Population
The HELENA cross-sectional study is a population-based, multi-centre investigation of the nutritional and lifestyle status of adolescents, carried out in ten European cities (Vienna in Austria, Ghent in Belgium, Lille in France, Dortmund in Germany, Athens and Heraklion in Greece, Pecs in Hungary, Rome in Italy, Zaragoza in Spain and Stockholm in Sweden). Data were collected from October 2006 to December 2007. A detailed description of the HELENA study design and sampling procedure has been published elsewhere [
19,
20]. The study was approved by the ethical committees and performed following the ethical guidelines of the Declaration of Helsinki. All study participants and their parents provided a signed informed consent form.
The total HELENA population consisted of 3,528 eligible adolescents (52.3 % females) aged 12.5–17.5 years. For the current analyses, adolescents were included if data were available for two non-consecutive 24-h dietary recalls and the food frequency questionnaire (FFQ), resulting in 1,215 eligible subjects. Since for Heraklion, Pecs and Rome no full set of dietary data were available (i.e. two 24-h dietary recalls and a FFQ), subjects from these cities were excluded from the present study. In the HELENA population, non-participants did not differ in sex or age. For this paper, age and BMI did not differ between excluded and included cases, while less boys and more adolescents from high socio-economic status (SES) were included (p = 0.002 and p = 0.010, respectively). Data on glucose and lipid homoeostasis were only available in a subpopulation (N = 387); they did not differ in sex, BMI and SES but had a higher age (p < 0.001; mean 14.5 vs. 14.2 years) compared with the other eligible subjects.
Socio-demographic determinants
Collected demographic data included information on sex, age, city and SES by means of a standardised self-reported questionnaire. SES was examined by parental education and by the Family Affluence Scale (FAS). Education level of mother and father was reported as “lower education”, “higher secondary education” and “university education”. A modified version of the FAS developed by Currie et al. [
21] was used as a proxy of SES status; the scale is based on the concept of material conditions in the family. For the purposes of the HELENA study, the FAS was slightly modified by replacing the item on frequency of family holidays by Internet availability at home. The adolescents completed a questionnaire asking about the number of cars (0–3 depending on amount) and computers at home (0–3 depending on amount), having access to Internet at home (0 no, 1 yes), and whether the adolescent had his or her own room (0 no, 1 yes). Adolescents were scored from 0 (very low SES) to 8 (very high SES). For some analyses, countries were organised in geographical regions, as agreed on in the HELENA study: Greece, Italy and Spain represented the “Southern” region (Mediterranean); (2) Sweden and Belgium represented the “Northern” regions and (3) France, Germany and Austria represented the “West/central” region.
Body composition
The protocol used to collect anthropometric data has been previously described [
33]. Measurements were done while participants were barefoot and in underwear.
Weight was measured to the nearest 0.1 kg using electronic scales (Type SECA 861). Height was measured to the nearest 0.1 cm with the head aligned in the Frankfort plane using a telescopic height-measuring instrument (Type SECA 225). BMI was calculated as body weight in kilograms divided by the square of height in metres. In addition, BMI was adjusted for age and sex to give a BMI
z-score using the British Growth Reference Data from the Child Growth Foundation [
34], and overweight was classified following the International Obesity Task Force [
35].
The triceps and subscapular skinfold thickness and waist and hip circumferences were measured on the left side of the body. A Holtain calliper (Crymmych, UK) was used to measure skinfold thickness to the nearest 0.2 mm, and a non-elastic tape was used to measure circumference to the nearest 0.1 cm [
33].
The waist–hip ratio was calculated as marker of central adiposity. A puberty-adjusted body fat% was calculated from the two skinfolds using the Slaughter formulae as marker of overall adiposity [
36]. Pubertal status (stages I–V) was assessed by a medical doctor according to the scale developed by Tanner and Whitehouse [
37], based on breast development and pubic hair status in females and genital and pubic hair development in males.
Biomarkers for glucose and lipid profile
Blood samples were collected in a randomly selected subset of the total HELENA study population by venipuncture at school between 8 and 10 A.M. after a 10-h overnight fast. Blood was collected in tubes for serum (blood lipid profile) and heparinised tubes for plasma (insulin), centrifuged at 3,500 rpm, aliquoted and transported at 4–7 °C (for a maximum of 14 h) to the central laboratory in Bonn (Germany). Triglycerides (TG), total cholesterol (TC), high-density lipoprotein cholesterol (HDL), low-density lipoprotein cholesterol (LDL), very low-density lipoprotein cholesterol (VLDL), lipoprotein A (LpA) and glucose were measured using enzymatic methods (Dade Behring, Schwalbach, Germany) from fresh serum within 24 h of blood extraction. Heparin plasma was stored at −80 °C until assayed. Insulin concentrations were measured using an Immulite 2000 analyser (DPC Bierman GmbH, Bad Nauheim, Germany).
The following indices of lipid status were calculated: non-HDL/HDL, LDL/HDL, TC/HDL, LDL corrected for LpA, and TG/HDL. Insulin resistance was measured by calculating the homoeostasis model assessment (HOMA) index as follows: fasting insulin (μIU/mL) × fasting glucose (mg/dL)/405 [
38].
Statistics
Analyses were performed with PASW Statistical Program version 19.0 (SPSS Inc, IBM, IL, USA). Two-sided level of significance was set at p < 0.05. Non-normal data were transformed using the logarithmic or square root transformation to perform analyses, but data were back-transformed in the original units for representation. The regression analyses were all corrected for age, sex, SES, city and breakfast skipping. Estimated marginal means from the regression were used for the graphs. For significance levels between consecutive categories of RTEC consumption, the “repeated” contrasts were run with RTEC non-consumers as reference category. Socio-demographic differences (age, sex, city and SES) between different RTEC consumption groups were examined by means of a logistic regression with RTEC consumption as outcome. All covariates were included simultaneously to test their independent influence.
Differences in overall daily nutrient intakes and DQI were examined between RTEC consumer categories by linear regression corrected for age, sex, city, SES and breakfast skipping. Apart from energy, the following nutrients were studied: fat, protein, carbohydrates (also separate for glucose, fructose, galactose, sucrose, lactose, maltose and polysaccharides), fibre (also separately for water soluble and water insoluble fibres), minerals (calcium, iron, magnesium, phosphorus, potassium, sodium and zinc) and vitamins A, B1 (thiamine), B2 (riboflavin), B3 (niacin), B5 (pantothenic acid), B6 (pyridoxine), B7(biotin), B9 (folic acid), B12 (cobalamin), C, D, E, K.
The daily prevalence of milk/yoghurt and fruit intake in the different RTEC consumer categories was examined using
χ
2 statistics. The daily quantity of milk/yoghurt and fruit intake was examined with linear regression, excluding those that did not consume milk/yoghurt or fruit. Logistic regression was used to detect differences in fulfilling the daily intake recommendation for milk and fruit, based on the age- and sex-specific recommendations for European adolescents [
39].
Linear regression was used to test the association of RTEC consumption with body composition (BMI, body fat%, waist circumference and waist–hip ratio). The linear regressions with glucose homoeostasis (insulin, glucose and HOMA-index) and blood lipid (TC, LDL, HDL, VLDL, non-HDL/HDL, LDL/HDL, TC/HDL, LDL corrected for LpA cholesterol, TG or TG/HDL) as outcome were additionally adjusted for BMI z-scores and physical activity. A logistic regression was run to determine RTEC consumption difference in their risk for overweight.