¹Universidade Federal de São Paulo, Escola Paulista de Medicina, São Paulo, Brazil

²Department of Preventive Medicine, Universidade Federal de São Paulo, Escola Paulista de Medicina; São Paulo, Brazil

Correspondence to: M A Sampei, Postgraduate in Nutrition, Universidade Federal de São Paulo, Escola Paulista de Medicina, Rua Marselhesa 630, Vila Clementino, CEP: 04020-060, São Paulo, SP, Brazil. E-mail: misampei@osite.com.br

OBJECTIVE: The aim of this study was to assess the body mass index (BMI) and its relationship with other methods of body fat evaluation in pre- and post-menarcheal, Japanese and Caucasian female adolescents, using two different cut-off points for obesity: 28% and 30%.

DESIGN: A cross-sectional study with incomplete sampling, using the subject as the evaluation unit.

SUBJECTS: A total of 436 Japanese and Caucasian female adolescents in two age groups: 10-11 (pre-menarcheal adolescent); and 16-17 (post-menarcheal adolescents).

METHODS: For the BMI the cut-off point for thinness was set at the 5th percentile of the BMI distribution of the NCHS reference population and the cut-off point for overweight and obesity was set at the 85th percentile. Body composition was assessed using foot-to-foot bioelectrical impedance analysis (BIA), near-infrared interactance (NIR) and Slaughter skinfold equations (SKI). The statistical comparison of the methods was performed using the kappa agreement test and the McNemar disagreement test.

RESULTS: In the 10- and 11-y-old girls, the BMI was considerably and significantly correlated with the other methods. The major agreements were: in Japanese adolescents BMI´NIR=82.3% (cut-off point of 28%), BMI´BIA=85.7% (cut-off point of 30%); in Caucasian adolescents BMI´NIR=80.7% (cut-off point of 28%), BMI´BIA=87.4% (cut-off point of 30%). The disagreement above the diagonal between BMI´NIR was higher within the two groups for both the cut-off points, revealing that the girls identified as obese by the BMI were considered eutrophic by NIR. In the 16- and 17-y-old adolescents, the BMI demonstrated low or no correlation with the other methods. Furthermore, it presented disagreements below the diagonal, revealing that the BMI identified fewer obese subjects than the other methods.

CONCLUSION: Among the 10- and 11-y-olds, the BMI presented a good correlation with the other methods, independent of ethnicity. The BMI can therefore be used in place of these methods, although it may underestimate obesity. Among the 16- and 17-y-olds, the BMI presented low or no agreement with the other methods, suggesting that it is probably not a suitable index for this age-group in studies focusing on the identification of obesity. In such cases the choice of one of the other methods, depending on availability, cost or technical experience, may represent a better approach.

International Journal of Obesity (2001) 25, 400-408

ethnic groups; menarche; foot-to-foot bioelectrical impedance; near-infrared interactance; cutaneous skinfold; body mass index

Introduction

Many difficulties are encountered in evaluating the anthropometric variables and body composition of adolescents. This is due to socio-economic differences, the wide variability in the start time of the growth spurt and the variation in maturation rate and growth patterns among different ethnic groups within the population.

The effect of environmental conditions on development, including growth, maturation and the fulfilment of genetic potential, can be identified through the study of the variations found among different ethnic groups in the same population. In this respect, population studies of body composition can be useful, although the impracticality of using precise and sophisticated laboratory methods on large samples presents a serious disadvantage.

An alternative to these methods is the body mass index (BMI), which is frequently used for evaluations in nutritional studies of individuals, especially those focusing on obesity. Despite its wide use, few studies have analyzed the accuracy of the BMI in evaluating body fat in children and adolescents.¹ Furthermore, although various studies have shown a weak relationship between BMI and percentage of fat,^1,2 there is strong evidence that ethnicity can interfere in this relationship.^3,4,5,6

Apart from BMI, there are other methods, such as foot-to-foot bioimpedance (BIA)⁷ and near-infrared interactance (NIR),⁸ that use portable equipment for body composition analysis. However, the substitution of BMI with these methods is something which requires further investigation.

Although comparative analysis of the different anthropometric and body composition evaluation methods has been performed in the white population, few studies have been performed in other ethnic groups.^9,10 It has been suggested that taking ethnicity into consideration can aid in the planning of health measures, and can sometimes give rise to important data on health aetiology, diagnosis and treatment.¹¹ The main aim of the present study was, therefore, to assess the BMI and its relationship to other methods of body fat evaluation (foot-to-foot bioelectrical impedance, BIA; near-infrared interactance, NIR; and skinfold, SKI) in 436 Japanese and Caucasian adolescents aged 10-11 (pre-menarcheal) and 16-17 (post-menarcheal) from private schools in São Paulo, Brazil. The analysis of the relationships between BMI and the other methods of body fat evaluation took into account ethnicity, the presence or absence of menarche and the effect of taking two different cut-off points for the degree of obesity.

