J EisenkoÈlbl 1, M Kartasurya 1 and K Widhalm 1 * Introduction

Transcription

1 (2001) 55, 423±429 ß 2001 Nature Publishing Group All rights reserved 0954±3007/01 $ Original Communication Underestimation of percentage fat mass measured by bioelectrical impedance analysis compared to dual energy X-ray absorptiometry method in obese children J EisenkoÈlbl 1, M Kartasurya 1 and K Widhalm 1 * 1 Department of Pediatrics, Division of Neonatology, Intensive Care and Congenitas Anomalies, University of Vienna, Vienna, Austria Objective: The aim of the study was to investigate whether there is a difference between body fat mass percentage measured by BIA and DXA method. Design: Transversal study, randomized. Setting: Lipid and Obesity Outpatient Clinic, Department of Pediatrics, University of Vienna, Austria. Subjects: Twenty-seven children and adolescents from the Lipid and Obesity Outpatient Clinic, Department of Pediatrics, University of Vienna, were included in the study (14 boys and 13 girls between 6 and 18 y; mean age 12.6 and 13.1 y). Methods: The body fat percentage was measured with BIA (bioelectrical impedance analyzer BIA 2000-M) and DXA (dual energy X-ray absorptiometry) methods on the same day. Results: The mean difference between the body fat mass percentage measured by BIA and DXA was 4.48 with a standard deviation of The results measured by BIA were almost always lower than that by DXA by about 12%. The lower and upper limit of the difference in 95% con dence interval was 5.64 and By paired t-test, these results were signi cantly different (P < 0.001). The correlation between the two measurements was The mean percentage of body fat mass measured by BIA was % and by DXA %. The differences were not changed by age and body fat percentage but they were by sex. Conclusions: The results of the study show that the body fat percentages measured by BIA and DXA method were signi cantly different. This is very important because BIA technique is a routine technique for clinical purposes. Adjustments to the formula used for calculating the total fat mass in obese children and adolescence are necessary. Underestimation of body fat percentage measured by bioelectrical impedance analysis compared to dual X-ray absorptiometry method in obese children is three times higher with boys than with girls. Sponsorship: Data Input GmbH, Hofheim, Germany. Descriptors: bioelectrical impedance analysis; dual energy X-ray absorptiometry; body mass index; children; obesity (2001) 55, 423±429 Introduction *Correspondence: K Widhalm, WaÈhringer GuÈrtel 18-20, A-1090 Vienna, Austria. Kurt.Widhalm@akh-wien.ac.at Guarantor: K Widhalm. Received 7 July 2000; revised 2 January 2001; accepted 10 January 2001 Bioelectrical impedance analysis (BIA) is a rapid, safe and simple technique for measuring body composition both in the eld and in clinical settings. The device is relatively small and portable, and can be carried in a suitcase. It is also much cheaper than other sophisticated methods such as dual energy X-ray analysis (DXA) or total body electrical conductivity (TOBEC). This method measures the impedance of the body tissues to get the total body water (TBW). From these values, FFM (fat-free mass) can be calculated based on the assumption that TBW is a xed part of FFM. Finally body fat mass is calculated from TBW and FFM. (Fischer & Lembcke, 1991) Knowledge of body composition is important to de ne and manage obesity and undernutrition. The goal of obesity management is primarily to reduce fat stores and to save protein tissue. Therefore a device to determine the fat content of the body is needed to evaluate the success of the treatment. There are no data available comparing results of measurements by means of BIA technique and more precise methods in obese children and adolescents. It has been shown that BIA technique underestimates the fat mass in extremely obese adult subjects (Deurenberg, 1996;

