European Journal of Pediatrics

, Volume 170, Issue 6, pp 731–739

Comparison of several anthropometric indices with insulin resistance proxy measures among European adolescents: The Helena Study

Authors

  • Katerina Kondaki
    • Department of Nutrition and DieteticsHarokopio University
    • Department of Public HealthGhent University
  • Evangelia Grammatikaki
    • Department of Nutrition and DieteticsHarokopio University
    • Department of Public HealthGhent University
  • David Jiménez Pavón
    • Department of Physiology, School of MedicineUniversity of Granada
  • Yannis Manios
    • Department of Nutrition and DieteticsHarokopio University
  • Marcela González-Gross
    • Department of Health and Human Performance, Facultad de Ciencias de la Actividad Física y del DeporteUniversidad Politécnica de Madrid
  • Michael Sjöstrom
    • Karolinska Institutet
  • Frédéric Gottrand
    • Faculté de médecineUniversity of Lille 2
  • Dénes Molnar
    • Department of PediatricsUniversity of Pécs
  • Luis A. Moreno
    • Escuela Universitaria de Ciencias de la SaludUniversidad de Zaragoza
  • Anthony Kafatos
    • Preventive Medicine and Nutrition UnitUniversity of Crete School of Medicine
  • Chantal Gilbert
    • Department of Consumer and Sensory SciencesCampden & Chorleywood Food Research Association
  • Mathilde Kersting
    • Research Institute of Child NutritionRheinische Friedrich-Wilhelms-Universität Bonn
    • Department of Public HealthGhent University
Original Paper

DOI: 10.1007/s00431-010-1322-4

Cite this article as:
Kondaki, K., Grammatikaki, E., Pavón, D.J. et al. Eur J Pediatr (2011) 170: 731. doi:10.1007/s00431-010-1322-4

Abstract

The aim of the current study was to compare the association of several anthropometric indices, with insulin resistance (IR) proxy measures in European adolescents. The present study comprises 1,097 adolescents aged 12.5–17.5 from ten European cities participating in the HELENA study. Weight, height, waist circumference (WC) and hip circumference, skinfolds thickness, fat mass (FM), fasting plasma glucose (GF) and serum insulin (IF) levels were measured. HOMA (as indicator of IR body mass index (BMI), waist to hip ratio (WHR) and waist to height ratio (WHtR) were calculated. IF and HOMA were statistically significantly related to BMI, WC, skinfold sum, WHtR, WHR and FM. BMI, WC, WHtR, skinfold sum and FM displayed similar correlation with IF and HOMA as opposed to WHR where lower correlation with IR indices was detected in the overall sample. Similar results were found for boys, girls and underweight/normal weight adolescents. On the other hand, WC and WHtR were found to be more strongly associated with IR proxy measures compared to the rest of anthropometric indices among overweight/obese subjects. Based on the current findings, WC and WHtR could be used, alternatively, to identify the overweight/obese adolescent at risk for developing IR. In addition, all aforementioned anthropometric indices, except WHR, could be used among the underweight/normal weight adolescents.

Keywords

Body mass indexSerum insulinWaist circumferenceWaist-to-height ratioWaist-to-hip ratioSkinfold sum

Abbreviations

WC

Waist circumference

IR

Insulin resistance

WHR

Waist-to-hip ratio

WHtR

Waist-to-height ratio

FM

Fatt mass

BMI

Body mass index

MVPA

Moderate to vigorous physical activity

PA

Physical activity

CVD

Cardiovascular disease

IF

Fasting serum insulin

GF

Fasting serum glucose

Introduction

The prevalence of obesity in children and adolescents is increasing worldwide and has reached epidemic proportions [6]. Moreover, several studies conducted in children and adolescents have reported that obesity is associated with increased prevalence of cardiovascular (CVD) risk factors and metabolic disorders (i.e. hypertension, hyperlipidemia, abnormal glucose tolerance, insulin resistance (IR) and type 2 diabetes) [4, 7, 8, 10, 33, 36]. Particularly, it has been shown that visceral adiposity is independently associated with CVD risk factors compared to total adiposity [2, 12, 15]. In addition, there are limited data regarding the association between visceral adiposity and IR development [1, 22, 26].