Methods and sample

Sampling and design of the study

A cross-sectional methodological design with incomplete sampling was used with the subject as the evaluation unit. The study was approved by the ethics committee of the Universidade Federal de São Paulo, Escola Paulista de Medicina and consent for the participation of the adolescents was given by the directors of the included schools.

The anthropometric and body composition measures were collected from all 10 and 11-y-old pre-menarcheal adolescents and 16 and 17-y-old post-menarcheal adolescents, independent of ethnicity, who attended 15 private schools in the city of São Paulo, Brazil. On those occasions when time constraints made it impossible to measure all the girls in every class, such as in the period preceding the school holidays, all the Japanese adolescents were measured, together with an equivalent number of girls of other ethnicities chosen by allocating each one a number and then drawing the numbers from a bag. In these cases an additional three or four Caucasian girls were also measured to act as substitutes.

The following selection criteria were adopted for inclusion of adolescents in the study: (1) those of Japanese origin had to have three or four grandparents born in Japan; (2) those of Caucasian origin had to be descended from Caucasians with no ancestors of Negro, Asian or other ethnic origin; (3) they had to be free from health problems such as cardiac diseases, critical renal diseases, diabetes, severe infections, fractures, etc; (4) they could not habitually ingest alcoholic drinks; (5) they had to have had no pregnancies. Using these criteria, 113 Japanese and 145 Caucasian 10 to 11-y-old girls, and 57 Japanese and 121 Caucasoid 16 to 17-y-old girls were selected. The data collected from adolescents of other ethnicities (Negro, mixed-race, Caucasian, Asian, etc) is to be used in other studies as the samples of these groups were insufficient for inclusion in this report.

Anthropometry and body composition

The measurements, which took 15-20 min per subject, were taken during a gym class. Height was measured to the nearest millimetre with a tape-measure affixed to a wall. The weight and percentage of body fat were measured simultaneously using a Tanita bioelectrical impedance analyzer (model TBF 521, Tanita Corporation of America Inc., Arlington Heights, IL). The estimation of body composition using electrical impedance equipment is based upon the principle that the conduction of low-frequency electric current in the fat-free mass is higher than the conduction in fat.¹² In the present study the foot-to-foot system was employed as it has operational advantages over the conventional BIA method⁷ when used with large samples.

The adolescents were weighed in their school uniform, which consisted of T-shirt and gym-suit, as it was not possible to weigh them in underwear only. After cleaning their feet with a piece of cotton-wool soaked in alcohol, the girls stood upright and barefooted on the electrodes of the analyzer. The electrodes were also cleaned before each measurement. The following criteria were adopted for body fat measurement using the Tanita equipment: (1) weight and percentage of body fat were determined before the adolescents started to perform any physical activity in the gym class; (2) no measurements were taken from the 16- and 17-year-old adolescents during the period from one week before menstruation to the end of menstruation; (3) none of the adolescents were using diuretic drugs in the 12 h preceding the measurement.

For the BMI the cut-off point for thinness was set at the 5th percentile of the BMI distribution of the NCHS reference population¹³ and the cut-off point for overweight and obesity was set at the 85th percentile.

The NIR method was performed using the FUTREX-5000 A (Futrex Inc., Gaithersburgs, MD). This apparatus consists of a monochromatic wave emitter and a fibre optic probe, which both conducts radiation from the emitter to a site selected on the body (biceps) and picks up the interactive radiation.⁸ The analyser measures the percentage of fat and the number of kilograms of fat and fat-free mass. Two measurements were taken from the biceps mid-line. If the measurements disagreed by more than 3% they were repeated until agreement was reached.

The Slaughter equations⁹ were chosen for the analysis of the relationship between body fat and skinfolds. The tricipital and subscapular skinfolds used in the equations were measured conventionally on the right side of the adolescents using Lange skinfold callipers with a constant pressure of 10 g/mm².¹⁴ Each measure was determined three times and the final result was the mean of the three values obtained. If one of the values disagreed by more than 5% with the others in the same site then a new series of three measures was performed.

The cut-off point considered as a low percentage of fat was 15%.¹⁵ Two cut-off points for a high percentage of fat of 28%¹⁵ and 30%¹⁶ were separately analyzed for the SKI, BIA and NIR methods. All the measurements were taken by the same person.

Statistical analysis

The data analysis was performed using the program EPI INFO, version 6. The statistical comparison of the methods was evaluated by the kappa agreement test¹⁷ and complemented by the McNemar disagreement test.¹⁸ Spearman's correlation analysis was also applied to the skewed distribution variables. For all the tests, the level of statistical significance was set at 5% (P£0.05). The results for weight, height and percentage of body fat provided by the different methods are listed in Table 1. Although percentage of fat is a skewed distribution variable and therefore mean and standard deviation are not strictly appropriate, these figures have been included to allow a better visualization of the results.

Results

To help clarify the results a full table of the kappa and McNemar tests for BMI´BIA with a cut-off point of 28% for the 10 to 11-y-olds, has been included (Table 2). The other results of the kappa and McNemar tests are presented in a summarized form only, in charts 1, 2, 3 and 4.

For the 10- and 11-y-old girls, the BMI agreed considerably and significantly with the other methods (Table 2, charts 1 and 2). As concerns the cut-off points, the agreements were generally better when the 30% fat cut-off was used.

Although the BMI presented high values of agreement, significant disagreements were also found. Using either the 28% or 30% cut-off point the disagreement between the BMI and NIR was higher above the diagonal, revealing that, within both groups, the girls identified as obese by the BMI were identified as eutrophic by the NIR. Among the Japanese adolescents, the correlation with the other methods was the inverse, ie the disagreement below the diagonal was higher, demonstrating that adolescents classified as eutrophic or malnourished by the BMI were respectively obese or eutrophic by the BIA and SKI methods. Among the Caucasian adolescents, the disagreement below the diagonal was only higher for the BMI´BIA comparison.

The Spearman's correlation analysis (Table 3) showed that, among pre-menarcheal girls, the BMI presented a good correlation with the other methods, mainly with BIA (Japanese, BMI´BIA, r=0.95, r²=0.90; Caucasian, BMI´BIA, r=0.91, r²=0.83), although the correlation with NIR was poor.

For the 16- and 17-y-old adolescents the comparison of the BMI with the other methods mostly demonstrated a low or absent correlation irrespective of the cut-off point adopted (charts 3 and 4). Only NIR, in Caucasian adolescents, with a cut-off point of 30%, showed agreement with the BMI (70.2%). Moreover, the analysis presented significant disagreements below the diagonal, indicating that the BMI identified fewer obese adolescents than the other methods. On the other hand the agreement between NIR, BIA and SKI was high.

For 16- and 17-y-olds the 28% cut-off point generally presented better results in the agreement tests between BIA, SKI and NIR than did the 30% cut-off. The most substantial correlation identified by the Spearman's test (Table 3) was BMI´NIR (Japanese, r=0.81, r²=0.66; Caucasian, r=0.84, r²=0.71). However, the agreement analysis revealed that the BMI´NIR presented an absence of agreement in the Japanese and low agreement in Caucasoid adolescents for the 28% cut-off point.

Although high values of agreement were encountered between the other methods (BIA, NIR and SKI), the Spearman's correlation test showed relatively low values (Table 3).

Discussion

The main objective of this study was to evaluate the ability of the BMI to identify obese adolescents, bringing into consideration ethnicity and menarche. To this end, the BMI was compared with other widely known methods of body composition analysis.

A major problem for this investigation is the lack of a 'gold standard' method for populational studies of this nature. It is clear that without the existence of such a 'gold standard' it is difficult to establish which method is the most precise for evaluating body composition. Nevertheless, this does not prevent interesting correlations between the various methods from being analysed.

As regards ethnicity, interpretation of the results did not reveal any striking differences between the Japanese and Caucasian groups of the same age. On the whole, the methods which presented the best correlations and agreements in the analysis of the Japanese adolescents also demonstrated satisfactory results for the Caucasian adolescents in the same age-group.

Although the comparison of BMI with the other methods didn't reveal any substantial differences between the ethnic groups, other studies have shown that the relationship between BMI and percentage of fact is affected by ethnicity. Some studies have emphasized that abnormal trunk/leg proportions can lead to distortions in the results of BMI and body fat.³ As Asian individuals tend to have a greater trunk/leg proportion, this could lead to a variation in the relationship between the BMI and the percentage of fat. Furthermore, various studies have shown that, for the same BMI, the percentage of fat in Asians is greater than it is in whites.^4,5,6,19

In the present study, the comparison of the methods revealed small differences between the Japanese and Caucasian girls; however, it is difficult to establish the nature of these variations without comparison to a 'gold standard'. Therefore, it is not possible to know whether these variations were due to the methods used, which could have measured Japanese and non-Japanese girls in different ways, or whether they were real differences inherent to the ethnic groups.

As concerns the age-groups, there were remarkable differences. The BMI represented a good index of body fat evaluation in pre-menarcheal girls, presenting strong agreement and correlation with the other methods. Conversely, the BMI in post-menarcheal adolescents did not reveal satisfactory results, mainly in regard to the diagnosis of nutritional status. In all cases, the BMI identified fewer obese adolescents than the other methods, which is in agreement with several other studies which also reported that the BMI sometimes fails to identify substantial numbers of obese individuals.^1,14,20,21 Despite this poor agreement, the Spearman's correlation of the BMI with the other methods was relatively good. On the other hand, the comparisons between the other methods (NIR, BIA and SKI) had good agreement results, yet the Spearman's correlation was poor. This discrepancy in the analyses of correlation and agreement may be due to the fact that in that Spearman's correlation analysis the data are individualized, although the degree of the relationship between the variables may be quite diverse. For example, a wide variation in the percentage of fat was found for the same BMI score. However, in the agreement tests, the data are categorized, which provides for a more precise analysis that is better suited to diagnosing nutritional status.

Although many studies have attained good results with the NIR method,^8,22,23 others have not.^24,25,26,27 Most of these studies demonstrated that NIR underestimates the values of body composition. The present study presented good agreement between NIR and the other methods, although NIR did present a bias towards underestimation.

Numerous studies have reported good results obtained with BIA.^28,29,30,31 Some authors have demonstrated that the foot-to-foot method adopted in the present study presents good correlations with the methods of hydrostatic weight, dual-energy X-ray absorptiometry (DEXA) and the conventional method of BIA.^7,32,33,34 In the present study, the BIA values were generally higher than those obtained by the other methods. Tsui et al³² also reported that the system of foot-to-foot BIA overestimated fat percentage in women, in comparison to DEXA. However, these authors stated that this difference, although significant, had little clinical relevance. In the present study, the overestimate may have been partly associated with the fact that measurements were taken with the subjects in school uniform rather than just in underwear, as the weight of these clothes may have increased the values given for percentage of fat. Nevertheless, taking into account the statement of Tsui et al,³² it is suggested that this increase does not radically change the diagnosis of the adolescents nutritional status.

The Slaughter equations⁹ attempt to minimize, to the greatest extent possible, the skew known to affect the prediction of the percentage of fat by skinfold. These equations take into account the influence of the subjects' ethnicity, sex and sexual maturation, and for this reason they were included in the present study. However, the Slaughter equations have only been developed in subjects of white and black ethnicities and care must therefore be exercised when analysing results from subjects from other ethnic groups. Nevertheless, another factor which favoured the use of these equations in the present study is that, in female subjects, the relationship between skinfolds and the percentage of fat has been shown not to be significantly affected by ethnic group by several studies that have validated these equations.^9,35,36 In the present study, the Slaughter equations produced good results in comparison with the other methods. However, it is important to emphasize that the collection of skinfold measures must be performed by a well-trained professional and that obtaining these measures in obese subjects is very difficult, which can consequently lead to potentially serious distortions in the results for percentage fat that they provide.

With regard to the cut-off points, it was observed that the best value for the 10 to 11-y-old age-group was 30%, whereas in the post-menarcheal adolescents the 28% cut-off point seemed to provide better results. Further studies are necessary to investigate the correlations between these cut-off points and other biochemical, physiological and metabolic variables so that definitive limits, which correspond with the aims of different nutritional evaluation programmes, can be established.

In conclusion, this study shows that the BMI can be used to substitute for other methods of evaluation of nutritional status in pre-menarcheal girls, without regard to ethnic group, although the fact that the BMI may sometimes underestimate obesity should be taken into account. On the other hand, in Japanese and Caucasian post-menarcheal adolescents, the BMI is probably not a good index for studies that are focused on the identification of obesity. In this case, the choice of one of the other methods based on availability, cost or technical experience may represent a better approach.

Acknowledgements

The authors would like to thank the directors, assistants, gymnastics teachers and participant pupils of the following São Paulo schools for their collaboration in the present study: Anglo Latino, Mater et Magistra, Itamaraty, Roberto Norio, São José, Pioneiro, Ursa Maior, Anglo-Brasileiro, Brasilia, Radial, Oshiman, Montessori, Bilac, Renovação and Madre Cabrini, as well as the Caramuru and Hongwanji scout groups and the Shohaku language school. We would also like to thank CAPES for the provision of grants and financial aid.

1 Ellis KJ, Abrams AS, Wong WW. Monitoring childhood obesity: assessment of weight/height² index. Am J Epidemiol 1999; 150: 939-946, MEDLINE

2 Gallangher D, Marjolein V, Sepúlveda D, Pierson RN, Harris T, Heymsfield SB. How useful is body mass index for comparison of body fatness across age, sex, and ethnic groups? Am J Epidemiol 1996; 143: 228-239, MEDLINE

3 Norgan NG. Population differences in body composition in relation to the body mass index. Eur J Clin Nutr 1994; 48: (Suppl 3) S10-S27, MEDLINE

4 Wang J, Thornton JC, Russel M, Burastero S, Heymsfield S. Asians have lower body mass index (BMI but higher percent body fat than do whites: comparisons of anthropometric measurements. Am J Clin Nutr 1994; 60: 23-28, MEDLINE

5 Gurrici S, Hartriyanti Y, Hautvast JGAJ, Deurenberg P. Relationship between body fat and body mass index: differences between Indonesians and Dutch Caucasians. Eur J Clin Nutr 1998; 52: 779-783, MEDLINE

6 Deurenberg P, Deurenberg Yap M, Wang J, Lin FP, Schmidt G. The impact of body build on the relationship between body mass index and percent body fat. Int J Obes Relat Metab Disord 1999; 23: 537-542, MEDLINE

7 Nuñez C, Gallangher D, Visser M, Pi-Sunyer FX, Wang Z, Heymsfield SB. Bioimpedance analysis: evaluation of leg-to-leg system based on pressure contact foot-pad electrodes. Med Sci Sports Exerc 1997; 29: 524-531, MEDLINE

8 Conway JM, Norris KH, Bodwell CE. A new approach for the estimation of body composition: infrared interactance. Am J Clin Nutr 1984; 40: 1123-1130, MEDLINE

9 Slaughter MH, Lohman TG, Boileau CA, Horswil CA, Stillman RJ, Van Loan MD, Bemben DA. Skinfold equations for estimations of body fatness in children and youth. Hum Biol 1988; 60: 709-723, MEDLINE

10 Must A, Dallal GE, Dietz WH. Reference data for obesity: 85th and 95th percentiles of body mass index (wt/ht²) and triceps skinfolds thickness. Am J Clin Nutr 1991; 53: 839-846, MEDLINE

11 Senior PA, Bhopal R. Ethnicity as variable in epidemiological research. Br Med J 1994; 309: 327-330,

12 Baumgartner RN. Electrical impedance and total body electrical conductivity. In: Roche AF, Heymsfield SB, Lohman TG (eds). Human body composition. Human Kinetics Books: Champaign, 1996, pp 79-107.

13 Himes JH, Dietz WH. Guidelines for overweight in adolescent preventive services: recommendations from an expert committee. Am J Clin Nutr 1994; 59: 307-316, MEDLINE

14 Durnin JVGA, Rahaman MM. The assessment of the amount of fat in the human body from measurements of skinfolds thickness. Br J Nutr 1967; 21: 681-688, MEDLINE

15 Sigulem DM, Veiga GV, Priore SE. Obesidade em adolescentes de baixa renda. In: Fisberg M (ed). Obesidade na Infância e Adolescência. Fundação BYK: São Paulo, 1995, pp 80-83.

16 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, MEDLINE

17 Landis JR, Kock GG. The measurement of observer agreement for categorical data. Biometrics 1977; 33: 159-174, MEDLINE

18 Siegel S. Nonparametric statistics for behavioral sciences. McGraw-Hill: New York, 1956.

19 Potts J, Simmons D. Sex and ethnic group differences in fat distribution in young United Kingdom South Asians and Europids. J Clin Epidemiol 1994; 47: 837-841, MEDLINE

20 Himes JH, Bouchard C. Validity of anthropometry in classifying youths as obese. Int J Obes 1989; 13: 183-193, MEDLINE

21 Smalley KJ, Knerr NA, Kendrick ZV, Colliver JA, Owen OE. Reassessment of body mass indices. Am J Clin Nutr 1990; 52: 405-408, MEDLINE

22 Davis PO, Dotson CO, Manny PD. NIR evaluation for body composition analysis. Med Sci Sports Exerc 1988; 20: S8,

23 Dotson CO, Davies O. Technical Note: evaluation of near-infrared spectrometry for body composition analysis in children and youth. FUTREX, Inc.: Gaithersburg, MD, 1989,

24 Mclean KP, Skinner JS. Validity of Futrex-5000 for body composition determination. Med Sci Sports Exerc 1992; 24: 253-258, MEDLINE

25 Elia M, Parkinson AS, Diaz E. Evaluation of near infrared interactance as a method for predicting body composition. Eur J Clin Nutr 1990; 44: 113-121, MEDLINE

26 Thomas DW, Ryde SJ, Ali PA, Birks JL, Evans CJ, Saunders NH, Al Zeibak S, Dutton J, Hancock DA. The performance of an infra-red interactance instrument for assessing total body fat. Physiol Measurement 1997; 18: 305-315,

27 Wilmore KM, McBride PJ, Wilmore JH. Comparison of bioelectric impedance and near-infrared interactance for composition assessment in a population of self-perceived overweight adults. Int J Obes Relat Metab Disord 1994; 18: 375-381, MEDLINE

28 Bandini LG, Vu DM, Must A, Dietz WH. Body fatness and bioelectrical impedance in non-obese pre-menarcheal girls: comparison to anthropometry and evaluation of predictive equations. Eur J Clin Nutr 1997; 51: 673-677, MEDLINE

29 Lukaski HC, Johnson PE, Bolochuk WW, Lykken GI. Assessment of fat-free mass using bioelectrical impedance measurements of human body. Am J Clin Nutr 1985; 41: 810-817, MEDLINE

30 Loan M, Mayclin P. Bioelectrical impedance analysis: is it a reliable estimator of lean body mass and total body water? Hum Biol 1987; 59: 299-309, MEDLINE

31 Kuczmarski RJ. Bioelectrical impedance analysis measurements as part of national nutrition survey. Am J Clin Nutr 1996; 64: (Suppl) 453S-458S, MEDLINE

32 Tsui EYL, Gao XJ, Zinman B. Bioelectrical impedance analysis (BIA) using bipolar foot electrodes in the assessment of body composition in type 2 diabetes mellitus. Diabetic Med 1998; 15: 125-128, MEDLINE

33 Bell NA, McClure PD, Hill RJ, Davies PSW. Assessment of foot-to-foot bioelectrical impedance analysis for the prediction of total body water. Eur J Clin Nutr 1998; 52: 856-859, MEDLINE

34 Utter AC, Nieman DC, Ward AN, Butterworth DE. Use of the leg-to-leg bioelectrical impedance method in assessing body composition change in obese women. Am J Clin Nutr 1999; 69: 603-607, MEDLINE

35 Janz KF, Nielsen DH, Cassady SL, Cook JS, Wu Y-T, Hansen JR. Cross-validation of the Slaughter skinfold equations for children and adolescents. Med Sci Sports Exerc 1993; 25: 1070-1076, MEDLINE

36 Roemmich JN, Clark PA, Weltman A, Rogol AD. Alterations in growth and body composition during puberty. I. Comparing multicompartment body composition models. J Appl Physiol 1997; 83: 927-935, MEDLINE

Figure 1 Kappa and McNemar tests between the various methods, using the same analysis pattern as the comparison between BMI´BIA (Table 2), in 10- and 11-year-old Japanese and Caucasian adolescents with a cut-off point of 28% of fat.

Figure 2 Kappa and McNemar tests between the various methods, using the same analysis pattern as the comparison between BMI´BIA, in 10- and 11-year-old Japanese and Caucasian adolescents with a cut-off point of 30% of fat.

Figure 3 Kappa and McNemar tests between the various methods, using the same analysis pattern as the comparison between BMI´BIA, in 16- and 17-year-old Japanese and Caucasian adolescents with a cut-off point of 28% of fat.

Figure 4 Kappa and McNemar tests between the various methods, using the same analysis pattern as the comparison between BMI´BIA, in 16- and 17-year-old Japanese and Caucasian adolescents with a cut-off point of 30% of fat.

Table 1 Mean and standard deviation of anthropometric measures of Japanese and Caucasian female adolescents aged 10-11 and 16-17 from private schools in São Paulo

Table 2 Kappa agreement test (k) between BMI´BIA, in 10- and 11- y-old Japanese adolescents with a cut-off point of 28% of fat

Table 3 Spearman's correlations (r and r²) between the BMI, BIA, NIR and SKI methods