Validation of a new pediatric air-displacement plethysmograph for assessing body composition in infants 1 4

Transcription

1 Validation of a new pediatric air-displacement plethysmograph for assessing body composition in infants 1 4 Guansheng Ma, Manjiang Yao, Yan Liu, Aiwei Lin, Hui Zou, Alessandro Urlando, William W Wong, Laurie Nommsen-Rivers, and Kathryn G Dewey ABSTRACT Background: The accurate measurement of body composition is useful in assessments of infant growth and nutritional status. Objective: This study evaluated the reliability and accuracy of a new air-displacement plethysmography (ADP) system for bodycomposition assessment in infants. Design: Between- and within-day reliability was assessed by comparing the percentage body fat (%BF) obtained on consecutive days and on the same day, respectively, in 36 full-term infants. Accuracy was assessed by comparing %BF measured with the use of ADP and %BF measured with the use of deuterium ( 2 H 2 O) dilution in 53 infants. Results: There were no significant differences in %BF between days ( %BF) or within days ( %BF). Mean between- and within-day test-retest SDs of 0.69 and 0.72%BF, respectively, indicated excellent reliability. The %BF measurements obtained by using ADP were not significantly influenced by infant behavioral state. Mean %BF obtained by using ADP (20.32%BF) did not differ significantly from that obtained by using 2 H 2 O dilution (20.39%BF), and the regression line [%BF( 2 H 2 O) 0.851%BF (ADP) 3.094] gave a high R 2 (0.76) and a low SEE (3.26). The 95% limits of agreement between ADP and 2 H 2 O( 6.84%BF, 6.71%BF) were narrower than those reported for other bodycomposition techniques used in infants. Individual differences between the 2 methods were not a function of body mass or fatness. Conclusion: ADP is a reliable and accurate instrument for determining %BF in infants, and it has the potential for use in both research and clinical settings. Am J Clin Nutr 2004;79: KEY WORDS Body composition, percentage body fat, infants, air-displacement plethysmography, deuterium dilution, total body water INTRODUCTION Accurate assessment of body composition during infancy provides key information for monitoring and evaluating growth patterns, efficacy of diet and medical interventions, progression of chronic disease, and recovery from malnutrition. In addition, because of the significant association between early infant development and childhood obesity (1), and in light of the remarkable increase in the prevalence of childhood obesity in the United States and worldwide (2, 3), body-composition assessment during infancy has the potential to become an important research and diagnostic tool for the study and prevention of child and adult obesity. Despite the potential usefulness of body-composition assessment in infants, the methods that are available to researchers and clinicians, such as isotope dilution, dual-energy X-ray absorptiometry (DXA), total-body electrical conductivity, bioelectrical impedance analysis, total body potassium, and skinfoldthickness measurements, have been used sporadically and with very limited success. Among the reasons for this are ethical and practical considerations, complexity, cost, lack of sufficient data regarding validity, or any combination of those factors. Researchers and clinicians therefore continue to seek a bodycomposition assessment method for use in the pediatric population that is safe, noninvasive, easy to perform, comfortable for subjects, relatively inexpensive, reliable, and accurate. Air-displacement plethysmography (ADP) is a densitometric technique in which the percentage body fat (%BF) is assessed from the direct measurement of subject mass and volume. Subject mass is measured on an electronic scale. Subject volume is measured in an enclosed chamber by applying gas laws that relate pressure changes to volumes of air in the enclosed chamber. With the use of a subtraction technique, these air volumes are used to calculate the subject s volume. Body density is then computed from the measured subject mass and volume and inserted into a standard formula for estimating %BF according to a 2-compartment model, such as the models of Siri (4) or Brozek et al (5) for adults, Lohman (6) for children, and Butte et al (7) for infants. ADP has been successfully used to measure the body composition of children and adults (8 14), and a study by Yee et al (15) suggests that it is also a valid method for use in the elderly. 1 From the National Institute for Nutrition and Food Safety, Chinese Center for Disease Control and Prevention, Beijing (GM); Life Measurement, Inc, Concord, CA (MY and AU); the Maternity and Child Care Hospital (YL and HZ) and the School of Public Health (AL), Shandong University, Jinan, China; the US Department of Agriculture/Agricultural Research Service Children s Nutrition Research Center, Department of Pediatrics, Baylor College of Medicine and Texas Children s Hospital, Houston (WWW); and the Department of Nutrition, University of California, Davis (LN-R and KGD). 2 The contents of this publication do not necessarily reflect the views or policies of the US Department of Agriculture, nor does mention of trade names, commercial products, or organizations imply endorsement by the US Government. 3 Supported by Life Measurement, Inc, Concord, CA. 4 Address reprint requests to KG Dewey, Department of Nutrition, University of California, One Shields Avenue, Davis, CA kgdewey@ucdavis.edu. Received May 15, Accepted for publication September 2, Am J Clin Nutr 2004;79: Printed in USA American Society for Clinical Nutrition 653

