Department of Pediatrics, University of Verona, Polyclinic, Verona, Italy

Correspondence to: C Maffeis, Department of Pediatrics, University of Verona, Polyclinic, 37134 Verona, Italy. Email: maffeis@borgoroma.univr.it

OBJECTIVE: The aim of this study was to identify in a group of 8-y-old prepubertal children the anthropometric parameter with the highest prediction power of overweight, measured 4 y later.

SUBJECTS: One-hundred and twelve Caucasian children (54 males, 58 females), aged 8.7±0.9 y, were studied.

RESULTS: An analysis of the association between relative body mass index (BMI) at follow-up (%) and some indexes of adiposity like relative BMI (%), waist circumference, subscapular and triceps skinfolds, the sum of four skinfolds and percentage fat mass measured at baseline, showed that relative BMI (relBMI) at baseline had the highest association with relBMI at follow-up (r=0.77; P<0.001); waist circumference had a slightly lower significant association with relBMI at follow-up (r=0.74; P<0.001). In a multiple regression analysis, waist circumference (adjusted for age) accounted for ~64% of the variation of relBMI at follow-up (P<0.001). RelBMI measured at baseline accounted for ~59% of the variation of relBMI at follow-up (P<0.001). Multiple logistic regression analysis included waist circumference, adjusted for age, mother's BMI and relBMI measured at baseline as independent variables in the final equation. In particular, each centimeter increase of waist circumference at the age of 8 y doubled the risk of having a relBMI greater than 120% at the age of 12 y.

CONCLUSION: The results of this study, the first which has approached this investigation in children, showed that waist circumference measured at the age of 8 y, which is simple to perform and easy to reproduce, may be a promising index to assess adiposity as well as to predict overweight at puberty.

International Journal of Obesity (2001) 25, 978-983

obesity; children; waist circumference

Introduction

The prevalence of pediatric obesity is increasing in the United States and Europe as well as in developing and newly industrialised countries.^1,2,3,4 In Italy, obesity affects approximately 14% of 6 to 14-y-old children and its prevalence is progressively increasing.² The epidemic of obesity in childhood is cause of concern for several reasons: first, evidence exists that childhood obesity continues into adulthood;⁵ second, obesity is associated with higher morbidity even in children.⁶ In particular, overweight in childhood is associated with several cardiovascular risk factors such as high blood pressure, hyperinsulinemia, type 2 diabetes and adverse lipid profile, all of which cause mortality in adulthood.⁶ Third, childhood obesity is associated with metabolic complications in adulthood, no matter what the weight status as adults.⁷ However, the consequences of obesity are not only physical, overweight has a big impact on the psycho-social life of an individual, on his/her quality of life as well as on personal and social economy.⁸ Recent data have shown that the economic costs of obesity reach $100 billion per year in the US.^9,10 Moreover, obesity is a disease that is very difficult to treat and the long-term results of treatment are frequently discouraging.¹¹

On the basis of the above-mentioned considerations, it is urgently important to put more effort into planning prevention programs for obesity. The available markers for selecting children at higher risk of obesity are parents' body mass index (BMI) and children's overweight.¹² In spite of the epidemic of childhood obesity, there is still no internationally accepted definition of obesity for children and adolescents. For this purpose, an anthropometric index like BMI has been proposed.¹³ However, there are some reservations in using BMI as an indicator of adiposity in children.^14,15,16 In particular, Ellis et al, comparing the BMI of 979 children with their fat mass measured by dual-energy X-ray absorptiometry (DXA), concluded that BMI can provide a general description of adiposity characteristics in a healthy pediatric population, but the accuracy in predicting fatness is poor in an individual child.¹⁴ Therefore, other adiposity indexes, which are easy to measure, economical and which have a sufficient degree of accuracy, may be useful. Moreover, the identification of an anthropometric index measured in young childhood and which has a good prediction level of weight gain in late childhood would be of interest, especially for prevention purposes.