2 424 Pietrobelli et al, 1998); therefore this study is designed to test the validity of this technique in obese children and adolescents because it is necessary to identify groups of subjects in which BIA results may be misleading. DXA is acknowledged as the standard (Dezenberg et al, 1999) and most precise (Fusch et al, 1999) method to assess body fat mass. Therefore in this study it is used as the standard method comparison to BIA technique, although it can only be used in a hospital and requires the use of a very low dose of radiation (Pietrobelli et al, 1998). Subjects and methods Subjects were outpatients of the Obesity and Metabolic Disorder Clinic for children and adolescents in the Department of Pediatrics, University of Vienna. Twenty-seven children and adolescents (14 boys and 13 girls) between 6 and 18 y (mean age 12.6 and 13.1 y) with a mean body mass index (BMI) of (boys) and (girls) were included in the study. The children were examined clinically by an experienced pediatrition and were regarded as healthy (free of any infection, etc). A routine chemistry pro le (Cholesterol, LDL-C, HDL-C, triglycerides, homocysteine, lipoproteine(a), ApoA 1, ApoB, GOT, GPT, electrolytes, blood picture, bilirubin, urea) was done. All results showed normal or obesity-related values. The BIA measurement of the children and adolescents took place after an overnight fast (Rodriguez et al, 1999) in a supine position with relaxed arms and legs without any contact with other body parts. The electrodes were placed on the dominant side of the body (right side for right- Figure 1 Comparison of body fat percentage measured by BIA and DXA. Scatterplot: boys and girls.

3 handed and left side for the left-handed children). Four skin-electrodes were placed on the dorsal side of the hand and foot proximal the metacarpalphalangeal and the metatarsalphalangeal joint and also between the distal prominences of the radius and ulna and between the medial and lateral melleoli of the ankle, respectively (Lukaski, 1987). BIA 2000 M (manufactured by Data Input GmbH, Hofheim, Germany with multiple frequencies 1, 5, 50 and 100 khz) was used to measure the resistance (R) and the reactance (Xc) values. These values and the weight and height measurements (measured to the nearest 0.1 kg and 1 cm, respectively) of the children were used to calculate the body fat mass percentage by Nutri4 software program. The formula used in this program is an optimized version of the formula published by Kushner et al, (1986) with single frequency analysis at 50 khz. As the standard method we used the DEXA system of Hologic (QDR 4500 X-Ray Bone Densitometer). For calibration a Step-Phantom scan was performed every morning. This unit is not transportable. Afterwards the values of body fat mass percentage and also the other data needed, such as DXA results, age and sex were collected to calculate the following results by SPSS windows 8.0 program. To show the proportion of the two measurements visually, some graphs were created, especially Bland ± Altmann Plots (1986). Paired t-test, standard deviations, 95% con dence intervals, signi cances and correlation coef cients were used for the comparison. Results The mean body fat mass percentage measured by BIA was 35.22% 7.34 and by DXA 39.41% Therefore BIA results appeared to be generally lower than the DXA results by about 10.6%. The underestimation of body fat percentage by BIA compared to DXA is shown in Figure 1. The t-test comparison for a paired group showed that the correlation coef cient between body fat percentages measured by BIA and DXA was (P < 0.001). Dividing the subjects by sex, the mean difference between the two measurements of boys was 6.27% 2.64, and the difference in girls was only 1.94% 2.69, while the standard deviation was nearly the same. The girls had a slightly better correlation coef cient (0.923) between the BIA and DXA results than the boys (0.919). By paired t-test, these results were signi cantly different. Figures 2 and 3 show the comparison in boys and girls, respectively. These differences can also be seen in Tables 1 and 2. Discussion It has been shown that a body fat measurement in obese adolescents exhibit markedly different results using BIA or DXA. This is especially true for boys. The difference between BIA and DXA results in measuring body fat percentage in girls was smaller than that of boys (Figures 425 Figure 2 The difference in body fat percentage measured by DXA and BIA.

4 426 Table 1 The difference between boys and girls in BIA and DXA results Boys Girls Variables Mean s.d. Mean s.d. Age (y) Body fat percentage by BIA 32.21* * 6.69 Body fat percentage by DXA Body mass index (BMI) (kg=m 2 ) Difference (DXA BIA) percentage fat ** ** Correlation coef cient between BIA and DXA results *P-value ˆ 0.023; **P-value < 0.001; s.d. ˆ standard deviation. Figure 3 The difference of body fat percentages measured by BIA and DXA method in boys.