2 654 MA ET AL Previous attempts to apply ADP to the infant population have failed because of technical and practical issues (16, 17). The only commercially available ADP system for use in children, adults, and the elderly is the BOD POD Body Composition Tracking System (Life Measurement, Inc, Concord, CA). The success of the BOD POD system in terms of its high accuracy and reliability in body-composition assessment, as indicated in published research and clinical studies (11), prompted the development of a new ADP system for infants, the PEA POD Infant Body Composition System (Life Measurement, Inc). Two background studies, one using inanimate objects (18) and the other using bovine tissue phantoms (19), were conducted to evaluate the performance of this ADP system. The design of the ADP system and the reliability and accuracy results from these studies indicate that ADP has the potential to provide a bodycomposition measurement tool that is easily used by operators, comfortable for infants, reliable, and accurate. In addition, results from our preliminary evaluation of the ADP system used in human infants indicated that the reliability of ADP was significantly better than that of currently available techniques for bodycomposition assessment of infants (20). In the present study, a comprehensive evaluation of the reliability and accuracy of the ADP system in assessing infant body composition was conducted. Deuterium ( 2 H 2 O) dilution was the reference technique used to assess the accuracy of the ADP system. SUBJECTS AND METHODS Subjects In total, 80 full-term healthy infants took part in the present study. The study was conducted at the University of California, Davis (n 17), and at the Jinan Maternity and Child Care Hospital, Jinan, China (n 63). The racial distribution was 64 Asian and 16 white infants. The reliability of the ADP system was assessed in a subsample of 36 infants (20 male, 16 female): 17 from Davis and 19 from Jinan. The racial distribution was 20 Asian and 16 white infants. The ranges of age, body mass, and length were wk, kg, and cm, respectively. Subjects were selected so that the 3 categories of body mass would include similar numbers of subjects: n 14 weighing 2 4 kg, 12 weighing 4 6 kg, and 10 weighing 6 8 kg. The accuracy of the ADP system was assessed in a subsample of 63 Asian infants (32 male, 31 female), all from Jinan. Of these 63 infants, 2 dropped out during the study, and results from another 5 were excluded from data analysis because they regurgitated 1 g milk within2hofthe 2 H 2 O dose administration. In addition, the %BF values determined from 2 H 2 O dilution in 3 of these subjects were excluded from data analysis because the values were extreme outliers: 2 were negative ( 8.38 and 2.25), and 1 value was considered impossibly low (1.89). Because these 3 subjects were involved only in the accuracy portion of the study, their corresponding %BF values obtained by using ADP (13.43, 13.84, and 6.08, respectively) were also excluded from data analysis. Consequently, results from 53 subjects were used for the analysis of ADP accuracy. For these infants, the age, body mass, and length ranges were wk, kg, and cm, respectively. Of the 53 subjects who took part in this portion of the study, 19 were also involved in the assessment of ADP reliability. At both study sites, approval to conduct the study was obtained from the respective human subjects review committees. Written informed consent was obtained from the mothers of the subjects before any measurements were conducted. General protocol The reliability of ADP measurements was assessed by using the following protocol: 1 ADP test was conducted on study day 1 [test 1, day 1 (T1D1)], and 2 ADP tests were conducted on study day 2 [test 1, day 2 (T1D2) and test 2, day 2 (T2D2), respectively]. The between- and within-day reliability of the ADP system were then evaluated by comparing %BF from T1D1 with that from T1D2 and that from T1D2 with that from T2D2, respectively. For the 19 subjects who participated in both the reliability and accuracy portions of the study, an additional ADP test was performed on study day 1 [test 2, day 1 (T2D1)], and the data from that test were used with data from T1D1 to assess the accuracy of ADP measurements. The accuracy of ADP measurements was assessed by using the following protocol: 2 ADP tests were performed on study day 1 (T1D1 and T2D1), and those tests were followed by performance of a 2 H 2 O dilution measurement. The accuracy of the ADP system was evaluated by comparing the mean %BF from the 2 ADP tests performed on day 1 with the %BF calculated from the total-body water value determined by using 2 H 2 O dilution on the same day. ADP system A detailed description of the physical design and operating principles of the ADP system is provided elsewhere (18, 19). Briefly, the ADP system is a pediatric air-displacement plethysmograph designed to assess the body composition of infants by the direct measurement of body mass and body volume and the application of the principles of whole-body densitometry. The ADP system components are mounted on or housed in a movable cart. The test chamber, electronic scale, and monitor are mounted on the cart s top surface, and the reference chamber, calibration volume container, and air circulation and air heating systems