Does the BMI reflect body fat in obese children and adolescents? A study using the TOBEC method

Article metrics

Abstract

OBJECTIVE: Due to the fact that obesity is defined as excess of body fat mass, we tested the hypothesis whether the body mass index (BMI) can be used as a valid measure for the detection of the degree of obesity in individual obese children and adolescents.

METHODS: A total of 204 obese children and adolescents (105 boys, 99 girls) aged 6–17 y, using total body electrical conductivity (TOBEC) for fat measurement, were included into this study.

 A multiple regression analysis was performed with percentage body fat (PBF) as dependent variable and BMI, age and sex as independent variables. First- and second-order interaction terms were also included. Since all interaction terms showed a significant influence on PBF, regression analysis was performed separately for boys and girls, dividing each group into two age subgroups (subjects younger than 10 y, and subjects 10 y or older).

RESULTS: BMI and PBF were observed to be positively correlated (overall: r=0,65; P=0.0001; boys r=0.63 and girls: r=0.68). Through a multiple regression analysis 57% of the variance of PBF could be explained by the independent variables. In boys younger than 10 y 73% and in girls younger than 10 y 63% of the variance of PBF was explained by the BMI. In subjects 10 y or older the association was poor (boys: 27%; girls: 38%). It should be emphasized that there is a wide range in the relationship between PBF and BMI in the obese subjects.

CONCLUSION: From these results we conclude that BMI might be a useful parameter for epidemiological studies: however in the individual pediatric patient, especially from 10 y onwards, it gives only a limited insight to the degree of obesity based on the definition.

Introduction

There are many facts that make it necessary to have a thorough look at the subject of childhood obesity. The dramatic rise in the prevalence of obesity in European countries and the USA has become a major health concern.1,2 Childhood obesity is linked to obesity in adulthood and furthermore it is associated with increased mortality, coronary heart disease, hypertension, dyslipidemia and diabetes mellitus.3,4,5,6,7,8 Thus, the importance of achieving a reliable and accurate estimate of body fatness is essential not only for the prevention but also for the treatment of overweight in children and adolescents.

Until now there has been no consensus with regard to the definition and classification of obesity in childhood and adolescence.9 So far the most widely used reference criteria have been based on the US National Center for Health Statistics (NCHS).

Obesity in the pediatric age group has been defined as an increase of total body fat out of proportion to other tissues.10

The most common estimate of body composition in adult populations is the body mass index (BMI: body weight (kg)/height2 (m2). It is a measure of weight with a low correlation to height.

In the past few years there have been several attempts to introduce the BMI into the pediatric age group as well, despite the disadvantage of its age- and sex-dependency, because it could represent a continuum of indices from childhood to adult life.9 The BMI is considered to be a simple parameter to define obesity.11,12 However there is a lack of data showing that the BMI is able to define obesity, based on its definition, for the whole pediatric age group.

Objective

Because obesity is defined as excess of body fat mass, we tested the hypothesis whether the BMI can be used as a valid measure for the detection of the degree of obesity in individual obese children and adolescents.

Methods

Subjects and protocol

To test this hypothesis 204 obese (above the 95th percentile, according to the percentiles of Rolland-Cachera et al13) but otherwise healthy children and adolescents aged 6–17 y were included in this study. All subjects were white Austrians and came to the Obesity Outpatient Clinic at the Department of Pediatrics University of Vienna because of overweight problems. Most of them came of their own volition, some were referred by school doctors, pediatricians or family doctors. Anthropometric and body composition measurements were performed on the same day.

Anthropometrical and body composition measurements

Anthropometrical and body composition measurements were performed by two experienced examiners. Height was measured with a fixed stadiometer (Holtain Ltd, UK), and weight was measured without shoes and heavy clothing on a standard clinical scale (Seca, Hamburg) to the nearest of 0.1 kg and 0.1 cm, respectively.

