Age (year) 13.93 ± 2.38 10–18

HT (cm) 159.41 ± 9.95 134.90–180.10

BM (kg) 52.50 ± 11.07 30.50–75.70

BSA (cm2) 15236.60 ± 1880.73 10843.39–18357.07

Measured VTG (L) (n = 37) 2.81 ± 0.78 1.47–4.85

Predicted VTG (L) 3.00 ± 0.38 1.77–3.95

DbHW (g/cc) 1.04032 ± 0.01872 0.99214–1.07369

DbBOD POD (g/cc) 1.04657 ± 0.01825 0.9953–1.0707

%BFDXA 25.23 ± 9.73 9.10–49.60

%BFHWa 25.97 ± 8.65 11.03–48.92

%BFBODPODa 23.12 ± 8.39 12.31–47.34

a %BF calculated using two-component model conversion formula (Siri, 1961).

75LOCKNER et al.: BODY COMPOSITION OF CHILDREN

There was a significant difference between average DbBOD POD (1.04657 ±0.01825 g/cc) and average DbHW (1.04032 ± 0.01872 g/cc), p <0.0005; the validitycoefficient between them was r = 0.85, p <0.0005, and the SEE was 0.0100 g/cc. Theslope (0.869) of the regression line relating DbBOD POD to DbHW did not differ from 1.0(p >0.05), and the intercept (0.131) was not significantly different from zero (p >0.05).FIGURE 1 shows the relationship between DbHW and DbBOD POD. The bias of Db esti-mated by the BOD POD compared to HW was related to height (r = 0.33), body mass(r = 0.40), and body surface area (r = 0.41), with the largest overprediction for thesmallest children.

Thirty-seven children (69%) achieved a valid measurement of thoracic gas vol-ume (Vtg). Their Db was calculated using predicted Vtg (3.00 ± 0.38 L), instead ofmeasured Vtg (2.81 ± 0.78 L) and resulted in a significant mean difference of0.0027 g/cc higher, p = 0.008.

When Db was converted to %BF for comparison to DXA values, %BFBOD POD (23.12± 8.39 %BF) was significantly lower than %BFDXA (25.23 ± 9.73 %BF), p <0.0005.There was no significant difference between %BHW (25.97 ± 8.65 %BF) and %BDXA.Further analysis of this relationship, however, showed that the correlation between%BFDXA and %BFBOD POD (r = 0.94) was greater than the correlation between%BFDXA and %BFHW (r = 0.89), whereas the SEE was smaller for BOD POD (SEE= 3.41 %BF) compared to HW (SEE = 4.55 %BF) (FIG. 2). Neither slopes of the linesrelating %BFBOD POD and %BFHW to %BFDXA differed from one, and the interceptsdid not differ from zero.

FIGURE 1. Relationship between DbBOD POD and DbHW. The dashed line indicatesline of identity.

76 ANNALS NEW YORK ACADEMY OF SCIENCES

DISCUSSION

With the interest in the body composition of children, the validation of a rapid,convenient method of assessment is needed. Although hydrodensitometry is oftenused as a reference method for assessing body composition in children, its inconve-nience and the difficulty of the protocol may lead to errors; DXA may be a more ap-propriate reference method due to its reduced error compared to HW. However, thecost of DXA and the exposure to radiation during each scan may be problematic forthe pediatric population. Air-displacement plethysmography has the potential toovercome these problems, but few validation studies with children have been con-ducted. Therefore, this study compared body composition results from the air-dis-placement plethysmograph, the BOD POD, to HW and DXA.

The testing protocol for HW, including measurement of residual lung volume, re-quired approximately 75 minutes for each child. Some children required additionalcoaching for underwater weighing, due to mild apprehension of submersion. By con-trast, air displacement plethysmography was easily accomplished. All children in thepresent study seemed comfortable with sitting in the BOD POD while breathing nor-mally, and no training was needed for a successful Db trial, which took approximate-ly 10 minutes. Although the measurement of thoracic gas volume in the BOD PODwas difficult for some children, this factor did not seem to significantly impact Dbmeasurement. Using predicted instead of measured volume of thoracic gas increasedthe mean Db by 0.002 g/cc, similar to the results of Collins et al.,9 who reported alower measured Vtg compared to predicted Vtg in their study of collegiate footballplayers. The use of measured Vtg in calculating Db slightly overpredicted it in thatstudy, as well as for the children in this study.

The similarity of mean %BF estimated from HW and DXA could be misinterpret-ed as close agreement between these two methods. However, the large prediction er-ror, (SEE >4.5 %BF) suggests that HW is not a good method for determining bodycomposition in children. Despite the significant mean difference in %BFBOD POD and%BFDXA (with %BFBOD POD being 3.8% BF lower, on average), the SEE was smaller(3.41% BF) than that for the relationship between %BFHW and % BFDXA (4.55%BF). Although %BFBOD POD was underpredicted for 89% of the sample, there wasoverall closer agreement between %BFBOD POD and %BFDXA than between %BFHWand %BFDXA (FIG. 2).