are housed inside the cart along with the electronic components, printer, and computer. A volume-perturbing diaphragm is mounted between the test and reference chambers. A calibration valve allows the test chamber to be connected to the calibration volume container. Pressure transducers are connected to the 2 chambers. The air circulation system continuously circulates air from the outside environment to the test chamber. The temperature of the air circulated within the test chamber is also maintained constant by the air heating system in this study, at 31 C. The heating system can be set, however, to keep the temperature of the air in the test chamber constant at temperatures between 31 and 37 C. The relations between pressure and volume expressed by Boyle s law and Poisson s law, which describe the behavior of air under isothermal and adiabatic conditions, respectively, are used in the ADP system to measure body volume. When the system is in operation, the diaphragm s oscillations create in the 2 chambers sinusoidal volume perturbations that are equal in magnitude but opposite in sign. For a known reference chamber volume,

3 PEDIATRIC AIR-DISPLACEMENT PLETHYSMOGRAPH 655 varying test chamber volumes are a linear function of the ratios of the pressure perturbations in the 2 chambers. Two volume calibrations, one before and one after the measurement of subject s volume, are performed to increase the accuracy of ADP volume measurements. Because ADP measures volume with the assumption that all the air in the test chamber behaves adiabatically, a correction for air that is maintained under isothermal conditions (ie, air close to the subject s surface and in the subject s lungs) is then performed automatically. Using the body mass and volume measurements, the ADP system then automatically calculates %BF by using a classic densitometric approach and age- and sex-specific fat-free mass (FFM) density values. For this study, the FFM density values used were obtained from a multicompartmental study by Butte et al (7). Before an ADP system test, the following specific steps must be performed to ensure the accuracy of the mass and volume measurements. The subject s length is measured, and that measurement is used to quantify the isothermal volumes of air close to the subject s surface and in the subject s lung. On day 1 of this study, subject length was measured twice to the nearest 0.1 cm with the use of a length board (CMS Weighing Equipment, London). The mean of the 2 length measurements was used for all subsequent ADP system tests performed on days 1 and 2. To allow for a precise quantification of the amount of air behaving isothermally in the test chamber, subjects must not be wearing any clothing during the ADP test, and the amount of air behaving isothermally in the proximity of their hair must be reduced. In this study, subjects were tested while nude, and their hair was smoothed down with the use of baby oil. Before initiating data collection for this study, the logistic aspects of ADP system testing were assessed at Davis in a sample of 15 infants. This assessment comprised the evaluation of the optimal timing of measurements, of the procedures for collecting data, and of the initial studies of reliability by using a firstgeneration ADP instrument. The data obtained were used to modify the instrument to enhance reliability and comfort through altering the frequency of the diaphragm s oscillations and adding a heating system and an air circulation system. At both locations, the same ADP system was used, and the ability of the ADP system to measure volume was evaluated before each testing session by measuring the volume of a National Institute of Standards and Technology (Gaithersburg, MD) traceable aluminum cylinder. Test results indicated consistent and similar performance at both locations. The scale was also calibrated before each testing session. Each ADP test was performed according to the following sequence. Subject data (sex, length, and age) were entered, and subject mass was measured on the ADP electronic scale while the first automatic volume calibration was taking place. The volume calibration procedure lasted 2 min, and at its end, the test chamber door opened automatically. The tray was then pulled out of the clear acrylic test chamber, and the subject was placed in it. To start the first of 3 volume measurements, the tray and subject were pushed back into the chamber, and the test chamber door was closed. Pressure changes were then measured in the test and reference chambers for 2 min. Data collection ended with the automatic opening of the test chamber door. The same volume measurement procedure was performed twice more. After the subject was removed from the test chamber, a second automatic volume calibration, identical to the first one, was performed. The subject s body volume was computed as the average of the volume values obtained during the 3 measurement periods. The %BF results were then displayed. For each of the 3 volume measurements performed in each test, the subject s behavioral state was recorded according to the following categories: deep sleep, sleep with rapid eye movement, drowsy, quiet and alert, awake and active, and crying intensely. It was also noted whether subjects urinated or defecated during a measurement. On 3 occasions, as a result of subjects crying and at the request of their parents, tests were interrupted. These tests were repeated after the infants were