5 Table 2 The difference between BIA and DXA results in all subjects Paired differences % con dence interval of the difference Mean s.d. s.e.m. Lower Upper t d.f. Signi cance (two-tailed) Percentage fat by BIA measurement percentage fat by DXA measurement < Figure 4 The difference of body fat percentages measured by BIA and DXA method in girls.

6 428 3 and 4). The mean difference between the two measurements in girls and boys was signi cant (P ˆ 0.023; Table 1). The correlation coef cient between the two measurements in girls was also slightly higher than in boys. In this study, the girls had a relatively higher body fat percentage than the boys, but the difference was not signi cant (Table 1). Therefore the higher correlation coef cient in girls was not caused by their higher body fat percentage. Figure 2 also shows that the higher body fat percentage did not result in a greater difference between these two measurements. One reason might be that this study was performed on subjects in an obese state. However, the children who had higher body fat percentage did not have a larger difference between the two measurements. One study with 10 obese children found a trend to underestimate BIA in relation to DXA measurements (Okasora et al, 1999). In healthy adults another study found a difference of 3.3% between BIA and DEXA, but a gender difference could not be detected (Levenhagen et al, 1999). Another study with 196 adults detected a difference between BIA and DXA of 6.8% in males and 8.8% in females. The bias observed in obese subjects was similar to that observed in non-obese subjects (Wattanapenpaiboon et al, 1998). A study with 40 black children showed no signi cant difference between males and females. Six of the 21 female subjects were obese with polycystic ovary syndrome (PCOS). In the case of PCOS females who were obese, body weight was an independent signi cant variable contributing to the predictability of FFM (Lewy et al, 1999). Another study with 99 subjects (5 ± 22 y) showed a difference of BIA and DXA measurements of 1.2% in males and 4.7% in females; so this group of subjects with only 15% obese found contrary results; the difference was four times higher with females than with males. This study concludes that it is dif cult to ensure that similar groups of subjects can be selected for the same degree of fatness when different measurement methods are used to classify the population (Ellis, 1996). The different results between BIA and DXA in this current study could be caused by a false assumption in predicting lean body mass from TBW. The assumption that 73.2% lean body mass consists of TBW (Data Input GmbH, 1998) is possibly not correct. In healthy persons the water content of lean body mass (LBM) is constant after the age of 20 (Schoeller, 1989). In the obese, however, the amount of water might be more than 73.2%. This condition could cause an overestimation of the lean body mass which in turn could underestimate the body fat mass (Deurenberg, 1996), which is counted by total body weight minus lean body mass. The other false assumption (Deurenberg, 1996) is, that the body water distribution, the relative ratio of extracellular water to TBW, is higher in obese children. This condition results in a larger difference in more obese children (Deurenberg, 1996). This was not the case in this study, as the higher body fat percentage did not increase the difference between the two measurements. In addition, the use of multifrequency BIA in this study (BIA 2000-M) accounts for this condition. The extracellular water was best predicted by resistance measured at 5 khz whereas TBW was best predicted at 100 khz (Data Input GmbH, 1998; Segal et al, 1991). BIA results were not as precise as DXA in measuring body fat mass in obese children, but the results were superior than the use of BMI in predicting body fat. The correlation coef cient between BMI values and body mass percentage by DXA method was only 0.605, compared to BIA, which was Even though the calculation of body fat mass percentage was done from the BMI (by the formula by Garrow and Webster; Heber et al, 1996), the correlation coef cient was only slightly improved to Conclusion BIA measurements of body fat percentage of obese children and adolescents were 10.6% lower than DXA results. The difference was three times higher in boys than in girls, but was not in uenced by a higher fat percentage. The differences could be caused by false assumptions made in predicting body fat by BIA. Although BIA is not as precise as DXA in measuring body fat percentage in obese children, BIA was superior than BMI and can still be used for measuring body fat with a standard error of about 10%. It is necessary to identify groups of subjects in which BIA results may be misleading. A larger study has to be done to produce a more precise formula for measuring body fat mass by BIA which ts for obese children and adolescents and classi es the difference between girls and boys. References Bland JM & Altman DG (1986): Statistical methods for assessing agreement between two methods of clinical measurement. Lancet i, 307 ± 310. Data Input GmbH (1998): BIA-Kompendium. Biophysikalische Grundlagen, KoÈrperzusammensetzung, ErnaÈhrungszustand, Klinische Anwendungsbereiche, Fallbeispiele. Hofheim, Germany. Deurenberg P (1996): Limitations of the bioelectrical impedance method for the assessment of body fat in severe obesity. Am. J. Clin. Nutr. 64(Suppl), 449S ± 452S. Dezenberg CV, Nagy TR, Gower BA, Johnson R Goran MI (1999): Predicting body composition from anthropometry in pre-adolescent children. Int. J. Obes. Relat. Metab. Disord. 23, 253 ± 259. Ellis KJ (1996): Measuring body fatness in children and young adults: comparison of bioelectric impedance analysis, total body electrical conductivity, and dual-energy X-ray absorptiometry. Int. J. Obes. Relat. Metab. Disord. 20, 866 ± 873. Fischer H & Lembcke B (1991): Die Anwendung der Bioelektrischen Impedananalyse (BIA) zur Beurteilung der KoÈrperzusammensetzung und des ErnaÈhrungszustandes. Innere Medizin Aktuell. 18(1=1991), S13 ± S17. Fusch C et al (1999): Neonatal body composition: dual energy X-ray absorptiometry, magnetic resonance imaging, and three-dimensional chemical shift imaging versus chemical analysis in piglets. Pediatr. Res. 46, 465 ± 473. Heber D, Ingles S, Ashley JM, Maxwell MH, Lyons RF & Elashoff RM (1996): Clinical detection of sarcopenic obesity by bioelectrical impedance analysis. Am. J. Clin. Nutr. 64, 472S ± 477S.