Body composition measurement by means of total body electrical conductivity (TOBEC) is a noninvasive, well-established method for the assessment of body composition without any radiation exposure.14 The TOBEC instrument (EM-Scan, Inc., Springfield, IL, model HA-2) consists of a chamber surrounded by a solenoidal coil driven by a 2.5 MHz radio frequency generator to produce a uniform electromagnetic field within the open-air core. When a conductor (in this case a patient) is placed within the core, it will interact with the external electromagnetic field, resulting in an energy loss for the outer coil which can be measured and is called the TOBEC number (E). According to calibration equations for different age groups, which have been extensively validated by Van Loan et al, the following two equations were used to derive an estimate for body fat mass in our study group:

Children(5−18y):fat=weight−(0.2772

×(E×height)0.5+1.232)

Young adults(19−40y):fat=weight−(0.2884

×(E×height)0.5−0.3951)

Two TOBEC scans of the whole body were performed and averaged for each subject. Fat mass was calculated as: Fat mass (kg)=body weight (kg)−fat-free mass (kg). Percentage body fat was calculated as: PBF=(fat mass/body weight)×100. The s.e.e. for the prediction of FFM as measured by TOBEC averages about 2.2 kg (range 1.5–3.0 kg), which translates to 4% uncertainty for PBF. 15,16,17,18

Statistical analysis

Percentiles were calculated for the description of the sample. Student's t-test was applied for comparison of age, weight, height, PBF and BMI between girls and boys. Pearson correlation coefficients were used to assess univariate association among PBF, BMI and age. Multiple regression analysis was performed with PBF as dependent variable and BMI, age and sex as independent variables. First- and second-order interaction terms were also included. Furthermore, BMI2 was added to the model. Since all interaction terms showed significant influence on PBF, regression analysis for PBF on BMI was performed separately for girls and boys, where each sex group was divided into two age groups (age <10 y and age ≥10 y, respectively). In addition BMI2 was included in the model for girls <10 y of age to test for nonlinearity.

The SAS® statistical software system was used for computation. A P-value <0.05 was considered to indicate statistical significance.

Results

Characteristics of study sample

The mean value and standard deviation of age, weight, height, PBF and BMI are shown in Table 1. Boys and girls younger than 10 y differed significantly in weight and BMI.

Table 1 Characteristics of study sample; values are given as mean±s.d.

Correlation analysis

BMI and PBF were observed to be positively correlated (overall: r=0.65, P<0.0001) (Figure 1) both in boys (r=0.63, P<0.0001) and girls (r=0.68, P<0.0001).19 BMI was also correlated with age (overall: r=0.43, P<0.0001; boys; r=0.33, P<0.0006; girls: r=0.54, P<0.0001). However the association between PBF and AGE was not significant (overall: r=0.10, P=0.154; boys: r=0.01, P=0.894; girls: r=0.19, P=0.061).

Figure 1
figure1

Correlation between BMI and PBF over the whole sample of 204 obese children and adolescents.

Terms of interaction

Multiple regression analysis with PBF as dependent variable and BMI, age and sex as independent variables and the first-and second-order interaction terms reached an r2 of 0.57. This means that approximately 57% of the variance of PBF can be explained by the independent variables. The main effects BMI (P<0.0001) and sex (P<0.0002) showed a significant influence on PBF. Age (P=0.053) was only borderline significant.

The P-value of the first order interaction BMI×sex (P=0.022) and BMI×age (P=0.014) indicates that the association between fat and BMI has a different pattern for boys and girls and different levels of ages. The interaction between BMI×age×sex (P=0.0145) also reached significance. Inclusion of the quadratic term BMI2 (P<0.001) in the model improved r2 only slightly to 0.59 with the interaction still being highly significant. Therefore subgroup analyses were performed.

Subgroup analyses

The results of the four regression analyses for PBF on BMI are presented in Table 2. The results reveal a strong association between PBF and BMI (P<0.0001) for boys younger than 10 y, with BMI explaining 73% of the variance of PBF. However, for boys 10 y or older, the association between BMI and PBF, although still significant, is observed to be weaker, with BMI explaining only 27% of the variance of PBF (Figures 2 and 3).

Table 2 Regression analysis (model 1)
Figure 2
figure2

Subgroup analysis: correlation between BMI and PBF in boys younger than 10 y.