comforted. Deuterium dilution The following procedure was followed to measure subjects total body water by using 2 H 2 O dilution. On completion of the 2 ADP tests on day 1, a baseline blood sample ( ml) from each infant was drawn into a potassium- and EDTA-coated microvette (0.5 ml; Sarstedt Inc, Princeton, NJ) by heel-stick (Quickheel Lancet; Becton Dickinson and Company, Franklin Lakes, NJ). Each infant then received by mouth a dose of1gof 10% 2 H 2 O/kg body mass through a 10-mL plastic syringe to which a short length of feeding tube was attached. The exact dose given was determined by weighing the feeding apparatus on an analytic balance before and after dose administration. Two hours after dose administration, a second blood sample ( ml) was collected by heel-stick. The exact times of dosing and of the baseline and 2-h postdose blood samplings were recorded. Within 3hofblood sample collection, the plasma was separated by centrifugation at 2500 rpm for 20 min at 4 C. All plasma samples were transferred to O-ring screw-top microtubes (1.25 ml; Sarstedt Inc) and stored at 20 C before they were shipped on dry ice to the US Department of Agriculture/Agricultural Research Service Children s Nutrition Research Center in Houston for gas isotope ratio mass spectrometry analysis (21). For hydrogen isotope ratio measurements, 10 L of plasma without further treatment was reduced to hydrogen gas with the use of 200 mg zinc reagent at 500 C for 30 min (22). The 2 H: 1 H isotope ratios of the hydrogen gas were measured by using a Finnigan -E gas isotope ratio mass spectrometer (Finnigan MAT, San Jose, CA). The results are expressed in delta ( ) per mil ( 0 / 00 ) units, which are defined as follows: 2 H( 0 / 00 ) (R x /R s 1) 10 3 (1) where R x and R s are the 2 H: 1 H isotope ratios of the sample and standard, respectively. The 2 H values were normalized against 2 international water standards: Vienna-Standard Mean Ocean Water and Standard Light Antarctic Precipitation (23). Total body water was calculated as follows: Total body water (mol) (d A E a )/(a E d 18.02) (2) where d is the dose of 2 H 2 O in grams, A is the amount of laboratory water in grams used in the dose dilution, a is the amount of 2 H 2 O in grams added to the laboratory water in the dose dilution, E a is the rise in 2 H abundance in the laboratory water after the addition of the isotopic water, and E d is the rise in 2 H abundance in the plasma after 2hofequilibration. FFM was determined from the total body water by using ageand sex-specific hydration coefficients for FFM calculated with

4 656 MA ET AL the use of data from a multicompartmental study performed in infants by Butte et al (7). The %BF was computed as %BF [(body mass FFM)/body mass] 100% (3) Statistical analysis Data were analyzed by using SPSS for WINDOWS (version 10.0; SPSS Inc, Chicago) and SAS for WINDOWS (version 8e; SAS Institute Inc, Cary, NC). Statistical significance was accepted at P Values are expressed as x SD. Paired-sample t tests were performed to detect significant between-day (T1D1 compared with T1D2) and within-day (T1D2 compared with T2D2) differences in %BF. To assess the agreement of the between- and within-day %BF estimates, linear regression analyses and Bland-Altman analyses were carried out. Between- and within-day SDs and CVs for %BF estimates were calculated, and the nonparametric Kruskal-Wallis H test was used to assess whether SDs and CVs were a function of body mass. Considering separately each of the 3 volume/%bf measurements during each ADP test (with a total of 3 ADP tests: T1D1, T1D2, and T2D2), we conducted a mixed-effect linear model procedure (using PROC MIXED of SAS), which uses a repeated-measures analysis approach, and Tukey s honestly significant difference multiple-comparison procedure to examine the least-squares means and to determine whether within-subject %BF estimates obtained from ADP differed by behavioral state and by urination. We made comparisons among the behavioral states of quiet and alert, awake and active, and crying intensely. Because the percentage of ADP measurements completed in other behavioral states (ie, sleep, drowsy, and defecation) was 3%, and thus the sample sizes in these states were very low, these rare states were excluded from the analysis. In addition, regression analysis was performed to examine whether %BF estimates obtained by using ADP differed significantly between white and Asian infants. Potential confounding variables, including age, sex, and body weight, were included in the regression model. An independent-sample t test was performed to examine whether the reliability of the ADP system, indicated by betweenand within-day SDs and CVs, was influenced by race. A paired-sample t test was conducted to detect a significant difference between the %BF values determined by ADP and by 2 H 2 O dilution. To determine how well %BF obtained from ADP reflected 2 H 2 O-derived %BF, %BF determined by 2 H 2 O was regressed on %BF obtained from ADP. Bland-Altman analysis was also conducted to assess the agreement between the 2 methods in individual subjects and for trends associated with subjects body fatness. Moreover, regression analysis was performed to assess whether the individual differences in the %BF values obtained by using the 2 methods were a function of body mass. RESULTS The physical characteristics of the subjects involved in both the reliability and accuracy portions of the study are shown in Table 1. The