7 Kushner RF & Schoeller DA (1986): Estimation of total body water by bioelectrical impedance analysis. Am. J. Clin. Nutr. 44, 417 ± 424. Levenhagen DK, Borel MJ, Welch DC, Piasecki JH, Chen KY & Flakoll PJ (1999): A comparison of air displacement plethysmography with three other techniques to determine body fat in healthy adults. J. Parent. Enteral. Nutr. 23, 293 ± 299. Lewy VD, Danadian K & Arslanian S (1999): Determination of body composition in African-American children: validation of bioelectrical impedance with dual energy X-ray absorptiometry. J. Pediatr. Endocrinol. Metab. 12, 443 ± 448. Lukaski HC (1987): Methods for the assessment of human body composition: traditional and new. Am. J. Clin. Nutr. 46, 537 ± 556. Okasora K, Takaya R, Tokuda M, Fukunaga Y, Oguni T, Tanaka H, Konishi K & Tamani H (1999): Comparison of bioelectrical impedance analysis and dual energy X-ray absorptiometry for assessment of body composition in children. Pediatr. Int. 41, 121 ± 125. Pietrobelli A, Wang ZM & Heyms ed SB (1998): Techniques used in measuring human body composition. Curr. Opin. Clin. Nutr. Metab. Care. 1, 439 ± 448. Rodriguez G, Moreno LA, Sarria A, Fleta J & Bueno M (1999): Diurnal variation in the assessment of body composition using bioelectrical impedance in children. Eur. J. Clin. Nutr. 53, 244. Schoeller DA (1989): Changes in total-body water with age. Am. J. Clin. Nutr. 50, Segal KR, Burastero S, Chun A, Coronel P, Pierson RN & Wang J (1991): Estimation of extracellular and total body water by multiple-frequency bioelectrical impedance measurement. Am. J. Clin. Nutr. 54, 26 ± 29. Wattanapenpaiboon N, Lukito W, Strauss BJG, Hsu-Hage BH-H, Wahlqvist ML & Stroud DP (1998): Agreement of skinfold measurement and bioelectrical impedance analysis (BIA) methods with dual energy X-ray absorptiometry (DEXA) in estimating total body fat in Anglo-Celtic Australians. Int. J. Obes. Relat. Metab. Disord. 22, 854 ±