ChildSpecific Thoracic Gas Volume Prediction ... - Wiley Online Library

1 downloads 0 Views 109KB Size Report
TGV prediction equations using anthropometric variables after completing a measured TGV and air-displacement plethysmography test in 224 healthy boys and ...
Child-Specific Thoracic Gas Volume Prediction Equations for Air-Displacement Plethysmography David A. Fields,*§ Holly R. Hull,* A.J. Cheline,† Manjiang Yao,† and Paul B. Higgins‡

Abstract FIELDS, DAVID A., HOLLY R. HULL, A.J. CHELINE, MANJIANG YAO, AND PAUL B. HIGGINS. Childspecific thoracic gas volume prediction equations for airdisplacement plethysmography. Obes Res. 2004;12: 1797–1804. Objective: To develop child-specific thoracic gas volume (TGV) prediction equations for use in air-displacement plethysmography in 6- to 17-year-old children. Research Methods and Procedures: Study 1 developed TGV prediction equations using anthropometric variables after completing a measured TGV and air-displacement plethysmography test in 224 healthy boys and girls (11.2 ⫾ 3.2 years, 45.3 ⫾ 18.7 kg, 149.9 ⫾ 18.5 cm). Study 2 cross-validated the prediction equations in a separate cohort of 62 healthy boys and girls (11.2 ⫾ 3.4 years, 44.2 ⫾ 15.3 kg, 149.4 ⫾ 19.3 cm). Results: In Study 1 (development of TGV prediction equations), the quadratic relationship using height as the independent variable and the measured TGV as the dependent variable yielded the highest adjusted R2 and the lowest SE of estimate in both genders, thus producing the following prediction equations: TGV ⫽ 0.00056 ⫻ H 2 ⫺ 0.12422 ⫻ H ⫹ 8.15194 (boys) and TGV ⫽ 0.00044 ⫻ H 2 ⫺ 0.09220 ⫻ H ⫹ 6.00305 (girls). In Study 2 (cross-validation), no significant difference between the

Received for review December 16, 2003. Accepted in final form July 26, 2004. The costs of publication of this article were defrayed, in part, by the payment of page charges. This article must, therefore, be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. *Department of Health and Sport Sciences, University of Oklahoma, Norman, Oklahoma; †Life Measurement Incorporated, Concord, California; ‡Department of Nutrition Sciences, University of Alabama in Birmingham, Birmingham, Alabama; and §Department of Pediatrics, School of Medicine, University of Oklahoma Health Science Center, Oklahoma City, Oklahoma. Address correspondence to David A. Fields, Department of Health and Sport Sciences, University of Oklahoma, 1401 Asp Avenue Room 104, Norman, OK 73019. E-mail: [email protected] Copyright © 2004 NAASO

predicted and measured TGVs (⫺0.018 ⫾ 0.377 liters) was observed. The regression between the measured TGV and the predicted TGV yielded a slope and intercept that did not significantly differ from the line of identity. Prediction accuracy was good as indicated by a high R2 (0.862) and low SE of estimate (0.369 liters). Discussion: The new child-specific TGV prediction equations accurately, precisely, and without bias estimated the actual TGV of 6- to 17-year-old children. Key words: BOD POD, pediatric, adolescent, body composition

Introduction

Air-displacement plethysmography (ADP)1 is gaining popularity in the determination of pediatric body composition (1– 8). This is partly because of the ease of the testing procedure but also because it has been validated with both traditional (DXA and hydrostatic weighing) and multicompartment (four-compartment) models of body composition (2,8). The measurement of body volume by ADP requires the volume of air in the lungs during normal breathing to be determined; this measurement is part of the normal testing procedures and does not require a separate machine. This air is characterized as the thoracic gas volume (TGV) defined as one-half of tidal volume (TV) plus functional residual capacity (FRC). If the TGV is not measured, total body volume will be underestimated by ⬃40% of TGV, causing body density (Db) to be overestimated and percentage fat to be underestimated (9). However, the measurement of the TGV requires the subject to perform a somewhat complex breathing maneuver. First, the subject is instructed to breathe normally through a plastic tube for several cycles of

1 Nonstandard abbreviations: ADP, air-displacement plethysmography; TGV, thoracic gas volume; TV, tidal volume; FRC, functional residual capacity; Db, body density; %BF, percentage body fat; SEE, SE of estimate; CI, confidence interval.