mean and range values for age, body mass, and height of the infants who were involved in the reliability portion of the study were comparable with those of the subjects who were involved in the accuracy portion of the study. Reliability Subjects mean %BF values obtained from T1D1, T1D2, and T2D2 were , , and %BF, respectively (Table 2). The mean difference in %BF between days ( , P 0.08) and within days ( , P 0.51) did not differ significantly from zero. Linear regression and Bland-Altman analyses (Figure 1) of %BF estimates obtained on different days and within the same day indicated good agreement. The regression lines (Figure 1, A and C) did not differ significantly from the line of identity (Y X), and the regression equations indicated a relatively low SEE (1.22 and 1.41 for between- and within-day tests, respectively) and a very high R 2 (0.96 and 0.94 for between- and within-day tests, respectively). Between- and within-day differences in %BF for individual subjects were within narrow ranges [95% CI (Figure 1, B and D): 2.9, 1.9%BF and 2.7, 3.1%BF for between- and within-day tests, respectively], and the individual differences were not a function of body fatness [r 0.06, P 0.74 for between-day tests (Figure 1, B); r 0.17, P 0.33 for within-day tests (Figure 1, D)]. The mean within-subject SDs and CVs for %BF estimates were %BF and %, respectively, for between-day tests and %BF and %, respectively, for within-day tests. There were no significant differences in within-subject SDs among the 3 body mass groups (2 4, 4 6, and 6 8 kg) between-day: , , and %BF (P 0.28); within-day: , , and %BF (P 0.70), respectively. The corresponding within-subject CVs for the 2 4, 4 6, and 6 8kg groups were %, %, and %, respectively (P 0.09), for between-day tests and %, %, and %, respectively (P 0.08), for within-day tests. The proportions of ADP measurements completed in each behavioral state were as follows: sleep, 0.6%; drowsy, 0.9%; quiet and alert, 30.3%; awake and active, 45.4%; crying intensely, 22.8%; urination, 10.1%; and defecation, 2.9%. As shown in Table 3, there were no significant differences in %BF among the 3 main categories of behavioral state (P 0.59, 0.10, and 0.10 for comparisons of quiet and alert with awake and active, awake and active with crying, and quiet and alert with crying, respectively). Urination during the measurement did not significantly influence the ADP %BF estimates (P 0.69). In addition, %BF estimates by ADP did not differ significantly between white and Asian infants after we controlled for age, sex, and body weight. Furthermore, the between- and within-day SDs and CVs did not differ significantly between the white and the Asian infants, which indicated that the reliability of the ADP system was not influenced by race. TABLE 1 Subjects physical characteristics 1 Reliability (n 20 M, 16 F) Accuracy (n 25 M, 28 F) 2 Age (wk) ( ) ( ) Body mass (kg) ( ) ( ) Height (cm) ( ) ( ) 1 Values are x SD; range in parentheses. 2 Nineteen of these infants also participated in the reliability study.

5 PEDIATRIC AIR-DISPLACEMENT PLETHYSMOGRAPH 657 TABLE 2 Percentage body fat (%BF) estimates determined by using air-displacement plethysmography (ADP) and deuterium dilution 1 Reliability Accuracy ADP %BF (test 1, day 1) ( ) ( ) %BF (test 2, day 1) ( ) Mean %BF (day 1 tests) ( ) %BF (test 1, day 2) ( ) %BF (test 2, day 2) ( ) Deuterium dilution %BF ( ) 1 All values are x SD; range in parentheses. Accuracy Mean %BF values determined by ADP and by 2 H 2 O dilution were and %BF, respectively (Table 2). The mean difference ( %BF) did not differ significantly from zero (P 0.89). Linear regression analysis (Figure 2, left) showed that %BF determined by ADP explained 76% of the variance in 2 H 2 O-derived %BF, and the SEE of 3.26 was considered good according to Lohman s criteria (24). The 95% limits of agreement from Bland-Altman analysis (Figure 2) were 6.84%BF, 6.71%BF. There was no trend in individual differences in %BF between the 2 methods as %BF varied (r 0.06, P 0.68). Moreover, the differences in %BF between the 2 methods were not significantly related to body mass (r 0.04, P 0.80). The proportion of ADP measurements completed in each behavioral state was as follows: sleep, 0.4%; drowsy, 3.7%; quiet and alert, 32.7%; awake and active, 40.5%; crying intensely, 22.8%; urination, 9.8%; and defecation, 2.1%. Mean %BF values obtained from ADP ( %BF) and expected %BF values ( %BF) compared favorably; the latter values, based on the subjects age and sex, were calculated by using data from a multicompartmental study in infants conducted by Butte et al (7). DISCUSSION This study represents the first comprehensive comparison of the reliability and accuracy of the ADP system and of 2 H 2 O dilution in an infant population whose age ( wk), body FIGURE 1. A and C: Linear regressions of percentage body fat (%BF) results from air-displacement plethysmography (ADP) for between-day tests [test 1, day 1 (T1D1) compared with test 1, day 2 (T1D2)] and within-day tests [T1D2 compared with test 2, day 2 (T2D2)]. The linear regression equations for between- and within-day tests are %BF T1D %BF T1D (R , SEE 1.22, P 0.001) and %BF T2D %BF T1D (R , SEE 1.41, P 0.001), respectively. B and D: Corresponding Bland-Altman plots for between- and within-day tests, respectively. The solid lines represent mean differences between tests (bias), and the dotted lines represent 2 SDs from the mean differences (95% limits of agreement).