Figure 3
figure3

Subgroup analysis: correlation between BMI and PBF in boys aged 10 y or older.

The results for girls are very similar to those of boys. For girls younger than 10 y of age the association between PBF and BMI (P<0.0001) is rather strong (r2=0.63). For girls aged 10 or above, an r2 of only 0.38 is reached (P<0.0001) (Figures 4 and 5). Since the scatter plot for girls younger than 10 y of age suggests a curvilinear association between PBF and BMI, the quadratic term BMI2 was included in the model. The results are also given in Table 2. Addition of BMI2 only improved the r2 from 0.63 to 0.71 in the group of girls younger than 10 y. In the other groups it did not show any improvement.

Figure 4
figure4

Subgroup analysis: correlation between BMI and PBF in girls younger than 10 y.

Figure 5
figure5

Subgroup analysis: correlation between BMI and PBF in girls 10 or older.

Discussion

In the present study carried out in a remarkably large sample of obese children and adolescents it can be demonstrated that the BMI of boys and girls aged younger than 10 y explains 73 and 63% of the variance of PBF. A rather poor explanation of variance has been observed in children aged older than 10 y (boys 27%, and girls 38%). These data indicate that the BMI is not a reliable measure for the determination of body fat in the individual obese pediatric patient.

A further goal of this study was to investigate whether age and sex contribute to the prediction of fatness. By means of correlation (BMI and sex, BMI and age), it was demonstrated that the association between PBF and BMI is different for boys and girls and changes with age. There is an increase of PBF with increasing age, and boys have a lower fat content than girls.

In the pediatric age group the BMI seems to be an inaccurate measure for the estimation of weight, without considering body composition. The BMI itself does not contribute to the definition of overweight or obesity, but through defining relative overweight the use of BMI might form a continuum of index from childhood to adult life for epidemiological purposes.

The predictive value of childhood BMI for overweight at age 35 y, defined as BMI>28 for men, and >26 for women, has been reported to be excellent at age 18, good at age 13, and moderate at ages below 13 y.6

For the German-speaking countries age-specific percentiles for boys and girls aged 10 y or older have been presented by Von Heye Coners et al. As cut-off point for screening for adiposity he recommended the 85th percentile.20 Different cut-offs have been recently presented. Another German group recommended the 75th percentile because it was the most appropriate cut-off value to screen for the 15% most obese patients by PBF.21 Thus, even on a national basis there is still no consensus for the appropriate cut-off points.

For that reason it has been emphasized that additional research is needed to identify BMI cut-off points throughout childhood and adolescence. These efforts could lead to a greater sensitivity, specifity and predictive value for adult obesity and the associated health problems. Further, anthropometrical, physiologic and psychological factors have to be included to augment the use of BMI for the identification of the most appropriate children for treatment.22

The BMI seems to be a convenient and easy method, as it is easily calculated from weight and height, which are routinely measured in the clinical setting. Its reliability refers to the reproducibility of measurement, although it is considered to be a highly specific, but relatively insensitive test for fatness.23

The correlation of BMI to PBF has been described differently in other studies.24 In this study in obese young subjects the overall correlation of BMI and PBF was significant (for all, r=0.65; girls, r=0.68, boys, r=0.63), which is almost identical to the results presented by Pietrobelli et al in normal-weight children and adolescents.25 In our study the degree of obesity was far beyond the 95th percentile of BMI. Thus a possible age- or maturation-dependent change of body composition could not influence the description of the amount of obesity. Therefore we divided the sample into children attending primary school and children/adolescents attending high-school, according to age.

It is an interesting phenomenon that both healthy and obese children and adolescents of the same age, sex and BMI showed a wide range of different individual values of PBF. This fact clearly demonstrates that the reliability to the BMI–fatness association is restricted.

Body fat can be determined from the difference between body weight and fat-free mass (FFM). There are various methods of estimating FFM, including anthropometry and hydrodensiometry, 40K spectrometry and TOBEC. The measurement of body composition using TOBEC is quick, allowing for the study of large numbers of subjects. The instrument (HA-2) has been commercially available since 1989. At present TOBEC is one of the most reliable methods to estimate body composition from infancy through childhood and adolescence to adulthood, but is relatively expensive and difficult to move and therefore not suitable for field studies. However its good reproducibility, precision and accuracy justify the use of TOBEC as a reference method.26,27,28,29,30

TOBEC has been demonstrated to give very similar results to the dual-energy X-ray absorptiometry (DXA), which has been used in Pietrobelli's study. In a study by Ellis, different body composition measurement methods, such as bioelectrical impedance analysis (BIA), bioelectrical impedance spectroscopy (BIS) and dual-energy X-ray absorptiometry (DXA) and TOBEC have been compared: a high correlation between the TOBEC method (r=0.918 for BIA, r=0.95 for DXA), and one to another has been shown.31 Therefore we suggest that comparison between different methods of body composition measurement is certainly possible, but caution is needed when integrating different measurements into one study.

A rather interesting result is the fact that the multiple regression analysis in this study showed a unsatisfactory correlation to PBF for children and adolescents aged 10 y or older. PBF was explained by the BMI with a variance of only 27 or 38% males and females, respectively. This emphasizes the strongly variable association of body fat and BMI until the age of 10.

The precision of direct fat measurements for the quantification of body fat and therefore the correct classification of the degree of obesity in an individual patient is a very important information for the evaluation of risk factors such as hypertension, cholesterol, etc, and the detection of obesity in the individual patient.

Conclusion

The BMI seems to be a useful parameter for epidemiological studies using sex- and age-specific BMI cut-offs for the approximate classification on the degree of obesity. As the BMI does not allow a quantification of body fat and the reliability of the fatness association in the individual pediatric patient is restricted, we do not recommend BMI for the monitoring of treatment in obese children.

References

  1. 1

    Troiano RP, Flegal KM, Kuczmarski RJ, Campbell SM, Johnson CL . Overweight prevalence and trends for children and adolescents Arch Pediatr Adolesc Med 1995 149: 1085–1091.

  2. 2

    Barth N, Ziegler A, Himmelmann GW, Coners H, Wabitsch M, Henninghausen K, Mayer H, Remschmidt H, Schärfer H, Hebe-brand J . Significant weight gains in clinical sample of obese children and adolescents between 1985 and 1995 Int J Obes Relat Metab Disord 1997 21: 122–126.

  3. 3

    DiPietro L, Mossberg H-O, Stunkard AJ . A 40-year history of overweight children in Stockhom: life-time overweight, morbidity, and mortality Int J Obes Relat Metab Disord 1994 18: 585–590.

  4. 4

    Kaufer Christoffele K, Ariza A . Commentary: The epidemiology of overweight in children: relevance for clinical care Pediatrics 1998 101: 103–105.

  5. 5

    Vanltallie TB . Predicting obesity in children Nutr Rev 1998 56: 154–155.

  6. 6

    Guo SS, Roche AF, Chumela WC, Gardener JD, Siervogel RM . The predictive value of childhood body mass index values for overweight at age 35 y Am J Clin Nutr 1994 59: 810–819.

  7. 7

    Troiano RP, Frongillo EA Jr, Sobal J, Levitsky DA . The relationship between body weight and mortality: a quantitative analysis of combined information from existing studies Int J Obes Relat Metab Discord 1996 20: 63–75.

  8. 8

    Power C, Lake JK, Cole TJ . Measurement and long-term health risks of child and adolescent fatness Int J Obes Relat Metab Disord 1997 21: 507–526.

  9. 9

    Poskitt EME . Defining childhood obesity: the relative body mass index (BMI) Acta pediatr 1995 84: 961–963.

  10. 10

    Wolff OH, Lloyd JK . Obesity Medicine 1974 27: 1583.

  11. 11

    Marshall JD, Hazlett CB, Spady DW, Conger PR, Quinney HA . Validity of convenient indicators of obesity Hum Biol 1991 63: 137–153.

  12. 12

    Hannan WJ, Wrate RM, Cowen SJ, Freeman CP . Body mass index as an estimate of body fat Int J Eat Disord 1995 18: 91–97.

  13. 13

    Rolland-Cachera M-F, Cole TJ, Sempé M, Tichet J, Rossignol C, Charrand A . Body mass index variations: centiles from birth to 87 y Eur J Clin Nutr 1991 45: 13–21.

  14. 14

    Puig M . Body composition and growth. In: Walker and Walkins (eds) Nutrition in Pediatrics, 2nd edn Becker: Hamilton 1997.

  15. 15

    VanLoan MD, Sagal KR, Bracco EF, Mayclin P, Vanltallie TB . TOBEC methodology for body composition assessment: a cross validation study Am J Clin Nutr 1987 46: 9–12.

  16. 16

    VanLoan MD, Mayclin P . A new TOBEC instrument and procedure for the assessment of body composition: use of Fourier coefficients to predict lean body mass and total body water Am J Clin Nutr 1987 45: 131–137.

  17. 17

    Keim NL, Mayclin P, Taylor SJ, Brown DL . Total body electrical conductivity method for estimating body composition: validation by direct carcass analysis of pigs Am J Clin Nutr 1988 47: 180–185.

  18. 18

    Horswill CA, Geesemann R, Boileau RA, Williams BT, Layman DK, Massey BH . Total body electrical conductivity (TOBEC): relationship to estimates of muscle mass, fat-free weight, and lean body mass Am J Clin Nutr 1989 49: 593–598.

  19. 19

    Widhalm K, Schönegger K . BMI: does it really reflect body fat mass? Pediatrics 1999 134: 522.

  20. 20

    Von Heye Coners, Himmelmann GW, Hebebrand J, Heseker H, Remschmidt H, Schärfer H . Perzentilenkurven für Body Mass Index zur Gewichtsbeurteilung bei Kindem and Jugendlichen ab einem Alter von 10 Jahren Der Kinderarzt 1996 8: 1002–1007.

  21. 21

    Schärfer F, Georgi M, Wühl E, Schärfer K . Body mass index and percentage fat mass in healthy German schoolchildren and adolescents Int J Obes Relat Metab Disord 1998 22: 461–469.

  22. 22

    Dietz WH . Use of the body mass index (BMI) as a measure of overweight in children and adolescents Pediatrics 1998 132: 1991–193.

  23. 23

    Dwyer T, Blizzard CL . Defining obesity in children by biological endpoint rather than population distribution Int J Obes Relat Metab Disord 1996 20: 472–480.

  24. 24

    Boot AM, Bouquet J, de Ridder MAJ, Krenning EP, de Muinck Keizer-Schrama S . Determinants of body composition measured by dual-energy X-ray absorptiometry in Dutch children and adolescents Am J Clin Nutr 1997 66: 232–238.

  25. 25

    Pietrobelli A, Faith MS, Allison DB, Gallagher D, Chiumello G, Heymsfield SB . Body mass index as a measure of adiposity among children and adolescents Pediatrics 1998 132: 204–210.

  26. 26

    Van Loan MD, Koehler LS . Utilization of total electrical conductivity for the assessment of body composition in middle aged and elderly individuals Am J Clin Nutr 1990 51: 548–552.

  27. 27

    Van Loan MD . Assessment of fat free mass in teen-agers: use of TOBEC methodology Am J Clin Nutr 1990 53: 586–590.

  28. 28

    Hergenroeder AC, Wong WW, Fiorotto ML, Smith EO, Kush WJ . Total body water and fat free mass in ballet dancers: comparing isotope dilution and TOBEC Med Sci Sports Exerc 1991 23: 534–541.

  29. 29

    Ellis JK, Yasumura S, Morgan WD . Changes in fat mass in postpartum women: a comparison of body composition models Int J Obes Relat Metab Disord 1997 21: 874–880.

  30. 30

    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 equation compared with total-body electrical conductivity Am J Clin Nutr 1995 61: 1195–1205.

  31. 31

    Ellis JK . 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 1996 20: 866–873.

Download references

Author information

Correspondence to K Widhalm.

Rights and permissions

Reprints and Permissions

About this article

Keywords

  • total body electrical conductivity (TOBEC)
  • body mass index (BMI)
  • obese children and adolescents
  • body composition

Further reading