OBESITY RESEARCH Vol. 12 No. 11 November 2004

1797

Child-Specific TGV Prediction Equations, Fields et al.

inhalation and exhalation. Then, at the point of midexhalation, determined by the systems software, the subject must puff gently against an occluded valve (9). In many instances, children are unable to comply; in one study, 31% of pediatric subjects could not obtain a measured TGV (4). This noncompliance forces investigators to predict TGV using the BOD POD software that uses equations developed in adults (17 to 91 years) (10). In an adult population, no significant difference was observed between the measured (3.591 liters) and predicted (3.537 liters) TGVs using these equations (11). However, several reports indicate that the equations of Crapo et al. are inappropriate for use in children (4,12). Therefore, the purpose of this study was 2-fold: to compare measured TGV with predicted TGV in 6- to 17-year-old children and adolescents and develop childspecific prediction equations and to cross-validate these new TGV equations in a separate cohort of children and adolescents.

Research Methods and Procedures Study Design The project was divided into two separate studies: Study 1 (development of TGV prediction equations), which took 7 weeks to complete; and Study 2 (cross-validation), which took 4 weeks to complete. Subjects for Study 1 were recruited by newspaper ads, study flyers, and a mass e-mail to all staff and faculty of the University of Oklahoma. At the completion of Study 1 and after the development of the new TGV prediction equations, a concerted effort was made to recruit a new cohort of children (using the same recruitment techniques as in Study 1) for Study 2. No child participated in both studies. Before the scheduled day of testing, subjects were asked to be fasted (i.e., no food or drink for at least 6 hours before testing) and were asked to refrain from physical activity on the day of testing. Testing was performed from 7:30 AM to 5:30 PM Monday to Saturday at the Human Body Composition Laboratory at the University of Oklahoma Norman campus by the same technician (H.R.H.) using the same BOD POD. Participants Study 1 included 224 healthy boys and girls (11.2 ⫾ 3.2 years, 45.3 ⫾ 18.7 kg, 149.9 ⫾ 18.5 cm), and Study 2 (cross-validation) included 62 healthy boys and girls (11.2 ⫾ 3.4 years, 44.2 ⫾ 15.3 kg, 149.4 ⫾ 19.3 cm). Children with obstructive lung diseases and asthma were excluded from the study. The study was approved by the University of Oklahoma Institutional Review Board for human use. Subject written informed assent and parent written informed consent were obtained before testing. ADP Total body volume was evaluated with the BOD POD MODEL 2000 using software version 1.69 (Body Compo1798

OBESITY RESEARCH Vol. 12 No. 11 November 2004

sition System; Life Measurement Incorporated, Concord, CA) as previously described (13). Briefly, the testing procedure involved the following steps. First, instrument calibration was conducted before subject entry into the BOD POD. Calibration involved the computation of the ratio of the pressure amplitudes (reference chamber and testing chamber) for the empty chamber and our calibration cylinder (49.556 liters) (9). Second, after the calibration was completed and the testing procedures fully explained, the subject entered the BOD POD for two body volume trials lasting ⬃45 seconds each. All study subjects wore a Speedo swimsuit provided by our laboratory. A measure of subject raw body volume was, thus, determined. Third, the TGV was measured. This stage necessitated that the subject be seated quietly in the BOD POD while breathing through a disposable tube and filter connected to the reference chamber in the rear of the BOD POD. After four or five normal breaths and at the point of midexhalation, the airway was occluded, and the subject was instructed to make two quick and light puffs. For the TGV to be considered successful, the following three criteria had to be met: airway pressure ⬍ 35.0 cm H2O, merit ⬍ 1.0, and a measured TGV ⬎ 1.0 but ⬍9.0 liters (14). The TGV was measured three times with all criteria having to be met; if the criteria were not met, the test was repeated until all three criteria were satisfied. Currently, the BOD POD software allows for the TGV to be estimated using equations developed by Crapo et al. (10). By definition, the TGV is measured at midexhalation, is comprised of the TV and FRC, and is calculated as: TGV ⫽ FRC ⫹ 0.5TV The specific gender equations are listed below: Men: FRC ⫽ 0.0472H ⫹ 0.0090A ⫺ 5.290 Women: FRC ⫽ 0.0360H ⫹ 0.00510A ⫺ 3.182 Where H is height in centimeters and A is age in years. The TV is assumed to be 1.2 in men and 0.7 liters in women. Db can then be calculated as: Db (grams per centimeter cubed) ⫽ M/(subject raw body volume ⫹ 0.40TGV ⫺ SAA) Where M is the mass of the subject and SAA is the surface area artifact. SAA is automatically computed by the computer software and is used to account for the isothermal behavior of air near the subject’s body surface. Percentage body fat (%BF) was then calculated from Db using the Lohman equations (15). Because total body water could affect %BF estimates due to dramatic changes during puberty, we determined %BF from Db by using Lohman’s age-gender adjusted equations (15).

Child-Specific TGV Prediction Equations, Fields et al.

Statistical Analysis The relationships between TGV and anthropometric variables and the associations among different anthropometric variables were first examined using scatter plots and by calculating Pearson correlation coefficients. Prediction equations for TGV were then developed for each gender, using anthropometric measures (height, age, and weight) as predictor variables. Various regression models including quadratic, log-transformed, and linear function were used to relate the predictor variables to measured TGV and were then compared. For models within each function, both main effects and interactions between independent variables were examined. Step-wise multiple regression analyses were used to determine which combination of parameters would best fit the model. In this regard, the coefficient of determination (R2) and the SE of estimate (SEE) were calculated, and predictor variables were retained in the regression model only if their addition resulted in a significant improvement of the explained variance of the dependent variable (as indicated by an increase in the adjusted R2 and a decrease in the SEE) and a substantial change (⬎10%) in ␤ coefficients of the independent variables in the model. The final prediction equations for TGV were further studied by comparing the residual plots (predicted value ⫺ measured value) with the predicted dependent variable and the independent variable(s). For regression models to be considered satisfactory, the equations needed to be symmetrical and homoscedastic. Accuracy of the prediction equations was assessed by comparing measured and predicted TGVs along with the difference in %BF, using each in an independent group (cross-validation group of 62 subjects). Mean values for predicted and measured variables were compared using a paired Student’s t test. The fitness between predicted and measured values was examined by linear regression analysis and by residual plots. For residual analysis, the residuals were plotted against the predicted values, and the plots were examined for curvilinearity (by adding a square term of the predicted variable to the original model and testing for its significance) and heteroscedasticity (by dividing the sample by a median split of predicted variable and testing for homogeneity of variances on the residuals). The 95% confidence intervals (CIs) around the mean difference between predicted and measured variables were also calculated to assess the range of agreement for individuals. Descriptive data are represented as mean ⫾ SD unless otherwise indicated. Statistical analyses were performed by using SPSS 10.0 and SYSTA 9.0 (SPSS Inc., Chicago, IL) with statistical significance set at p ⬍ 0.05. Sample Size Calculations To develop new TGV prediction equations for use in children 6 to 17 years old, a total number of 200 to 250 subjects may be considered appropriate, based on the pre-

vious similar study by Crapo et al. (10) who developed the currently used TGV prediction equations in 245 healthy adults between the ages of 17 and 91 years old. The number of subjects involved in the cross-validation study was determined using the following method: from the study by Demerath et al. (12), the mean difference between predicted and measured TGVs among children was 0.38 ⫾ 0.55 liters. Therefore, to detect a true mean difference between the predicted and measured TGVs, at a significant level of 0.05 and 95% power, 30 children are needed. Assuming a rate of TGV measurement failure of 30% and a possible attrition and noncompliance rate of 10% to 20%, our target sample size of 40 to 60 children may be considered adequate for cross-validating the developed TGV prediction equations for use in children.