6 658 MA ET AL TABLE 3 Percentage body fat (%BF) measured by using air-displacement plethysmography (ADP) according to infant behavioral state and urination during measurement Value 1 Measurements completed 2 n Behavioral state Quiet and alert Awake and active Crying intensely Urination Yes No Least-squares x SE (from mixed linear model). No significant effects on ADP %BF measurements (mixed linear model with Tukey s honestly significant difference multiple-comparison procedure). 2 The maximum n is 324 because 36 infants participated in the reliability study, and each had 9 ADP values (3 volume measurements 3 ADP tests). The actual n totals are 324 because measurement periods during which the infants were asleep or drowsy were excluded from the behavioral state comparisons, and measurement periods during which the infants defecated were excluded from the comparisons of urination and no urination. mass ( kg), and %BF ( %BF) represent the expected range for infants between birth and 6 mo of age, for whom the ADP system is designed. Reliability The ADP system was found to be a reliable instrument for determining %BF in infants. The mean between-day and withinday SDs for %BF estimates were 0.69%BF and 0.72%BF, respectively, which correspond to mean CVs of 4.9% and 5.1%, respectively. These CVs are comparable with those obtained by using DXA ( %; 25, 26) and total-body electrical conductivity ( %; 27 29), the 2 techniques that have been validated and are commonly used in infant populations (29 31). In addition, the ADP system showed narrow ranges of betweenday ( 2.9 to 1.9) and within-day ( 2.7 to 3.1) individual variations in %BF estimates in subjects spanning a wide range with respect to age, body mass, and overall %BF. These findings indicate that the ADP system has the potential to be used in comparative and longitudinal studies and in clinical settings where changes in body composition between groups or over time should be assessed periodically and monitored closely. Furthermore, the ADP %BF estimates were not significantly affected by infants behavioral states, including quiet and alert, awake and active, crying intensely, and urination. This suggests that, in a population in which compliance with a test protocol cannot be expected, the ADP system can provide consistent %BF estimates under a wide range of conditions. Accuracy Results of this study indicate that the %BF estimates obtained by using ADP and by using 2 H 2 O dilution had excellent agreement indeed, the means were virtually identical and the regression equation of %BF by 2 H 2 O against %BF by ADP gave high R 2 and low SEE values. These results are consistent with the recent findings of Sainz and Urlando (19), who conducted a study to evaluate the performance of the ADP system by using bovine tissue phantoms in which the range in mass ( kg) and percentage fat ( %) was similar to that expected in infants 0 6 mo of age. In that study, the ADP system provided a highly accurate estimate of percentage fat (R , SEE 0.60) compared with chemical analysis of percentage fat. In the present study, individual variations in agreement between %BF estimates determined by ADP and by 2 H 2 O dilution (95% limits of agreement: 6.84, 6.71%BF) were smaller than those from a previous study in which body-composition methods underwent pairwise comparison (32). In a comparison of 2 H 2 O and total-body electrical conductivity, the 95% limits of agreement for %BF in infants aged 0.5, 3, and 6 mo were 15.4, 10.5; 4.4, 13.7; and 7.7, 10.3%, respectively; in a comparison of 2 H 2 O and DXA, the 95% limits of agreement for %BF in infants aged 0.5 mo were 24.3, 4.8%BF (32). Results from this study are also comparable with those from studies of adults and children that compared the BOD POD Body Composition Tracking System with hydrostatic weighing (95% limits of agreement from various studies ranged from 4.0, 3.4 to 11.0, 5.4) and with DXA (95% limits of agreement ranged from 2.6, 7.8 to 11.9, 4.1; reviewed in 11). Individual differences between %BF obtained by using ADP and that obtained by using 2 H 2 O dilution may be attributable to biological sources of error. For example, biological variations in FIGURE 2. Left: Linear regression of percentage body fat (%BF) results obtained with deuterium ( 2 H 2 O) dilution compared with those obtained with air-displacement plethysmography (ADP). The linear regression equation is %BF( 2 H 2 O) %BF(ADP) (R , SEE 3.26, P 0.001). The 3 open circles represent 3 excluded extreme outliers for the 2 H 2 O dilution values. The dotted line is Y X. Right: The corresponding Bland-Altman plot. The solid line represents the mean difference between methods (bias), and the dotted lines represent 2 SDs from the mean difference (95% limits of agreement).