Therefore, the aim of this study was to identify in a group of 8-y-old children the anthropometric parameter with the highest prediction power of overweight, measured 4 y later.

Materials and methods

Subjects

We studied 112 prepubertal Caucasian children (54 males, 58 females) aged 8.7±0.9 y. The children were recruited through questionnaires distributed to teachers in primary schools; private and public schools were proportionally represented. When the children and their parents had agreed to enrol in the study, they were invited to come to the medical room at their school to meet the research team. After an initial briefing, the child underwent a physical examination by a pediatrician who took anthropometric measurements. A physical examination was also performed to reasonably exclude any health problems other than simple obesity. Self-reported weight and height of the parents were used to calculate their BMI, because self-reported weight and height are generally reliable in adults.¹⁷

Complete data were obtained from all the children and their parents at baseline. Four years later, each of the families were invited by letter to come for a second visit to the project laboratory at the hospital, where height and weight measurements for each child were recorded by the pediatricians. Puberty development was clinically assessed on the basis of Tanner stages.¹⁸ Obesity was defined as relative BMI (relBMI)>120%, where relBMI=(BMI/BMI at 50th percentile for age and gender)´100. The relBMI of the children was calculated using Rolland-Cacherà et al BMI tables as reference.¹⁹

Informed consent was obtained from the parents of each child. Our protocol was done according to the 1975 Helsinki Declaration and 1983 revision.

Physical characteristics

Weight and height were measured by the same two investigators. Height was measured to the nearest 0.5 cm on a standardised height board. Weight was determined to the nearest 0.1 kg on a standard physician's beam scale, with the child dressed only in light underwear and no shoes. The standardised height board and beam scale were calibrated each day before the first measurement. The BMI was calculated as weight (kg) divided by height squared (m²). Skinfold thickness was measured to the nearest millimeter in triplicate with a Harpenden skinfold calliper (CMS, Weighing Equipment Ltd, London, UK). The triceps skinfold was measured half-way between the acromion and olecranon on the back of the arm with the elbow bent, while the subscapular skinfold was measured just below the tip of the scapula.²⁰ Readings were taken 3 s after the calliper jaws were released. In the present analysis, we used the mean of right and left-sided measurements for each skinfold. Lohman's formulae were used to estimate fat mass percentage, based on the measurement of triceps and subscapular skinfolds.²⁰ Waist circumference was measured to the nearest centimeter with a flexible steel tape measure while the subjects were in a standing position at the end of gentle expiration. The following anatomical landmarks were used: laterally, midway between the lowest portion of the rib cage and iliac crest, and anteriorly, midway between the xiphoid process of the sternum and the umbilicus.²⁰ Our between-observer error was ~30% for triceps and subscapular measurements. Our inter-observer error was 1.6±0.3 cm for waist circumference measurement. All these data are very close to the inter-observer reliability data reported by the literature.²⁰

Statistical analysis

Baseline variables are described as the groups' mean and standard deviation. Zero-order correlations were performed first to assess any unadjusted association between relBMI at follow-up and baseline parameters and parents' BMI. Partial correlation between waist circumference and final relBMI, controlling for basal relBMI, was run. Several multiple regression analyses were run to investigate the association between relBMI at follow-up and baseline parameters. RelBMI at follow-up was used as the dependent variable and relBMI at baseline, waist circumference, fat mass percentage, triceps and subscapular skinfolds, and sum of four skinfolds (tricipital, bicipital, subscapular and suprailiac), respectively. Age, gender and mother's BMI were also included as covariates. Father's BMI and Tanner's stage were not included among the covariates because they did not correlate with relBMI at follow-up.

In order to assess the prediction level of certain variables (waist circumference and mother's BMI) on the risk of becoming obese, we performed a multivariate logistic regression analysis with backward stepping of variables and an evaluation of the model using three goodness-of-fit chi-square statistics, using age and gender as covariates. The children were divided into two groups, depending on relBMI at follow-up: group A with relBMI£120% and group B, with relBMI >120%.