Results Study 1 The physical characteristics of the subjects in Study 1 of the study are shown in Table 1. The predicted TGV using the equation by Crapo et al. (2.6 ⫾ 1.0 and 2.5 ⫾ 0.6) was significantly overestimated [as compared with the measured TGV (2.4 ⫾ 1.0 and 2.1 ⫾ 0.7)] in both boys and girls (p ⬍ 0.001). Further, the difference between predicted (from Crapo equation) and measured TGVs was significantly associated with BMI (r ⫽ 0.157, p ⫽ 0.019). However, the magnitude of TGV and other subject characteristics, such as age, weight, and height, did not significantly contribute to the difference between predicted and measured TGVs. The correlation between the measured TGV and anthropometric variables (height, age, and weight) for both genders ranged from r ⫽ 0.71 to 0.89 and were significant (p ⬍ 0.001). Figure 1 shows the relationship between height and measured TGV and was highly significant in both genders (r ⫽ 0.89 for boys; r ⫽ 0.79 for girls, p ⬍ 0.001). Because the correlations of the measured TGV with height and age were stronger than that with body weight in both genders, and height may also be a more reliable parameter than body weight in children under abnormal conditions (e.g., being sick), height and age were included as independent variables in various regression models (including linear, log-transformed, and quadratic functions) for developing the bestfitting model. It was found from step-wise regression analyses that the quadratic relationship using height as the significant independent variable and measured TGV as the dependent variable yielded the highest adjusted R2 and the lowest SEE in both genders (Table 2). Furthermore, the residuals between predicted and measured TGVs were not significantly related to the predicted dependent variable (TGV) (boys, r ⫽ 0.035, p ⫽ 0.715; girls, r ⫽ 0.023, p ⫽ 0.811) and the independent variable (height) (boys, r ⫽ 0.035, p ⫽ 0.712; girls, r ⫽ 0.023, p ⫽ 0.809). This indicates that the equations are symmetrical and homosceOBESITY RESEARCH Vol. 12 No. 11 November 2004

1799

Child-Specific TGV Prediction Equations, Fields et al.

Table 1. Physical characteristics of subjects (N ⫽ 224) for Study 1 Boys (N ⴝ 113)

Age (years) Weight (kg) Height (cm) BMI TGV (liters) %BF Number of subjects in each age interval (years) 6 to 8 9 to 11 12 to 14 15 to 17 Number of subjects in each ethnic group White American African American Native American Hispanic

Girls (N ⴝ 111)

Mean ⴞ SD

Range

Mean ⴞ SD

Range

11.3 ⫾ 3.2 47.0 ⫾ 21.4 152.0 ⫾ 20.4 19.2 ⫾ 4.6 2.4 ⫾ 1.0 19.9 ⫾ 9.0

6.0 to 17.0 16.4 to 108.7 110.0 to 191.5 12.8 to 37.1 0.3 to 5.2 5.1 to 47.7

11.2 ⫾ 3.2 43.6 ⫾ 15.5 147.7 ⫾ 16.2 19.3 ⫾ 3.8 2.1 ⫾ 0.7 22.5 ⫾ 8.3

6.0 to 17.0 17.9 to 90.4 112.0 to 177.8 12.0 to 29.8 0.9 to 4.3 7.9 to 46.7

29 31 31 22

23 39 29 20

100 5 4 4

96 5 6 4

dastic. Consequently, this equation was considered the best fitting equation for predicting TGV in both genders and was used in Study 2 (cross-validation). Study 2 The physical characteristics of the subjects in the cross validation group (Study 2) are shown in Table 3. On a group

basis, the measured and predicted TGVs were highly correlated (r ⫽ 0.93, p ⬍ 0.0001), whereas no significant difference between the predicted and measured TGVs (predicted TGV ⫺ measured TGV ⫽ ⫺0.018 ⫾ 0.377 liters) was observed. Regression of measured TGV on the predicted TGV is shown in Figure 2A. The high R2 (0.862) and low SEE (0.369 liters) indicated good agreement between

Figure 1: Measured TGV in relation to height in 113 boys (F) and 111 girls (E).

1800

OBESITY RESEARCH Vol. 12 No. 11 November 2004

Child-Specific TGV Prediction Equations, Fields et al.

Table 2. Prediction equations for TGV Sex

N

Equation

R2

SEE

Residual SD

Boys Girls

113 111

TGV ⫽ 0.00056 ⫻ H2 ⫺ 0.12422 ⫻ H ⫹ 8.15194 TGV ⫽ 0.00044 ⫻ H2 ⫺ 0.09220 ⫻ H ⫹ 6.00305

0.838 0.650

0.417 0.426

0.413 0.422

TGV, TGV in liters; H, height in centimeters.

the predicted and measured TGVs for individual subjects. The slope and intercept from the regression equation were not significantly different from 1 and 0, respectively (95% CI for slope ⫽ 0.991 to 1.217, 95% CI for the intercept ⫽ ⫺0.491 to 0.055). The TGV residuals plotted against the predicted TGV are shown in Figure 1B. The residual analysis showed that the differences between predicted and measured TGVs were not significantly related to the predicted values (r ⫽ ⫺0.232, p ⫽ 0.07), indicating no bias across the range of TGVs. Additionally, the 95% CI around the mean difference was ⫺0.771 to 0.735 liters. Test for curvilinearity of the residual plot was not significant (p ⫽ 0.18), whereas heteroscedasticity was significant (p ⫽ 0.015). On a group basis, %BF calculated from the measured and predicted TGVs was highly correlated (r ⫽ 0.99, p ⬍ 0.0001), and no significant difference between the %BF

calculated from the predicted and measured TGVs [%BF (predicted TGV) ⫺ %BF (measured TGV) ⫽ 0.02 ⫾ 1.57%BF) was observed. Regression analyses of %BF using the measured TGV on %BF using the predicted TGV are shown in Figure 2C. The prediction accuracy, determined by the R2 (0.974) and the SEE (1.520%BF), indicated excellent agreement between %BF using the predicted and measured TGVs for individual subjects. The slope and intercept from the regression equation were not significantly different from 1 and 0, respectively (95% CI for slope ⫽ 0.92 to 1.0, 95% CI for the intercept ⫽ 0.21 to 2.2). The %BF residuals plotted against %BF using the predicted TGV are shown in Figure 2D. The residual analysis showed that differences between %BF using the predicted and %BF using the measured TGV were significantly related to the predicted %BF values (p ⫽ 0.04). Additionally, the 95% CI around the mean difference was

Table 3. Physical characteristics of subjects (N ⫽ 62) for Study 2 Boys (N ⴝ 31)

Age (years) Weight (kg) Height (cm) BMI TGV (liters) %BF Number of subjects in each age interval (years) 6 to 8 9 to 11 12 to 14 15 to 17 Number of subjects in each ethnic group White American African American Native American Hispanic

Girls (N ⴝ 31)

Mean ⴞ SD

Range

Mean ⴞ SD

Range

11.32 ⫾ 3.5 45.0 ⫾ 15.9 151.9 ⫾ 21.2 19.3 ⫾ 4.7 2.3 ⫾ 1.2 18.8 ⫾ 2.9

6.0 to 17.0 17.7 to 78.8 106.9 to 189.5 12.8 to 37.1 0.7 to 5.0 14.7 to 26.4

11.0 ⫾ 3.3 43.3 ⫾ 14.9 146.9 ⫾ 17.2 19.3 ⫾ 3.8 1.7 ⫾ 0.6 19.5 ⫾ 3.8

6.0 to 17.0 20.6 to 75.5 114.3 to 173.2 11.9 to 29.8 0.9 to 2.9 13.5 to 29.3

9 6 8 8

6 13 5 7

26 3 2 0

25 4 2 0

OBESITY RESEARCH Vol. 12 No. 11 November 2004

1801

Child-Specific TGV Prediction Equations, Fields et al.

Figure 2: Linear regression and residual analyses for comparison of measured TGV and predicted TGV and %BF using measured TGV and %BF using predicted TGV in 62 children (31 boys F, 31 girls E). (A and B) Regression of measured TGV on predicted TGV (R2 ⫽ 0.862, SEE ⫽ 0.369 liters) and the TGV residuals plotted against predicted TGV (r ⫽ ⫺0.232, p ⫽ 0.07). (C and D) Regression of %BF using measured TGV on %BF using predicted TGV (R2) ⫽ 0.974, SEE ⫽ 1.52%BF) and the %BF residuals plotted against %BF using predicted TGV (r ⫽ 0.283, p ⫽ 0.03).

⫺3.12 to 3.15%BF. Test for curvilinearity was significant (p ⫽ 0.04), and heteroscedasticity was not significant (p ⫽ 0.429).

Discussion We developed and examined the validity of new childspecific TGV equations for use with the BOD POD. Gender specific equations were formulated from a sample of 224 children and adolescents, using height as the predictor variable. We then cross-validated the equations in an independent sample of 62 children and adolescents. The new prediction equations resulted in TGV and %BF estimates that 1802

OBESITY RESEARCH Vol. 12 No. 11 November 2004

did not significantly differ from measured TGV or its respective %BF. Because TGV measurements have proven challenging in pediatric populations, these equations represent an advancement in the application of the BOD POD to pediatric body composition assessment. With the rise in pediatric obesity, ADP is emerging as a popular technique for the assessment of body fatness in children. At first inspection, a typical ADP test seems simple and quick; however, obtaining a measured TGV in a pediatric population has proven challenging and difficult for many investigators (4). Through personal communications with other pediatric centers, we estimate that in 35% of tests, investigators are unable to attain a measured TGV

Child-Specific TGV Prediction Equations, Fields et al.

(data not shown). In this study, we were able to achieve a measured TGV in all subjects, although 75% of the children needed three trials, and 20% required ⱖfive trials before a valid TGV measurement could be obtained. Currently, investigators unable to obtain a valid TGV measurement in children are forced to use the prediction equations of Crapo et al. or other child-specific prediction equations (7,16). The following example demonstrates the effect of using those predicted TGV to estimate %BF: a 9-year-old boy who weighs 39 kilograms with an SAA of ⫺0.573 and an actual TGV of 2 liters has a %BF of 32.4%. An error in the TGV of ⫾200 mL would result in an error of 1.1% fat units (31.3 to 33.5%BF), and an error of ⫾400 mL (as seen in this study) would result in an error of 2.2% fat units (30.2 to 34.6%BF). Demerath et al. found that predicted TGV using the Crapo et al. equations was significantly higher than measured TGV in children and produced a significant overestimate of %BF (1.4%) (12). However, use of the Crapo et al. equations in another study in children did not affect body composition estimates by the BOD POD (8). Despite the presence of equivocal data, using the adult prediction equation by Crapo et al. in a pediatric population may produce invalid BF estimates and, therefore, is not recommended. The new gender-specific prediction equations adequately predicted TGV and yielded accurate %BF estimates in an independent sample of children and adolescents. These equations were derived from male and female subjects ranging from 6 to 17 years old. The age range of the cross-validation sample was similar. The ability of these equations to predict TGV in children younger than 6 years is not known and will require further investigation. Of note, the new equations were derived from a direct measure of TGV from the BOD POD system. Hence, they are not reliant on assumptions regarding TV, nor do they require independent prediction equations for estimating TV (as with current child-specific equations). Moreover, investigators unable to derive a measure of TGV can feel more comfortable with a straightforward child-/adolescent-specific prediction equation. There were some strengths specific to this study: the TGV was measured in 224 children and crossvalidated in a separate cohort of 62 children, and the same technician (H.R.H.) performed all testing procedures including the TGV measurement in all 286 children. A possible limitation of the study was that subject population was relatively homogenous, with white American, AfricanAmerican, Native American, and Hispanic individuals making up 87%, 6%, 5%, and 3%, respectively. Thus, the accuracy of these equations in non-white ethnicities should be evaluated. It is our recommendation that TGV be measured according to current guidelines whenever possible (14). However, if the subject becomes uncooperative or if time constraints exist, these “practical guidelines” may help in improving compliance: to alleviate confusion, keep the instructions

basic and simple; perform a practice TGV test before the actual test; tell the subject to “puff” as opposed to “pant”; and have the subject breathe rhythmically, with the inhalation and exhalation having a definite starting and ending point. This can be achieved quite simply by having the subject inhale and exhale according to your commands from outside the testing chamber. In conclusion, we generated child-/adolescent-specific TGV prediction equations for use in the BOD POD body composition testing. Our data support the use of these new equations for the estimation of the TGV in 6- to 17-year-old children when TGV is unattainable or impractical. We hope that these equations will improve the accuracy of the BOD POD for body composition assessment in children and adolescents.

