Introduction

Childhood obesity is a serious and growing problem in affluent countries as well as in populations undergoing economic transition (Koletzko et al, 2002), and represents a significant public health concern. Obesity is defined as an excessive amount of fat in the body. However, it is not possible to measure body fat content in a simple way. Various indices for identifying obesity in humans have therefore been developed, the most commonly used being body mass index (BMI). BMI is simple to measure, but its application is complicated, since the BMI distribution in children and adolescents varies markedly with age and gender (Cole et al, 2000). Furthermore, as pointed out by Wells (2000), BMI identifies subjects with an excessive body weight rather than subjects with an excessive amount of body fat.

Another simple and potentially useful method for estimating total body fat (TBF) is the skinfold technique, which can produce valid although rather imprecise results in adults (Durnin & Womersley, 1974). This technique measures the thickness of the subcutaneous fat layer directly, and therefore represents a seemingly attractive way to measure TBF. Skinfold measurements are relatively easy to obtain, but may involve some discomfort for the child. A procedure based on this technique and applicable to children below 2 y of age has been described (Weststrate & Deurenberg, 1989), but has thus far not been satisfactorily evaluated in this age group.

Evaluation of body composition methodology requires an appropriate reference method. Unfortunately, however, no golden standard for measuring human body composition 'in vivo' is available. For many years, underwater weighing was considered the best available method for use as a reference, but for obvious reasons this cannot be applied to children below 2 y of age. During the past decade, however, the doubly labelled water (DLW) method has been established for studies of human energy metabolism. This method produces estimates of so-called isotope dilution spaces, making calculation of total body water (TBW) possible. By using estimates of the degree of hydration of the fat-free mass (FFM), TBF can be calculated from TBW and body weight, a procedure known as the body water dilution (BWD)-method.

In this paper, we report on a study where body fatness, estimated using the skinfold technique, described by Weststrate and Deurenberg (1989), and BMI, was compared to body fatness estimated using the BWD-method in children 9 or 14 months of age. For this kind of comparison, the procedure described by Bland and Altman (1986) is generally recommended. However, this requires that the methods under comparison measure the same variable. This was not the case in our study since BMI, although intended to be an estimate of body fatness, is in fact not a numerical estimate of %TBF. A ranking and grouping procedure allowing for the intended comparisons was therefore used.

Subjects and methods

Subjects

In total, 65 children were recruited, as described by Tennefors et al (2003), to take part in the present study. Six of these children were excluded from the study due to minor illnesses or poor parent compliance. The 59 children shown in Table 1 participated at 9 or 14 months of age. This study was part of a survey of the health and nutrition of 300 children recruited from a well-educated, middle-class population in Umeå, Sweden. The 65 children, all of whom were full term and healthy, were randomly selected from those 300 for anthropometric measurements as well as assessment of total energy expenditure and TBW using the DLW-method. The study was approved by the ethics committee of the faculty of medicine and odontology at Umeå University.

Table 1 - Weight, length, weight-for-age z-score, length-for-age z-score, weight-for-length z- score for the children in the study.

Full table

Design

Weight and length were measured on the same day as the dose of stable isotopes was given for the DLW measurements. Skinfold thickness (SFT) measurements were taken 11 days later.

Assessment of body fatness

BMI, that is, weight (kg)/height (m)², was calculated from body weight (naked) measured to the nearest 0.02 kg (weighing scale for babies–adults 0–136 kg, CMS Weighing Equipment Ltd, London, UK) and recumbent length measured to the nearest 0.1 cm (Rollameter baby measure mat, CMS Weighing Equipment Ltd, London, UK).

Total body fat based on skinfold thickness measurements (TBF-SFT) was assessed using a Harpenden skinfold caliper (Harpenden skinfold caliper, CMS Weighing Equipment Ltd, London, UK). Skinfolds measured at bicipital, tricipital, subscapular and suprailiacal sites (Durnin & Rahaman, 1967) were used to calculate %TBF-SFT, that is, body fatness, according to Weststrate & Deurenberg (1989). All skinfolds were the average of three measurements assessed on the left side of the body, always by the same observer. The technical error of measurement (Moreno et al, 2002) ranged from 0.14 to 0.20 mm. The coefficient of reliability (Moreno et al, 2002) ranged from 99.5 to 99.8%.