All statistical analyses were carried out using SPSS v 9.0 software for Windows (SPSS Inc., Chicago, IL) package for personal computers. The probability level of P<0.05 was used to indicate statistical significance in all analyses.

Results

The physical characteristics of the 112 children measured at baseline (8 y of age) are shown in Table 1. At the time of the second measurement, 4 y after baseline, the children showed an increase of 16.4±5 kg in weight (final value 46.7±9.4 kg) and 22.9±4.9 cm in height (final value 156.1±9 cm). Fifty-seven children showed an increase in their relBMI, whereas 55 children had a decrease in their relBMI; mean relBMI did not change (109±16%).

An analysis of the association between relBMI at follow-up and certain indices of adiposity such as relBMI measured at baseline, waist circumference, subscapular and triceps skinfolds, sum of four skinfolds, and fat mass percentage measured at baseline, showed that relBMI at baseline had the highest association with relBMI at follow-up (r=0.77; P<0.001). Waist circumference had a slightly lower significant association with relBMI at follow-up (r=0.74; P<0.001). Fat mass percentage, tricipital and subscapular skinfolds, and the sum of the four skinfolds showed lower levels of association (r=0.70, 0.62, 0.61 and 0.63, respectively; P<0.001; Table 2).

RelBMI at baseline showed a significant correlation with waist circumference (r=0.89; P<0.001). However, a partial correlation analysis showed that waist circumference had a significant independent association with relBMI at follow-up, when basal relBMI was controlled for (r=0.23, P<0.02). Therefore, we ran a multiple regression analysis with relBMI at follow-up as the dependent variable, using baseline relBMI, waist circumference, mother's BMI, age and gender as independent variables. The model included just waist and age and rejected gender, relBMI and mother's BMI, explaining 64% of final relBMI interindividual variability (P<0.001; Table 3). Father's BMI and Tanner's stage were not included among the covariates because they did not correlate with relBMI at follow-up (r=0.18, P=NS and r=0.21, P=NS, respectively).

We ran four other multiple regression analyses using relBMI at follow-up as the dependent variable, and triceps and subscapular skinfolds, sum of four skinfolds and fat mass percentage as independent variables. They accounted for approximately 50%, 36%, 41% and 61% of the variation of relBMI at follow-up, respectively (P<0.001). All the models were adjusted for age, gender and mother's BMI (Table 4).

Finally, a multivariate logistic regression analysis was performed on two groups of children, selected on the basis of their relBMI at follow-up: group A with relative BMI£120% and group B, with relative BMI>120%. Waist circumference, age and the mother's BMI were included in the final equation, whereas gender fell out of the model (Table 5).

Discussion

The results of our study show that waist circumference measured at the age of 8 y is the best predictor of overweight (relBMI) at age 12 y. In fact, in a multiple regression analysis, waist circumference (adjusted for age, gender, mother's BMI and relBMI at baseline) was the adiposity index with the highest association with relBMI at follow-up (r²=0.64; P<0.001). Mother's BMI did not further explain the children's relBMI at follow-up. Several anthropometric measures are available to evaluate total adiposity for epidemiological purposes and clinical application.²⁰ The BMI was proposed as an index of adiposity in children and adults.¹³ There are several advantages to using the BMI. First, BMI has a good level of correlation with the percentage of body fat measured by DXA in both boys and girls.^21,22,23 Moreover, weight and height, from which it is obtained, are easily measured with a good level of reliability by both an inter- and intra-operator.²⁴ Finally, it is not costly or invasive. However, some BMI limitations have been mentioned. In particular, a comparison of BMI and fat mass obtained by DXA in 979 children showed that BMI, although it may describe the adiposity characteristics of a healthy pediatric population, is a poor predictor of fatness for the individual child, with a standard error for relative adiposity of 4.7-7.3%.¹³ Moreover, the level of association between adiposity and BMI is affected by race, as demonstrated by the comparison of the BMIs of subjects from different ethnic groups having comparable body fat composition.^15,25 Finally, the adjustment for height does not completely eliminate the stature effect so that the use of BMI in a clinical setting requires additional measures to confirm the diagnosis of obesity in children.¹⁴

Other indices of adiposity were proposed for children, in particular skinfolds and circumferences. Triceps and subscapular skinfolds and waist circumference are the most frequently used. Although skinfold thicknesses are well correlated with percentage body fat, inter- and/or intra-individual measurements and evaluation of overweight subjects are difficult to reproduce. On the other hand, an estimation of fat mass based on anthropometry involves the development of prediction models where anthropometric measures (ie sum of skinfolds) are related to body fat mass. This issue shows that we need to use specific equations previously cross-validated for our population. When associated with the relative difficulty of the measurement per se, enthusiasm for the use of skinfolds decreased. On the contrary, the measurement of waist circumference is easier, economical and offers more accurate results for the pediatrician.²⁶ In adults, waist circumference is often used as a marker of intra-abdominal adipose tissue (IAAT), measured by imaging techniques.²⁷ The importance of IAAT is its high correlation with cardiovascular risk factors and metabolic disorders (insulin resistance, type 2 diabetes and dyslipidemia).^28,29,30 This relationship is not as strong in prepubertal children and in adolescents as it is in adults.³¹ Waist circumference is not a good marker to discriminate between IAAT and subcutaneous abdominal adipose tissue in adults as it is in children. Moreover, few data are available on abdominal body fat distribution assessed by CT or MRI in children and they offer different results.^32,33 In particular, some studies suggest that IAAT in children increases in proportion to overall fatness,³² as measured in adults;³⁴ whereas other studies show that obese children tend to accumulate subcutaneous fat and not IAAT.³³ However, in spite of the important effect of where body fat is located on the development of metabolic disturbances, it has been demonstrated that waist circumference may help to identify children and adolescents with an adverse concentration of circulating lipids and other risk factors.³⁵ Finally, previous studies have shown that waist circumference is a good index of total adiposity in prepubertal children as well as trunkal fat.^27,36

Several studies have explored the level of association between adiposity and cardiovascular risk factors or insulin sensitivity in children.^37,38 On the contrary, no longitudinal studies have yet explored the level of association between waist circumference measured in young children and overweight at puberty. The results of this study, the first which has approached this investigation, show that waist measured at the age of 8 y, adjusted for age, is the best predictor of overweight at the age of 12 y. These findings are emphasised by the results of the multiple logistic regression analysis, which included waist circumference, adjusted for age, mother's BMI and relBMI measured at baseline as independent variables in the final equation. In particular, each centimeter increase of waist circumference at the age of 8 y doubles the risk of having a relBMI greater than 120% at the age of 12 y. The duration of this longitudinal study was 4 y and this is not enough to evaluate the predictive value of waist circumference measured in young childhood on obesity in adulthood. However, several studies have shown that childhood obesity tends to track into adulthood.¹² At the age of 10 y, an obese child has a 22-fold higher risk of being obese as an adult than a nonobese child.¹² Moreover, Dwyer and colleagues showed that children who where overweight or simply fat at baseline were more likely to be overweight and overfat at follow-up and to have more cardiovascular risk factors than their peers.³⁹ On the basis of this evidence, it is important to promote longitudinal studies using methods to measure fatness which are simple to perform and easy to reproduce, such as waist circumference, in order to identify early in life further signs of the risk of maintaining excess adiposity, as well as its metabolic complications.

Acknowledgements

This study was supported by the National Research Council, Rome, Italy, contract no. 96.03441.CT04 and by Nestlè Italiana Spa, Italy.

1 WHO. Obesity: preventing and managing the global epidemic. Report of a WHO Consultation on Obesity, Geneva, 3-5 June 1997. World Health Organization: Geneva, 1998, 33-35.

2 Maffeis C, Schutz Y, Piccoli R et al. Prevalence of obesity in children in north-east Italy. Int J Obes Relat Metab Disord 1993; 17: 287-294, MEDLINE

3 Troiano RP, Flegal KM. Overweight children and adolescents: description, epidemiology and demographics. Pediatrics 1998; 101: 497-504,

4 Livingstone MBE. Epidemiology of childhood obesity in Europe. Eur J Pediat 2000; 159: (Suppl 1) S14-S34,

5 Serdula MK, Ivery D, Coates RJ et al. Do obese children become obese adults? A review of the literature. Prev Med 1993; 22: 167-177, Article MEDLINE

6 Dietz WH. Health consequences of obesity in youth: childhood predictors of adult disease. Pediatrics 1998; 101: 518-525,

7 Must A, Jacques PF, Dallal GE et al. Long term morbidity and mortality of overweight adolescents: a follow-up of the Harvard Growth Study of 1922 to 1935. New Engl J Med 1992; 327: 1350-1355, MEDLINE

8 Friedman MA, Brownell KD. Psychological correlates of obesity: moving to the next research generation. Psychol Bull 1995; 117: 3-20, MEDLINE

9 Quesenberry CP, Caan B, Jacobson A. Obesity, health service use, and health care cost among members of a Health Maintenance Organization. Arch Intern Med 1998; 158: 466-472, MEDLINE

10 Wolf AA, Colditz GA. Current estimates of the economic cost of obesity in the United States. Obes Res 1998; 6: 97-106, MEDLINE

11 Wadden TA. The management of obesity: from past failure to future attainment. In: Guy-Grand B, Ailhaud G (eds). Progress in obesity research: 8. John Libbey: New York, 1999, 713-719.

12 Whitaker RC, Wright JA, Pepe MS et al. Predicting obesity in young adulthood from childhood and parental obesity. New Engl J Med 1997; 337: 869-873, MEDLINE

13 Dietz WH, Bellizzi MC. Introduction: the use of body mass index to assess obesity in children. Am J Clin Nutr 1999; 70: (Suppl) 123S-125S,

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

15 Franklin MF. Comparison of weight and height relations in boys from 4 countries. Am J Clin Nutr 1999; 70: (Suppl) 157S-162S, MEDLINE

16 Wang J, Thornton JC, Russell M et al. Asians have lower body mass index (BMI) but higher percent body fat than do whites: comparison of anthropometric measurements. Am J Clin Nutr 1994; 60: 23-28, MEDLINE

17 Stunkard AJ, Albaum JM. The accuracy of self-reported weight. Am J Clin Nutr 1981; 34: 1953-1959,

18 Tanner JM. Growth at adolescence 2nd edn. Blackwell: Oxford, 1962,

19 Rolland-Cacherà MF, Sempè M, Guillaud-Bataille M et al. Adiposity indices in children. Am J Clin Nutr 1982; 36: 178-184, MEDLINE

20 Lohman TG. Applicability of body composition techniques and constants for children and youth. Exerc Sport Sci Rev 1986; 14: 325-357, MEDLINE

21 Goran MI, Driscoll P, Jhonson R et al. Cross-calibration of body-composition techniques against dual-energy X-ray absorptometry in young children. Am J Clin Nutr 1996; 63: 299-305, MEDLINE

22 Gutin B, Litaker M, Islam S et al. Body-composition measurement in 9-11-year-old children by dual-energy X-ray absorptiometry, skinfold thickness measurements, and bioimpedance analysis. Am J Clin Nutr 1996; 63: 287-292, MEDLINE

23 Pietrobelli A, Faith M, Allison DB et al. Body mass index as a measure of adiposity among children and adolescents: a validation study. J Pediatr 1998; 132: 204-210, MEDLINE

24 Marks GC, Habicht JP, Mueller WH. Reliability, dependability, and precision of anthropometric measurements. The Second National Health and Nutrition Examination Survey 1976-1980. Am J Epidemiol 1989; 130: 578-587, MEDLINE

25 Daniels SR, Khoury PR, Morrison JA. The utility of body mass index as a measure of body fatness in children and adolescents: differences by race and gender. Pediatrics 1997; 99: 804-807, MEDLINE

26 Mueller WH, Kaplowitz HJ. The precision of anthropometric assessment of body fat distribution in children. Ann Hum Biol 1994; 21: 267-274, MEDLINE

27 Goran MI, Gower BA. Relation between visceral fat and disease risk in children and adolescents. Am J Clin Nutr 1999; 70: (Suppl) 149S-156S,

28 Pi-Sunyer FX. Health implications of obesity. Am J Clin Nutr 1991; 53: (Suppl) 1595S-1603S, MEDLINE

29 Goran MI, Gower BA. Abdominal obesity and cardiovascular risk in children. Coron Art Dis 1998; 9: 483-487,

30 Caprio S, Hyman LD, McCarthy S et al. Fat distribution and cardiovascular risk factors in obese adolescent girls: importance of the intraabdominal fat depot. Am J Clin Nutr 1996; 64: 12-17, MEDLINE

31 Maffeis C, Pietrobelli A, Grezzani A, Provera S, Tatò L. Waist circumference and cardiovascular risk factors in prepubertal children. Obes Res 2001; (in press).

32 Goran MI, Kaskoun MC, Shuman WP. Intra-abdominal adipose tissue in young children. Int J Obes Relat Metab Disord 1995; 19: 279-283, MEDLINE

33 Fox K, Peters D, Armstrong N et al. Abdominal fat deposition in 11-y-old children. Int J Obes Relat Metab Disord 1992; 16: 443-449, MEDLINE

34 Lemieux S, Prud'homme D, Bouchard C et al. Sex differences in the relation of visceral adipose tissue accumulation to total body fatness. Am J Clin Nutr 1993; 58: 463-467, MEDLINE

35 Freedman DS, Serdula MK, Srinivasan SR, Berenson GS. Relation of circumferences and skinfold thicknesses to lipid and insulin concentrations in children and adolescents: the Bogalusa Heart Study. Am J Clin Nutr 1999; 69: 308-317, MEDLINE

36 Taylor RW, Jones IE, Williams SM, Goulding A. Evaluation of waist circumference, waist-to-hip ratio, and the conicity index as screening tools for high trunk fat mass, as measured by dual-energy X-ray absorptiometry, in children aged 3-19 y. Am J Clin Nutr 2000; 72: 490-495, MEDLINE

37 Bandini LG, Schoeller DA. Total body fat distribution and glucose tolerance in adolescents. Am J Clin Nutr 1986; 43: 696,

38 Freedman DS, Srinivasan SR, Burke GL et al. Relation of body fat distribution to hyperinsulinemia in children and adolescents: the Bogalusa Heart Study. Am J Clin Nutr 1987; 46: 403-410, MEDLINE

39 Dwyer JT, Stone EJ, Yang M et al. Predictors of overweight and overfatness in a multiethnic pediatric population. Child and Adolescent Trial for Cardiovascular Health Collaborative Research Group. Am J Clin Nutr 1998; 67: 602-610, MEDLINE

Figure 1 Relationship between waist circumference and relBMI at follow-up in male and female children.

Table 1 Physical characteristics of the children at baseline and 4 y later at follow-up. Data are shown as mean (s.d.)

Table 2 Correlation matrix of the anthropometric parameters and relBMI at follow-up after 4 y in 112 children

Table 3 Multiple regression model: final equation. Dependent variable, relBMI at follow-up; independent variables, age and waist circumference

Table 4 Multiple regression models. Dependent variable, relBMI at follow-up; independent variables, age, gender, mother's BMI, percentage fat mass and skinfolds

Table 5 Multivariate logistic regression analysis. RelBMI at follow-up was used as grouping variables and age, gender, mother's BMI and waist circumference as independent variables