7 PEDIATRIC AIR-DISPLACEMENT PLETHYSMOGRAPH 659 FFM hydration (used to calculate %BF from total-body water values obtained by 2 H 2 O dilution) and FFM density (used to calculate %BF from body mass and volume values measured by ADP) could have affected 2 H 2 O dilution and ADP results. Some individual differences might also be the result of factors affecting the 2 H 2 O dilution space during the 2-h postdose period, such as insensible water losses (eg, crying and sweating), changes in dilution space resulting from water intake (eg, breastfeeding or bottle-feeding), and the high water turnover rate in infants (33, 34). Technical errors in isotope analysis were considered small because of the high measurement precision of the mass spectrometer used in this study. There was no systematic difference in agreement between ADP and 2 H 2 O dilution results across a wide range of body mass and fatness. This represents a significant improvement over DXA. When DXA was used to assess body composition in animal models in the pediatric ranges of body mass and fatness, several studies found that the magnitude of bias in estimates of body fatness (in grams or percentage of weight) was related to body size, fat content, or both (35 39). The scanning program and software solutions used for scan analysis can also influence the results (35 37). In several studies, calibration of DXA to the laboratory standard of carcass analysis (ie, using conversion formulas or correction factors for DXA) was applied to improve the accuracy of DXA measurements (36, 39 41), but this is not an option for most clinicians. The ADP system has several advantages over 2 H 2 O dilution. Although isotope dilution is considered one of the most reliable and accurate methods for estimating body composition in early infancy (30), it is not practical from a research or clinical perspective because of the difficulties in using it in infants (eg, administering the proper dose and collecting adequate plasma or urine samples) and because it is time-consuming and expensive and requires considerable expertise in sample analysis. In contrast, a complete ADP test can usually be completed within 10 min. The test procedure is noninvasive, it does not require the subjects to be sedated, and the test environment (eg, temperature maintained constant between 31 and 37 C and constant air circulation in the test chamber) is comfortable for the infants. In summary, ADP is a reliable and accurate method of measuring %BF in healthy full-term infants, and it has the potential for use in the infant population in both research and clinical settings. Future studies are needed to determine the biological and technical sources of individual variations between ADP and other reference methods and to further evaluate longitudinal changes in %BF in infants. We are grateful to C Chaparro, B Lu, and X Mu for technical assistance with data collection and to the subjects and their parents for participating in the study. WA Coward from the MRC Human Nutrition Center (Cambridge, United Kingdom) and JC Wells from the Institute of Child Health (London) provided expert advice regarding the use of deuterium dilution techniques in the determination of infant body composition. Coordination of specific training in the use of deuterium dilution techniques and sponsorship of expert consultants (WA Coward and JC Wells) were provided by the International Atomic Energy Agency (Department of Nuclear Sciences and Applications, Division of Human Health, Section of Nutritional and Health-Related Environmental Studies), Vienna. GM, MY, YL, LN-R, and KGD were responsible for the study design, data analysis, and writing of the manuscript. MY, AL, and HZ were responsible for data collection in China, and LN-R was responsible for data collection in California. AU helped with technical issues related to the PEA POD system and with the portions of the manuscript describing the PEA POD. WWW provided expert advice regarding the use of deuterium dilution techniques and was responsible for analyzing the blood samples. All authors approved the manuscript. GM, YL, AL, HZ, WWW, LN-R, and KGD have no affiliations with the institution sponsoring this study. MY and AU are employees of Life Measurement, Inc. REFERENCES 1. Ong KKL, Ahmed ML, Emmett PM, Preece MA, Dunger DB. Association between postnatal catch-up growth and obesity in childhood: prospective cohort study. BMJ 2000;320: Troiano RP, Flegal KM. Overweight children and adolescents: description, epidemiology, and demographics. Pediatrics 1998;101: National Center for Health Statistics Division of Data Services. Prevalence of overweight among children and adolescents: United States, National Health and Nutrition Examination Survey (NHANES) Internet: hestats/overwght99.htm (accessed 10 March 2003). 