Total body fat based on body water dilution (TBF-BWD) was calculated from TBW measured using the DLW-method (Tennefors et al, 2003). FFM was assumed to have the following degrees of hydration: 79% (girls) and 79.3% (boys) at 9 months of age, and 78.7% (girls) and 78.8% (boys) at 14 months of age. These figures represent reference data as described by Fomon et al (1982). TBF was calculated as body weight minus FFM, and FFM as TBW divided by the appropriate degree of hydration of FFM. To obtain body fatness, TBF–BWD was expressed in per cent of body weight, %TBF-BWD.

Evaluation of classification capacity

This procedure involved ranking the children in the study on the basis of their level of body fatness in a sequence where the child with the lowest level of body fatness had the lowest number, and where the difference in the level of body fatness between this child and the second child in the sequence was the smallest possible. This principle of the smallest possible difference was maintained for all children, producing a sequence with a gradually increasing level of body fatness. Based on this sequence, the children were then divided into five groups with increasing levels of body fatness comprising 12 (lowest), 12 (second lowest), 11 (middle), 12 (second highest) and 12 (highest) children, respectively. This ranking and grouping procedure was carried out for estimates of body fatness obtained using %TBF-BWD, %TBF-SFT and BMI. The classification capacity of %TBF-SFT and BMI was then evaluated as the number of children placed in the same (0), in the next higher (+1) or lower (-1), in the second next higher (+2) or lower (-2), in the third next higher (+3) or lower (-3), or in the fourth next higher (+4) or lower (-4) group when compared to the groups obtained when the classification was based on %TBF-BWD. The BWD-method was thus considered as the reference method capable of providing a correct classification of body fatness.

Statistical analyses

The statistical computation was performed using Excel 1997, Microsoft, Redmond, Seattle, USA. Results are expressed as means s.d. Values for the same variable obtained by two different methods were compared according to Bland and Altman (1986). Paired comparisons were used to test if means obtained using two independent methods were significantly different (Altman, 1999).

Results

Table 2 shows BMI, %TBF-SFT and %TBF-BWD of the children in the study. On average, the BMI of all children was 17.5 kg/m², while their average contents of TBF-SFT and TBF-BWD were 27.8 and 29.1%, respectively.

Table 2 - Body fatness by body mass index BMI, by skinfold thickness (%TBF-SFT) and body water dilution (%TBF-BWD) of the children in the study.

Full table

Table 3 presents the mean, s.d., range and CV for BMI, %TBF-SFT and %TBF-BWD of children in five groups representing five different levels of body fatness.

Table 3 - Body fatness of children in five groups with increasing levels of body fatness obtained on the basis of BMI, the skinfold technique (%TBF-SFT) and using body water dilution (%TBF-BWD).

Full table

Figure 1 compares estimates of %TBF-BWD and %TBF-SFT according to Bland and Altman (1986). The difference between results obtained by the two methods was 1.35% TBF, while limits of agreement (2 s.d.) were –6.76 to 9.47% TBF. The difference between the two methods was statistically significant (P<0.05) and the 95% confidence interval for the bias was 0.29–2.41% TBF. No significant correlation between the average of results obtained by the two methods and the difference between them was obtained.

Figure 1.

Comparison of body fatness obtained by the body water dilution method (%TBF-BWD) and by the skinfold technique (%TBF-SFT) according to Bland and Altman (1986) (n=59). Mean difference %TBF-BWD - %TBF-SFT=1.35% (2 s.d. 8.11%).