The overestimation of Db compared to HW for children in this study contrastswith previous studies using heterogeneous groups of adults4,5,10,11 where very closeagreement was reported. With more homogeneous subgroups, however, results havebeen mixed. As in the present study, Iwaoka et al.12 reported an overestimation ofDb compared to HW (with an underestimation of %BF determined by the BOD PODcompared to DXA) in Japanese men of short stature, as did Collins et al.9 for colle-giate football players. Nuñez et al.10 found no significant difference for average Dbestimates from HW and the BOD POD for children 8–19 years of age; however, asignificant bias was noted. Explanation for these inconsistent results is lacking; aswith any new technology, additional observations are needed.

The strong correlation between %BFBOD POD and %BFDXA suggests additionalresearch may identify techniques to improve the usefulness of the BOD POD forbody composition measurement of children. Our results support the results and con-clusion recently published by Nuñez et al.10 that Db and derived %BF assessed by

77LOCKNER et al.: BODY COMPOSITION OF CHILDREN

FIGURE 2A. Relationship between %BF DXA and %BF HW. The dashed line indi-cates line of identity.

FIGURE 2B. Relationship between %BF DXA and %BF BOD POD. The dashed lineindicates line of identity.

78 ANNALS NEW YORK ACADEMY OF SCIENCES

the BOD POD are highly correlated to values from HW and DXA. The bias in Dbmeasurement by the BOD POD related to body size indicates that a small correctionfactor may improve the accuracy of the BOD POD measurements of children com-pared to DXA measurement. In addition, with our poor results in assessing bodycomposition of children using HW, the convenience of the BOD POD encouragesfurther investigations to improve its accuracy for such assessments.

REFERENCES

1. WELLS. J.C., N.J. FULLER, O. DEWITT, M.S. FEWTRELL, M. ELIA & T.J. COLE. 1999.Four-compartment model of body composition in children: density and hydration offat-free mass and comparison with simpler models. Am. J. Clin. Nutr. 69: 904–912.

2. KOHRT, W.M. 1998. Preliminary evidence that DEXA provides an accurate assessmentof body composition. J. Appl. Physiol. 84: 372–377.

3. ELLIS, K.J. 1999. Measuring body fatness in children and young adults: comparison ofbioelectrical impedance analysis, total body electrical conductivity, and dual-energyX-ray absorptiometry. Int. J. Obesity 20: 866–873.

4. BIAGGI, R.R., M.W. VOLLMAN, M.A. NIES, C.E. BRENER, P.J. FLAKOLL, D.K. LEVEN-HAGEN, M. SUN, Z. KARABULUT & K.Y. CHEN. 1999. Comparison of air-displacementplethysmography with hydrostatic weighing and bioelectrical impedance analysis forthe assessment of body composition in healthy adults. Am. J. Clin. Nutr. 69: 898–903.

5. MCCRORY, M.A., T.D. GOMEZ, E.M. BERNAUR & P.A. MOLE. 1995. Evaluation of a newair displacement plethysmograph for measuring human body composition. Med. Sci.Sports Exercise 27: 1686–1691.

6. MOTLEY, H.L. 1957. Comparison of simple helium closed method with oxygen circuitmethod for measuring residual air. Am. Rev. Tuberc. 76: 601–615.

7. DEMPSTER, P. & S. AITKINS. 1995. A new air displacement method for determination ofhuman body composition. Med. Sci. Sports Exercise. 27: 1692–1697.

8. SIRI, W.E. 1961. Body composition from fluid spaces and density: analysis of methods.In Techniques for Measuring Body Composition. J. Brozek & A. Henschel, Eds.: 223–244. National Academy of Sciences, National Research Council. Washington, DC.

9. COLLINS, M.J., M.L. MILLARD-STAFFORD, P.B. SPARLING, T.K. SNOW, L.B. ROSSKOPF,S.A. WEBB & J. OMER. 1999. Evaluation of the BOD POD for assessing body fat incollegiate football players. Med. Sci. Sports Exercise 31: 1350–1356.

10. NUÑEZ, C., A.J. KOVERA, A. PIETROBELLI, S. HESHKA, M. HORLICK, J.J. KEHAYIAS, Z.WANG & S.B HEYMSFIELD. 1999. Body composition in children and adults by air dis-placement plethysmography. Eur. J. Clin. Nutr. 53: 382–387.

11. SARDINHA, L., T. LOHMAN, P.J. TEIXEIRA, D.P. GUEDES & S.B. GOING. 1998. Compari-son of air displacement plethysmography with dual-energy X-ray absorptiometry and3 field methods for estimating body composition in middle-aged men. Am. J. Clin.Nutr. 68: 786–793.

12. IWAOKA, H., T. YOKOYAMA, T. NAKAYAMA, Y. MATSUMURA, Y. YOSHITAKE, T. FUCHI, N.YOSHIKE & H. TANAKA. 1998. Determination of percent body fat by the newly developedsulfur hexafluoride dilution method and air displacement plethysmography. J. Nutr. Sci.Vitaminol. 44: 561–568.