Therefore, the need for a measure of central obesity has emerged. Although, several measures have been proposed as indicators of visceral adiposity (i.e. waist circumference-to-hip circumference ratio (WHR), waist circumference (WC) and waist circumference-to-height ratio (WHtR)), none of these has been accepted as a gold standard measure [19, 20, 32]. Moreover, none of the aforementioned obesity-related indicators is considered as the best one for predicting CVD risk, since the results of previous studies are inconsistent. For example, in studies conducted among children in Cyprus [35], Japan [17] and China [42], both WC and WHtR were found to be better predictors of children’s lipidemic profile and blood pressure levels compared with body mass index (BMI), with a small superiority of WHtR. In addition, Kahn et al. using data from a cross-sectional survey in a US population based sample, reported that WHtR identifies better than BMI youths with CVD risk factors [20].

Concerning the associations of various anthropometric indices with IR proxy measures and plasma glucose levels, limited data are available [18, 24, 25, 29, 34, 42]. In particular, Yan et al. have revealed that WC, WHtR and BMI are similarly correlated with fasting serum insulin (IF), and HOMA [42]. Similar results were declared by Misra et al. who compared BMI, WC and subscapular skinfold thickness with fasting hyperinsulinemia [29]. Finally, Manios et al., at “The Children Study” revealed that BMI, WC and WHtR were significantly related to IR proxy measures as opposed to WHR [25]

Therefore, the aim of this study is to identify which obesity-related measure is stronger associated with glucose metabolism and IR proxy measures in European adolescents. In this context, the associations of BMI, WC, WHR, WHtR, skinfold sum and fat mass (FM), with IF levels and HOMA were assessed.

Methods

Research design and methods

The HELENA-CSS is a multicenter investigation carried out in ten European cities: Athens (Greece), Dortmund (Germany), Ghent (Belgium), Heraklion (Greece), Lille (France), Pécs (Hungary), Rome (Italy), Stockholm (Sweden), Vienna (Austria) and Zaragoza (Spain). The main aim of the HELENA-CSS was to obtain reliable and comparable data on a broad battery of relevant nutrition and health-related parameters such as dietary intake, anthropometry, physical activity (PA), fitness, haematological and biochemical indices [30]. Data collection from the HELENA-CSS took place in 2006–2007.

All participants were recruited at schools and met the general HELENA-CSS inclusion criteria: age range 12.5–17.5 years, not participating simultaneously in another clinical trial and be free of any acute infection lasting less than 1 week before the inclusion, and having information on weight and height [28]. The present work is confined to a subset of 1,097 adolescents (583 females and 514 males) with complete data on glucose and insulin levels. Ethics committees from each country approved the HELENA-CSS protocols [3].

Physical examination

The anthropometric methods followed in the HELENA-CSS study were described in detail by Nagy et al. [31]. Weight was measured in underwear and without shoes with an electronic scale (Type SECA 861) to the nearest 0.05 kg, and height was measured barefoot in the Frankfort plane with a telescopic height measuring instrument (Type SECA 225) to the nearest 0.1 cm. BMI was calculated as body weight (kg) divided by the height squared (m2). Age- and sex-standardized BMI cut-off points according to the International Obesity Task Force were used to define normal weight, overweight, and obesity [11]. Skinfold thickness was measured to the nearest 0.2 mm in triplicate in the right side at biceps, triceps, subscapular, suprailiac, thigh, and medial calf with a Holtain Caliper (Crymmych, UK). The sum of six skinfold thickness was used as an indicator of total body fat. WC was measured in triplicate at the midpoint between the lowest rib and the iliac crest with an anthropometric tape SECA 200 [23] and was used as a surrogate of central body fat. Hip circumference was measured at the midpoint of hips with an anthropometric tape SECA 200. WHR and WHtR were calculated. FM was calculated from body weight deducting fat-free mass. Fat-free mass was assessed by bioelectric impedance (BIA) measurements, a classical tetrapolar technique by means of BIA 101 AKERN SRL. Standard instructions for BIA measurements were followed. Pubertal stage was recorded by a researcher of the same sex as the child, after brief observation according to Tanner and Whitehouse [38].

Blood samples

Serum concentrations of glucose (GF) and IF were measured after an overnight fast. The homeostasis model assessment (HOMA) was calculated as IF (μlU/mL) × GF (mmol/l)/22.5 (to convert IF values in μlU/mL to pmol/l multiply by 6.945) [27]. A detailed description of the blood analysis has been reported elsewhere [14]. These data were available in a subsample (n = 546).

Physical activity assessment

More details regarding PA assessment can be found in Hagstromer et al. [16]. In short, PA was measured using accelerometers (Actigraph MTI, model GT1M, Manufacturing Technology Inc., Fort Walton Beach, FL, USA) placed on each individual for several days. The monitor was secured underneath clothing at the lower back using an elastic belt and was worn for seven consecutive days. Adolescents were also instructed to wear the accelerometer during all time awake and only to remove it during water-based activities. It was initialized as described by the manufacturer and a 15-s epoch was used. The sum of accelerations was transformed into counts. Low PA was considered when the mean of time spent in activity was from 500 to 1,999 counts. Moderate PA was considered when the mean of time spent in activity was from 2,000 to 3,999 counts. High PA was considered when the mean of time spent in activity was more than 4,000 counts. The moderate to vigorous physical activity stands for the time spent on at least 2,000 counts or more for PA per day.

Dietary assessment

Food consumption

Population mean for intakes and distributions for participants in different European countries were collected with the 24-h recall method. Furthermore, usual intake was estimated by statistical modelling techniques using two non-consecutive 24-h recalls [5]. Following these recommendations and regarding the challenges to measure food consumption in adolescents, a computer-assisted self-administered tool (HELENA-DIAT, previously known as YANA-C), attractive for adolescents, was adapted for dietary assessment in the Helena Study [40] (Appendix). To calculate energy and nutrient intakes, food intake information were linked to local food composition databases [21].

Statistical analysis

Normally distributed continuous variables are expressed as mean values ± standard deviation and skewed variables are reported as median (25th and 75th percentiles). Normality of distribution was evaluated through the Kolmogorov–Smirnov test. IF and HOMA were non-normally distributed and were log-transformed. Correlations between anthropometric variables and GF were tested by the use of Pearson’s correlation coefficient, while the correlation between anthropometric indices and IF, HOMA were tested by the use of Spearman’s correlation. For each one of the following indices: IF, HOMA, (as dependent variables), six multiple regression models were applied. In each model, one of the six anthropometric indices as well as age, sex, tanner stage, physical activity, total energy, simple carbohydrate and fat intake were included as potential confounders. The adjusted R2 of each model (variance explained by each one model) was calculated in order to determine which of the anthropometric index is stronger associated with proxy IR measures even after controlling for potential confounders. Different models were performed for each anthropometric marker in order to avoid the problems of multi-colinearity, since all these anthropometric markers were strongly correlated. Stratified analysis by gender and by weight status were conducted. P values <0.05 from two-sided hypotheses are considered as statistically significant. All statistical calculations were performed using STATA 8.0

Results

Table 1 shows descriptive characteristics of the anthropometric characteristics, GF levels, IF levels, HOMA, PA, total energy intake, carbohydrate intake, and fat intake for the total study population and by gender. Among 546 adolescents, 5.5% were underweight, 71.4% were normally weighted, 16.7% were overweight and 6.4% were obese. The prevalence of obesity was higher among boys compared with girls although these results did not reach statistical significance (8.8% and 4.4%, respectively, p = 0.160).
Table 1

Characteristics of study population

 

Total (N = 1,089)

Males (N = 514)

Females (N = 583)

Age (years)a

14.5 (13.5, 15.5)

14.5 (13.5, 15.3)

14.4 (13.5, 15.4)

Anthropometric measuresb

Weight (kg)

59 ± 13

63 ± 14

56 ± 10*

Height (cm)

165 ± 9

170 ± 10

162 ± 7*

BMI (kg/m2)

21.4 ± 3.6

21.4 ± 4.0

21.4 ± 3.4

Waist circumference (cm)

72.3 ± 8.6

74.6 ± 8.9

70.8 ± 7.8*

Waist/hip ratio

0.79 ± 0.06

0.82 ± 0.05

0.76 ± 0.05*

Waist/height ratio

0.44 ± 0.05

0.44 ± 0.05

0.44 ± 0.05

Skinfold sum (cm)

68.2 ± 32.6

61.8 ± 31.9

84.3 ± 29.4*

Total fat mass (kg)

12.5 ± 7.1

10.3 ± 6.9

14.5 ± 6.8*

Tanner stage

1

0.7%

1.5%

0.0% *

2

5.4%

7.2%

3.9%

3

18.9%

18.5%

19.2%

4

44.0%

42.4%

45.5%

5

31.0%

30.5%

31.4%

MVPAa

55.0 (40.8, 70.9)

69.5 (53.3, 83.7)

48.8 (35.4, 62.0)*

Biochemical measurements

Glucose (mg/dl)b

91 ± 7

93 ± 7

89 ± 7

Insulin (μIU/mL)a

8.57 (6.12, 11.91)

8.14 (5.93, 11.59)

9.00 (6.46, 12.30)*

HOMAa

0.62 (0.44, 0.87)

0.60 (0.43, 0.87)

0.65 (0.45, 0.88)*

Energy and macronutrients intake

Total energy intake (kcal/day)a

1925 (1508, 2520)

2379 (1814, 3036)

1748 (1403, 2,180)*

Total carbohydrate intake (g/day)a

238 (187, 314)

278 (218, 402)

222 (169, 274)*

Total fat intake (g/day)a

75 (42, 102)

86 (64, 128)

65 (48, 91)*

BMI body mass index, MVPA moderate to vigorous physical activity

aData are presented as median (25th and 75th percentiles)

bData are presented as mean ± standard deviation for continuous variables

*p value < 0.05 for comparison between two genders

Univariate correlation coefficients between anthropometric indices and glycemic control indices are presented in Table 2 for the total sample and by gender and in Table 3 for underweight/normal weight and overweight/obese, separately. No anthropometric index was significantly correlated with GF, while IF and HOMA were significantly positively related to all anthropometric measures both in boys and girls and underweight/normal weight and overweight/obese. IF and HOMA were similarly correlated with BMI, WC, skinfolds, WHtR and FM, as opposed to WHR in the overall sample, boys and girls. The correlation coefficients were found to be higher among overweight/obese compared with underweight/normal weight adolescents (Table 3). Moreover, it was found that FM and skinfold sum were more strongly associated with IR measures (IF and HOMA) among overweight/obese and underweight/normal weight adolescents, respectively. The correlation coefficient between BMI and IF and HOMA was similar to those observed for skinfold sum and FM, while the correlation of WC and WHtR was lower both in overweight/obese and underweight/normal weight adolescents. Finally, WHR was not significantly correlated with IF and HOMA both in underweight/normal weight and overweight/obese.
Table 2

Pearson’s or Spearman’s correlation coefficients between insulin/insulin resistance and anthropometric indices for the total sample and by gender

Anthropometric measurements

Glucose

Insulina

HOMAa

BMI (kg/m2)

Total sample

0.028

0.349**

0.339**

Boys

0.100*

0.405**

0.401**

Girls

−0.059

0.286**

0.264**

Waist circumference (cm)

Total sample

0.076

0.313**

0.311**

Boys

0.071

0.367**

0.361**

Girls

−0.028

0.308**

0.290**

Waist/hip ratio

Total sample

0.101*

0.118**

0.128**

Boys

−0.071

0.138**

0.124**

Girls

0.010

0.223**

0.216**

Waist/height ratio

Total sample

0.038

0.334**

0.326**

Boys

0.079

0.364**

0.359**

Girls

−0.007

0.306**

0.291**

Skinfold sum (cm)

Total sample

−0.027

0.353**

0.335**

Boys

0.126*

0.398**

0.398**

Girls

−0.015

0.288**

0.272**

Total fat mass (kg)

Total sample

−0.060

0.343**

0.321**

Boys

0.085

0.374**

0.371**

Girls

−0.061

0.299**

0.276**

BMI body mass index

*p ≤ 0.05; **p < 0.01

aSpearman’s correlation

Table 3

Pearson or Spearman correlation coefficients between insulin/insulin resistance and anthropometric indices by weight status

Anthropometric measurements

Glucose

Insulina

HOMAa

BMI (kg/m2)

Underweight/normal weight

−0.065

0.170**

0.154**

Overweight/obese

0.170

0.437**

0.440**

Waist circumference (cm)

Underweight/normal weight

0.043

0.138**

0.140**

Overweight/obese

0.185*

0.367**

0.379**

Waist/hip ratio

Underweight/normal weight

0.118*

−0.010

0.008

Overweight/obese

0.118

0.057

0.072

Waist/height ratio

Underweight/normal weight

0.015

0.129**

0.127*

Overweight/obese

0.136

0.369**

0.374**

Skinfold Sum (cm)

Underweight/normal weight

−0.063

0.255**

0.239**

Overweight/obese

0.121

0.406**

0.408**

Total fat mass (kg)

Underweight/normal weight

−0.151

0.208**

0.179**

Overweight/obese

−0.040

0.496**

0.468**

BMI body mass index

*p ≤ 0.05; ** p < 0.01

aSpearman’s correlation

The results of the multiple linear regression for IF and HOMA (as dependent variables) are presented in Tables 4 and 5. The results confirmed the findings of the univariate analysis for the total sample and for boys and girls, separately. In particular, it was found that the models including skinfold sum, FM, WC, BMI and WHtR explained higher proportion of variance of IF and HOMA compared to that explained by the model with WHR. WHR and skinfold sum seem to have higher predictability for both HOMA and IF in girls compared to boys. Stratified analyses by weight status, revealed that among overweight/obese, WC and WHtR seem to have higher predictability for IR proxy measures compared with the rest of the anthropometric measures. On the other hand, the percent of variance of IR proxy measures explained by all anthropometric measures is similar among underweight/normal weight.
Table 4

Results from multiple linear regression using as dependent variable the log-transformed values of HOMA

Parameters

β-coefficients ± SE

Adjusted R2

Total sample

BMI (kg/m2)

0.070 ± 0.007

0.224

Waist circumference (cm)

0.030 ± 0.003

0.267

Waist/hip ratio (cm)

2.251 ± 0.440

0.119

Waist/height ratio (cm)

5.198 ± 0.477

0.257

Total fat mass(kg)

0.032 ± 0.003

0.217

Skinfold sum (cm)

0.007 ± 0.001

0.223

Boys

BMI (kg/m2)

0.086 ± 0.010

0.274

Waist circumference (cm)

0.034 ± 0.005

0.237

Waist/hip ratio (cm)

2.564 ± 0.861

0.061

Waist/height ratio (cm)

6.606 ± 0.828

0.259

Total fat mass(kg)

0.042 ± 0.006

0.221

Skinfold sum (cm)

0.010 ± 0.001

0.275

Girls

BMI (kg/m2)

0.057 ± 0.009

0.225

Waist circumference (cm)

0.028 ± 0.004

0.265

Waist/hip ratio (cm)

1.973 ± 0.543

0.154

Waist/height ratio (cm)

4.636 ± 0.642

0.267

Total fat mass (kg)

0.026 ± 0.005

0.215

Skinfold sum (cm)

0.005 ± 0.001

0.193

Underweight/normal weight

BMI (kg/m2)

0.050 ± 0.018

0.158

Waist circumference (cm)

0.018 ± 0.007

0.154

Waist/hip ratio (cm)

1.445 ± 0.829

0.142

Waist/height ratio (cm)

3.885 ± 1.177

0.173

Total fat mass (kg)

0.022 ± 0.009

0.147

Skinfold sum (cm)

0.005 ± 0.002

0.175

Overweight/obese

BMI (kg/m2)

0.069 ± 0.027

0.273

Waist circumference (cm)

0.033 ± 0.007

0.443

Waist/hip ratio (cm)

0.663 ± 0.758

0.168

Waist/height ratio (cm)

4.956 ± 1.545

0.332

Total fat mass (Kg)

0.030 ± 0.010

0.310

Skinfold sum (cm)

0.003 ± 0.003

0.192

After controlling for age, sex, Tanner stage, total energy intake, simple carbohydrate and fat intake, and physical activity

*p ≤ 0.05; **p < 0.01

Table 5

Results from multiple linear regression using as dependent variable the log-transformed values of insulin

Parameters

β-coefficients ± SE

Adjusted R2

Total sample

BMI (kg/m2)

0.069 ± 0.006

0.259

Waist circumference (cm)

0.030 ± 0.003

0.267

Waist/hip ratio (cm)

2.251 ± 0.440

0.125

Waist/height ratio (cm)

5.198 ± 0.477

0.269

Total fat mass(kg)

0.032 ± 0.003

0.229

Skinfold sum (cm)

0.007 ± 0.001

0.230

Boys

BMI (kg/m2)

0.083 ± 0.010

0.281

Waist circumference (cm)

0.033 ± 0.004

0.245

Waist/hip ratio (cm)

2.727 ± 0.816

0.069

Waist/height ratio (cm)

6.487 ± 0.781

0.272

Total fat mass(kg)

0.041 ± 0.006

0.223

Skinfold sum(cm)

0.009 ± 0.001

0.279

Girls

BMI (kg/m2)

0.058 ± 0.009

0.231

Waist circumference (cm)

0.028 ± 0.004

0.269

Waist/hip ratio (cm)

1.932 ± 0.523

0.152

Waist/height ratio (cm)

4.585 ± 0.620

0.269

Total fat mass(kg)

0.027 ± 0.004

0.224

Skinfold sum(cm)

0.005 ± 0.001

0.193

Underweight/normal weight

BMI (kg/m2)

0.052 ± 0.018

0.166

Waist circumference (cm)

0.018 ± 0.007

0.158

Waist/hip ratio (cm)

1.436 ± 0.803

0.145

Waist/height ratio (cm)

3.845 ± 1.140

0.178

Total fat mass(kg)

0.023 ± 0.009

0.155

Skinfold sum(cm)

0.005 ± 0.002

0.179

Overweight/obese

BMI (kg/m2)

0.073 ± 0.026

0.278

Waist circumference (cm)

0.033 ± 0.007

0.465

Waist/hip ratio (cm)

0.704 ± 0.719

0.145

Waist/height ratio (cm)

4.904 ± 1.453

0.426

Total fat mass(kg)

0.023 ± 0.009

0.155

Skinfold sum(cm)

0.003 ± 0.002

0.173

After controlling for age, sex, Tanner stage, total energy intake, simple carbohydrate and fat intake, and physical activity

*p ≤ 0.05; **p < 0.01

Discussion

The purpose of this study was to examine which of the following anthropometric and body composition indices, i.e. WHtR, WC, WHR, skinfold sum, BMI and FM, were more strongly related to GF levels and IR proxy measures among European adolescents. Our results showed that, BMI, WC, WHtR, skinfold sum and FM were more strongly associated with IR proxy measures, compared to WHR in the overall sample and underweight/normal weight adolescents. However, the association of WC and WHtR with IR proxy measures were stronger compare to the rest of anthropometric indices, among overweight/obese adolescents. Neither anthropometric indices nor FM were significantly correlated with GF. On the other hand, WHR and skinfold sum were explicitly more strongly associated with proxy IR measures in female rather than in male adolescents. That could be partly explained by the different fat storages between the two sexes and the different metabolic action of insulin in deposition of the abdominal and subscapular fat.

To our knowledge, there are some studies that have examined the associations between anthropometric measures of obesity and IR or GF levels in children or adolescents [18, 25, 29, 34, 39, 42]. Our findings are partially in line with those of other researchers. A study conducted in Chinese children highlighted WC as the best predictor of GF and reported similar correlation of WC, WHtR and BMI with IF and HOMA [42]. Manios et al. have also reported a stronger association of BMI, WHtR and WC with IR proxy measures compared to WHR ratio, among Greek pupils [25]. However, in accordance with our findings, Kahn et al. and Manios et al. have both demonstrated that no anthropometric index was a significant predictor of GF in youth [20, 25].

Similar results have been observed in adults, where a small number of studies have dealt with comparing the correlations between different anthropometric indices and IR. In these studies, both BMI and WC were identified as better predictors of IR compared to WHR [9, 13]. On the other hand, most studies have focused on identifying the best obesity-related indicator of type 2 diabetes and WC has emerged as the best predictor compared to BMI, WHR and WHtR [24, 28]. According to Yang L, a new significant indicator of IR is the neck circumference which is correlated with visceral adiposity in WHO grade III obesity [43].

The obesity-related metabolic factors in children, as in adults, (i.e. abnormal lipid or insulin concentration) are associated with an upper body or centralized deposition of excess body fat than with total obesity. In particular, the link between central adiposity and elevated fasting insulin could be attributed to the low hepatic insulin clearance due to high exposure of liver to free fatty acids from abdominal adiposity [37]. Furthermore, the increased intracellular fat accumulation, induced by high levels of circulating free fatty acids derived from the abdominal adipose tissue, could be another explanatory mechanism for the association observed in the present study between IR and central adiposity [41].

Still, there are some potential limitations in the current study. Firstly as a cross-sectional study, it is not appropriate for cause–effect relationships. Secondly, assessment of total body fat and central adiposity was indirect via the calculation of BMI and measurement of WC, respectively.

In conclusion, WC, BMI, WHtR, FM and skinfold sum are similarly correlated and more strongly associated with proxy IR measures compared to WHR, in underweight/normal weight adolescents, while WC and WHtR are more strongly related to IR proxy measures compared with the rest of anthropometric indices among overweight/obese. This implies that all these simple obesity indices could be used as a simple tool for identification of normal weight adolescents at risk for developing IR and type 2 diabetes in clinical practice, while only WC and WHtR among overweight/obese adolescents. Of course, further studies to investigate the association between several anthropometric indices and IR among adolescents are required, since data regarding this issue are limited.

Funding

The HELENA Study was carried out with the financial support of the European Community Sixth RTD Framework Programme (Contract FOOD-CT-2005-007034). The content of this article reflects only the authors’ views, and the European Community is not liable for any use that may be made of the information contained therein.

Conflicts of interest statement

The authors declare that there are no conflicts of interest.

Copyright information

© Springer-Verlag 2010