Full figure and legend (19K)

Figure 2a and b presents the classification capacity of %TBF-SFT and BMI, respectively. When classification was based on %TBF-SFT as well as on BMI, 21 of the 59 children were classified correctly. However, serious misclassification, that is, the number of children classified in two or more groups too high or too low, differed for %TBF-SFT and BMI. Thus, %TBF-SFT classified 17 (29%) of the 59 children in this way, while the corresponding figure for BMI was 10 children (17%).

Figure 2.

(a) Classification capacity of %TBF-SFT vs %TBF-BWD. (b) Classification capacity of BMI vs %TBF-BWD. The figure shows the number of children classified in the same (0), in the next higher (+1) or lower (-1), in the second next higher (+2) or lower (-2), in the third next higher (+3) or lower (-3), or in the fourth next higher (+4) or lower (-4) group as compared to the groups obtained when the classification was based on %TBF-BWD.

Full figure and legend (29K)

Figure 3 shows the classification capacity of %TBF-SFT (I) and BMI (II), for groups with the following levels of fatness on the basis of %TBF-BWD: lowest (a), second lowest (b), middle (c), second highest (d) and highest (e). For the 12 children with the lowest level of fatness based on %TBF-BWD, %TBF-SFT and BMI classified 58% (Figure 3aI) and 50% (Figure 3aII), respectively, of the children correctly. For the 12 children with the second lowest level of fatness, the corresponding figures were 25% for %TBF-SFT (Figure 3bI) and 8% for BMI (Figure 3bII). For the remaining three groups, that is, the middle fatness group, the group with the second highest level of fatness and the group with the highest level of fatness, the corresponding figures were 18% for both %TBF-SFT and BMI Figure 3cI and II), 33% (Figure 3dI) and 42% (Figure 3dII) as well as 42% (Figure 3eI) and 58% (Figure 3eII), respectively, for %TBF-SFT and BMI. In addition, the results presented in Figure 3 indicate that when compared to children with an average level of %TBF-BWD, there is a tendency for both methods to classify more children correctly in the groups, where %TBF-BWD is extremely high or low. The two methods seem to have a fairly similar capacity to classify children in these groups.

Figure 3.

Classification capacity of %TBF-SFT and BMI vs %TBF-BWD for children with different levels of body fatness. (a) Number of children classified by %TBF-SFT (I) and BMI (II) in the same (0), in the next higher (+1), in the second next higher (+2), in the third next higher (+3) or in the fourth next higher (+4) group for children with the lowest level of fatness on the basis of %TBF-BWD (n=12). (b) Number of children classified by %TBF-SFT (I) and BMI (II) in the same (0), in the next higher (+1) or lower (-1), in the second next higher (+2) or in the third next higher (+3) group for children with the second lowest level of fatness on the basis of %TBF- BWD (n=12). (c) Number of children classified by %TBF-SFT (I) and BMI (II) in the same (0), in the next higher (+1) or lower (-1), or in the second next higher (+2) or lower (-2) group for children with the middle level of fatness on the basis of %TBF-BWD (n=11). (d) Number of children classified by %TBF-SFT (I) and BMI (II) in the same (0), in the next higher (+1) or lower (-1), in the second next lower (-2), or in the third next lower (-3) group for children with the second highest level of fatness on the basis of %TBF-BWD (n=12). (e) Number of children classified by %TBF-SFT (I) and BMI (II) in the same (0), in the next lower (-1), in the second next lower (-2), in the third next lower (-3) or in the fourth next lower (-4) group for children with the highest level of fatness on the basis of %TBF-BWD (n=12).

Full figure and legend (91K)

Discussion

When compared to the National Center for Health Statistics growth reference data (WHO, 1983), the children in our study were slightly heavier and taller than reference children (Table 1). The BMIs of the 9- and 14-month-old children in our study are in good agreement with recently published Swedish population-based reference values, although the 50th-percentile values of our children are slightly lower and the s.d.'s slightly higher (He et al, 2000).