Acknowledgments We thank Megan A. McCrory for editorial comments in the development of this manuscript. This study was partially supported by Life Measurement Incorporated (Concord, CA). A.J.C. and M.Y. are employed by Life Measurement Incorporated. References 1. Nun˜ez C, Kovera AJ, Pietrobelli A, et al. Body composition in children and adults by air displacement plethysmography. Eur J Clin Nutr. 1999;53:382–7. 2. Fields DA, Goran MI. Body composition techniques and the four-compartment model in children. J Appl Physiol. 2000;89: 613–20. 3. Dewit O, Fuller NJ, Fewtrell MS, Elia M, Wells JCK. Whole body air displacement plethysmography compared with hydrodensitometry for body composition analysis. Arch Dis Child. 2000;82:159 – 64. 4. Lockner DW, Heyward VH, Baumgartner RN, Jenkins KA. Comparison of air-displacement plethysmography, hydrodensitometry, and dual X-ray absorptiometry for assessing body composition of children 10 to 18 years of age. Ann NY Acad Sci. 2000;904:72– 8. 5. Wells JCK, Fuller NJ. Precision of measurement and body size in whole-body air-displacement plethysmography. Int J Obes. 2001;25:1161–7. 6. Nicholson JC, McDuffie JR, Bonat SH, et al. Estimation of body fatness by air displacement plethysmography in African American and white children. Pediatr Res. 2001;50:467–73. 7. Wells JC, Fuller NJ, Wright A, Fewtrell MS, Cole TJ. Evaluation of air-displacement plethysmography in children aged 5-7 years using a three-component model of body composition. Br J Nutr. 2003;90:699 –707. 8. Gately PJ, Radley D, Cooke CB, et al. Comparison of body composition methods in overweight and obese children. J Appl Physiol. 2003;95:2039 – 46. 9. Dempster P, Aitkens S. A new air displacement method for the determination of human body composition. Med Sci Sports Exerc. 1995;27:1692–7. OBESITY RESEARCH Vol. 12 No. 11 November 2004

1803

Child-Specific TGV Prediction Equations, Fields et al.

10. Crapo RO, Morris AH, Clayton PD, Nixon CR. Lung volumes in healthy nonsmoking adults. Bull Eur Physiopath Respir. 1982;18:419 –25. 11. McCrory MA, Mole´ PA, Gomez TD, Dewey KG, Bernauer EM. Body composition by air-displacement plethysmography by using predicted and measured thoracic gas volumes. J Appl Physiol. 1998;84:1475–9. 12. Demerath EW, Guo SS, Chumlea WC, Towne B, Roche AF, Siervogel RM. Comparison of percent body fat estimates using air displacement plethysmography and hydrodensitometry in adults and children. Int J Obes. 2002;26: 389 –97.

1804

OBESITY RESEARCH Vol. 12 No. 11 November 2004

13. McCrory MA, Gomez TD, Bernauer EM, Mole´ PA. Evaluation of a new air displacement plethysmograph for measuring human body composition. Med Sci Sports Exerc. 1995;27: 1686 –91. 14. BOD POD Body Composition System: Operator’s Manual. Concord, CA: Life Measurement, Inc.; 2000. 15. Lohman TG. Assessment of body composition in children. Pediatr Exerc Sci. 1989;1:19 –30. 16. Rosenthal M, Cramer D, Bain SH, Denison D, Bush A, Warner JO. Lung function in white children aged 4 to 19 years: II. Simple breath analysis and plethysmography. Thorax. 1993;48:803– 8.