HBSC: study design and sample
The HBSC study is an international health study of children and adolescents sponsored by the World Health Organisation (WHO) and developed jointly in 1982 by scientists from Britain, Finland and Norway; since then it has been conducted every 4 years in a growing number of countries (41 in 2005/06) [1]. A consistent and structured procedure using a detailed research protocol must be followed to ensure international comparability. Both the methodology for compiling the samples and a core questionnaire with compulsory questions are pre-specified, but these can be supplemented by additional questions. The sampling begins at the level of schools and focuses on classes in the 5th, 7th and 9th grades. At least 1,500 students in each of the three age groups must be questioned per survey date and country to ensure a representative survey. The analysis presented here is based on the survey conducted in Germany in 2005–2006 (N = 7,274; 3,606 girls and 3,668 boys aged 11–15 years), in which the federal German states (Bundesländer) of Berlin, Hamburg, Hessen, North Rhine-Westphalia and Saxony took part [15].
Measurement of BMI in the HBSC study
In the HBSC study all participants were asked to fill in a questionnaire in which they self-reported their height and weight. Subjective BMI was calculated using the formula BMI = weight/ height2.
Kromeyer-Hauschild’s BMI reference [17] values are currently used to define overweight and obesity in children and adolescents in Germany, in line with the recommendations of the Study Group on Obesity in Childhood and Adolescence (AGA) (see http://www.a-g-a.de). According to these reference values, children are considered overweight if they have a BMI above the 90th age- and gender-specific percentile of the Kromeyer-Hauschild reference system. They are deemed to be obese if their BMI is above the 97th percentile. Children or adolescents with a BMI below the age- and gender-specific 10th percentile are defined as underweight; those below the third percentile are considered extremely underweight [17].
Body image in the HBSC study
In the HBSC questionnaire all participants were asked to report their body image on a 5-point Likert-type scale. They were asked: "Do you think you are
➜ much too thin
➜ a bit too thin
➜ exactly the right weight
➜ a bit too fat
➜ much too fat ?" [1, 13].
Responses were classified into the following categories: (1) ‘too thin’, (2) ‘right weight’, and (3) ‘too fat’.
The correction procedure
In the KiGGS study [14], randomly selected boys and girls aged between 11 and 17 years were asked to self-report their height and weight in face-to-face interviews at respective study centres before being measured and weighed in a standardized fashion. Trained staff measured body height without shoes to an accuracy of 0.1 cm using a portable Harpenden stadiometer (Holtain Ltd., UK). Body weight was measured to the nearest 0.1 kg, wearing underwear, using a calibrated electronic scale (SECA Ltd., Germany). BMI was calculated and classified as mentioned above [17]. Body image was measured with the official German translation which is also used in the German version of the HBSC study [1, 13].
The correction methods suggested in [13] were applied to the prevalence estimates from the HBSC study. To correspond with the age groups included in the HBSC study, the representative KiGGS sample was restricted to adolescents aged between 11 and 15 years (N = 2,565; 1,216 girls and 1,349 boys) for the analyses presented here.
Correction procedure I
As explained in [13] the coefficients aij linking subjective and objective BMI classifications found on the basis of representative surveys are transferable to other studies based only on subjective statements, as long as those studies were carried out in the same age group of the same population in the same time period as the representative survey. Using correction formula (14) in [13] allows an estimation of the unknown true prevalence of overweight, normal weight and underweight, correcting for (gender-specific) distortions associated with subjective body image. Applying this to the HBSC study the estimated prevalence of the respective BMI class will be
(I)
Where aijKiGGS are the conditional probabilities calculated by the validation study KiGGS (as defined in formula (II) in [13]) and QjHBSC is the prevalence of subjective BMI category I
j
found in the HBSC study.
Correction procedure II
In [13] a further correction formula (16) was introduced which can be applied for any study that has information not only on subjective BMI, but on body image as well. With data from KiGGS, it could be shown that the association between objectively and subjectively estimated BMI depends greatly on subjective body image. Adolescents who considered themselves “a bit too fat” or "much too fat" indicated a substantially lower subjective BMI on average than those who considered themselves “exactly the right weight” [13].
If a parallel validation study can provide estimates of the associations between objective and subjective BMI for different body image groups, this information can be used to further improve the prevalence estimates. This is the case for the HBSC study, using KiGGS as a corresponding validation study. For reasons of simplicity, the body image categories are combined into three groups as proposed and defined in [13]. Thus a further correction of the prevalence of BMI categories estimated by the HBSC study can be achieved by
(II)
where
is the prevalence of body image category k in the HBSC study, and is the prevalence of subjective BMI category j in the group of adolescents with Body Image BI = k in the HBSC study.