In this study, we assume that estimates of TBF obtained using the BWD-method represent accurate estimates of body fatness. Recently, Wells et al (1999) studied body composition of children aged 8–11 y and found underwater weighing to be associated with a larger error than the BWD-method. These authors also investigated the variation in estimates of the degree of hydration of FFM and presented data showing that for the children in their study, the interindividual variation in this variable is rather small. Their conclusion was that the BWD-method is appropriate for determination of body fat in individual as well as in groups of children between 8 and 11 y of age. Unfortunately, no similar study is available for children in the age group studied in this paper. However, Butte et al (2000) have recently provided estimates of the degree of hydration in FFM for children under 2 y of age. The values reported for children between 3 months and 2 years are in close agreement with the corresponding values published by Fomon et al (1982), that is, those used in the present study. Furthermore, the variation in these estimates (Butte et al, 2000) is reported to be rather small (CV=1.4–2%). Therefore, it is reasonable to assume that our estimates of %TBF-BWD represent appropriate reference data for evaluation of alternative methods for assessing body fatness in children of the same age as those studied in this paper.

Our results indicate that the method proposed by Weststrate and Deurenberg (1989) provides estimates of TBF that tend to underestimate body fatness for 9- and 14-month-old children. This is in contrast to the results by de Bruin et al (1995), who found that this method overestimated body fatness in infants when total body electrical conductivity was used as the reference method. Their infants were, however, considerably different from our subjects with respect to age and body fatness. Another factor of possible significance for the different results obtained in the two studies is that the fat content in adipose tissue varies considerably in infants and young children (Baker, 1969; Kabir & Forsum, 1993).

Owing to the need for simple methods for assessing body composition in young children, it is disappointing to find that neither skinfold measurements nor BMI was able to place more than about 35% of our children in the correct body fatness group. It is relevant to point out that 20% of the children, that is, about 12 of our 59 children, can be expected to be classified correctly just by random. The number of children classified in two or more groups too high or too low (that is, serious misclassification) was high for both methods, but was higher for %TBF-SFT than for BMI. The finding that the two methods were both able to classify more children correctly in groups where body fatness was at its lowest or highest is noteworthy. It is likely that a large proportion of the children in these groups would have been considered as substantially leaner or fatter than average without any measurements at all.

We have not been able to find any data in the literature where body fatness assessed by means of skinfolds and BMI has been evaluated in children below 2 y of age. However, in a study on boys 7–16.9 y of age, Sarría et al (1998) found that estimates based on skinfold measurements were better predictors than BMI of body fatness, a finding that is in contrast to the results of the present study. We believe that this discrepancy is likely to be due to a true difference between the children in the two studies since the precision of our skinfold measurements compares favorably with that reported by other authors (Moreno et al, 2002). The fact that skinfolds were measured 11 days later than the estimates of weight and length used to calculated BMI did not affect the results of the present study.

The results of the present study illustrate the difficulties encountered when assessing body fatness of individual children. We found that BMI was a rather unsatisfactory tool for that purpose, and a method proposed by Weststrate & Deurenberg (1989) based on skinfold measurements produced even less satisfactory results. Also, de Bruin et al (1995) concluded from their study that this method cannot be used to assess accurately TBF in an individual infant.

Obviously, there is a need for improved methodology to study body composition of children below 2 y of age. Development of such methodology is important, as there is a growing need to identify children at risk of developing obesity as early in life as possible.

References

1. Altman DG (1999): Practical Statistics for Medical Research, pp 179–223. London: Chapman & Hall.
2. Baker GL (1969): Human adipose tissue composition and age. Am. J. Clin. Nutr. 22, 829–835. | PubMed | ChemPort |
3. Bland JM & Altman DG (1986): Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. i, 307–331.
4. de Bruin NC, van Velthoven KAM, Stijnen T, Juttmann RE, Degenhart HJ & Visser HKA (1995): Quantitative assessment of infant body fat by anthropometry and total body electrical conductivity. Am. J. Clin. Nutr. 61, 279–286. | PubMed |
5. Butte NF, Hopkinson JM, Wong WW, Smith EOB & Ellis KJ (2000): Body composition during the first 2 years of life: An updated reference. Pediatr. Res. 47, 578–585. | PubMed | ChemPort |
6. Cole TJ, Bellizzi MC, Flegal KM & Dietz WH (2000): Establishing a standard definition for child overweight and obesity world wide: international survey. BMJ. 3320, 1–6.
7. Durnin JVGA & Rahaman MM (1967): The assessment of the amount of fat in the human body from measurements of skinfold thickness. Br. J. Nutr. 21,681–689. | Article | PubMed | ISI | ChemPort |
8. Durnin JVGA & Womersley J (1974): Body fat assessed from total body density and its estimation from skinfold thickness: measurements on 481 men and women aged from 16–72 years. Br. J. Nutr. 32, 77–97. | Article | PubMed | ISI | ChemPort |
9. Fomon SJ, Haschke F, Ziegler EE & Nelson SE (1982): Body composition of reference children from birth to age 10 years. Am. J. Clin. Nutr. 35, 1169–1175. | PubMed | ISI | ChemPort |
10. He Q, Albertsson-Wikland K & Karlberg J (2000): Population-based body mass index reference values from Göteborg, Sweden: birth to 18 years of age. Acta Paediatr. 89, 582–592. | Article | PubMed | ChemPort |
11. Kabir N & Forsum E. (1993): Estimation of total body fat and subcutaneous adipose tissue in full-term infants less than 3 months old. Pediatr. Res. 34, 448–454. | PubMed |
12. Koletzko B, Girardet J-P, Klish W & Tabacco O (2002): Obesity in children and adolescents worldwide: current views and future directions — working group report of the first world congress of pediatric gastroenterology, hepatology & nutrition. J. Pediatr. Gastroenterol. Nutr. 35, 205–212. | Article | PubMed |
13. Moreno LA, Rodríguez G, Guillén J, Rabanaque MJ, León JF & Ariño A (2002): Anthropometric measurements in both sides of the body in the assessment of nutritional status in prepubertal children. Eur. J. Clin. Nutr. 56, 1208–1215. | Article | PubMed | ISI | ChemPort |
14. Sarría A, García-Llop LA, Moreno LA, Fleta J, Morellón MP & Bueno M (1998): Skinfold thickness measurements are better predictors of body fat percentage than body mass index in male Spanish children and adolescents. Eur. J. Clin. Nutr. 52, 573–576. | Article | PubMed | ChemPort |
15. Tennefors C, Coward WA, Hernell O, Wright A & Forsum E (2003): Total energy expenditure and physical activity level in healthy young Swedish children 9 or 14 months of age. Eur. J. Clin. Nutr. 57, 647–653. | Article | PubMed | ChemPort |
16. Wells JCK (2000): A Hattori chart analysis of body mass index in infants and children. Int. J. Obes Relat Metat Disord. 24, 325–329. | Article |
17. Wells JCK, Fuller NJ, Dewit O, Fewtrell MS, Elia M & Cole TJ (1999): Four-component model of body composition in children: density and hydration of fat-free mass and comparison with simpler models. Am. J. Clin. Nutr. 69, 904–912. | PubMed | ISI | ChemPort |
18. Weststrate JA & Deurenberg P (1989): Body composition in children: proposal for a method for calculating body fat percentage from total body density or skinfold-thickness measurements. Am. J. Clin. Nutr. 50, 1104–1115. | PubMed | ISI | ChemPort |
19. WHO (1983): Measuring Change in Nutritional Status. Guidelines for Assessing the Nutritional Impact of Supplementary Feeding Programmes for Vulnerable Groups. Geneva: World Health Organization.
20. WHO (1995): Physical Status: The Use and Interpretation of Anthropometry. Report of a WHO Expert committee. WHO Technical Report Series, No. 854. Geneva: WHO.

Acknowledgements

We thank all the children and parents who participated in the study. We are also grateful to Margareta Henriksson for skillful assistance throughout this study. Olle Hernell is thanked for medical coverage. Financial support was received from Semper AB, Sweden, and the Swedish Research Council, Project Number 12172.