4. Siri WE. Body composition from fluid spaces and density: analysis of methods. In: Brozek J, Henschel A, eds. Techniques for measuring body composition. Washington, DC: National Academy Sciences, National Research Council, 1961: Brozek J, Grande F, Anderson JT, Keys A. Densitometric analysis of body composition: revision of some quantitative assumptions. Ann N Y Acad Sci 1961;110: Lohman TG. Assessment of body composition in children. Pediatr Exerc Sci 1989;1: Butte NF, Hopkinson JM, Wong WW, Smith EO, Ellis KJ. Body composition during the first 2 years of life: an updated reference. Pediatr Res 2000;47: 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: Fields DA, Goran MI. Body composition techniques and the fourcompartment model in children. J Appl Physiol 2000;89: 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 N Y Acad Sci 2000;904: Fields DA, Goran MI, McCrory MA. Body-composition assessment via air-displacement plethysmography in adults and children: a review. Am J Clin Nutr 2002;75: Nunez C, Kovera AJ, Pietrobelli A, et al. Body composition in children and adults by air displacement plethysmography. Eur J Clin Nutr 1999; 53: Weyers AM, Mazzetti SA, Love DM, Gomez AL, Kraemer WJ, Volek JS. Comparison of methods for assessing body composition changes during weight loss. Med Sci Sports Exerc 2002;34: Vescovi JD, Zimmerman SL, Miller WC, Hildebrandt L, Hammer RL, Fernhall B. Evaluation of the BOD POD for estimating percentage body fat in a heterogeneous group of adult humans. Eur J Appl Physiol 2001; 85: Yee AJ, Fuerst T, Salamone L, et al. Calibration and validation of an air-displacement plethysmography method for estimating percentage body fat in an elderly population: a comparison among compartmental models. Am J Clin Nutr 2001;74: Falkner F. An air displacement method of measuring body volume in babies. Ann N Y Acad Sci 1963;110: Taylor A, Aksoy Y, Scopes JW, demont G, Taylor BA. Development of an air displacement method for whole body volume measurement of infants. J Biomed Eng 1985;7: Urlando A, Dempster P, Aitkens S. A new air displacement plethysmograph for the measurement of body composition in infants. Pediatr Res 2003;53: Sainz RD, Urlando A. Evaluation of a new pediatric air-displacement plethysmograph for body-composition assessment by means of chemical analysis of bovine tissue phantoms. Am J Clin Nutr 2003;77: Yao M, Nommsen-Rivers L, Dewey KG, Urlando A. Preliminary evaluation of a new pediatric air displacement plethysmograph for body composition assessment in infants. Acta Diabetol 2003;40:S Wong WW, Lee LS, Klein PD. Deuterium and oxygen-18 measurements

8 660 MA ET AL on microliter samples of urine, plasma, saliva, and human milk. Am J Clin Nutr 1987;45: Wong WW, Clarke LL, Laurador M, Klein PD. A new zinc product for the reduction of water in physiological fluids to hydrogen gas for 2 H/ 1 H isotope ratio measurements. Eur J Clin Nutr 1992;46: Gonfiantini R. Report on advisory group meeting on stable isotope reference samples for geochemical and hydrological investigations. Vienna: International Atomic Energy Agency, Lohman TG. Advance in body composition assessment. Champaign, IL: Human Kinetics, Lapillonne A, Braillon P, Claris O, Chatelain PG, Delmas PD, Salle BL. Body composition in appropriate and in small for gestational age infants. Acta Paediatr 1997;86: Koo WWK. Body composition measurements during infancy. Ann N Y Acad Sci 2000;904: Fiorotto ML, Cochran WJ, Klish WJ. Fat-free mass and total body water of infants estimated from total body electrical conductivity measurements. Pediatr Res 1987;22: De Bruin NC, van Velthoven KAM, Stijnen T, Juttmann RE, Degenhart HJ, Visser HKA. Body fat and fat-free mass in infants: new and classic anthropometric indexes and prediction equations compared with totalday electrical conductivity. Am J Clin Nutr 1995;61: De Bruin NC, van Velthoven KAM, de Ridder M, et al. Standards for total body fat and fat-free mass in infants. Arch Dis Child 1996;74: Lapillonne A, Salle BL. Methods for measuring body composition in newborns a comparative analysis. J Pediatr Endocrinol Metab 1999; 12: Koo WWK, Walters JC, Hockman EM. Body composition in human infants at birth and postnatally. J Nutr 2000;130: Butte N, Heinz C, Hopkinson J, Wong W, Shypailo R, Ellis K. Fat mass in infants and toddlers: comparability of total body water, total body potassium, total body electrical conductivity, and dual-energy x-ray absorptiometry. J Pediatr Gastroenterol Nutr 1999;29: Davies PSW, Wells JCK. Calculation of total body water in infancy. Eur J Clin Nutr 1994;48: Salazar G, Infante C, Vio F. Deuterium equilibration time in infant s body water. Eur J Clin Nutr 1994;48: Mitchell AD, Rosebrough RW, Conway JM. Body composition analysis of chickens by dual energy x-ray absorptiometry. Poult Sci 1997;76: Mitchell AD, Scholz AM, Conway JM. Body composition analysis of small pigs by dual-energy x-ray absorptiometry. J Anim Sci 1998;76: Picaud JC, Rigo J, Nyamugabo K, Milet J, Senterre J. Evaluation of dual-energy X-ray absorptiometry for body composition assessment in piglets and term human neonates. Am J Clin Nutr 1996;63: Picaud JC, Nyamugabo K, Braillon P, et al. Dual-energy x-ray absorptiometry in small subjects: influence of dual-energy x-ray equipment on assessment of mineralization and body composition in newborn piglets. Pediatr Res 1999;46: Fusch C, Slotboom J, Fuehrer URS, et al. Neonatal body composition: dual-energy x-ray absorptiometry, magnetic resonance imaging, and three-dimensional chemical shift imaging versus chemical analysis in piglets. Pediatr Res 1999;46: Pintuaro SJ, Nagy TR, Duthie CM, Goran MI. Cross calibration of fat and lean measurements by dual-energy X-ray absorptiometry to pig carcass analysis in the pediatric body weight range. Am J Clin Nutr 1996;63: Mitchell AD, Scholz AM, Pursel VG, Evock-Clover CM. Composition analysis of pork carcasses by dual-energy x-ray absorptiometry. J Anim Sci 1998;76: