Future analytical plan The feasibility and the interrelationship of the different stress measurements (questionnaires, cortisol in different biological samples, HRV) will be studied. This will be done using regression analyses corrected for age and sex, a hierarchical Nilotinib Bcr-Abl inhibitor linear model for salivary cortisol (to model the diurnal pattern of cortisol) and a triad analysis allowing to compare more than two measurements at the same time (to examine which method may most accurately indicate true childhood stress) [58]. Secondly, regression models stratified for gender and corrected for age can verify the mutual relationships between stress and lifestyle related factors (sleep, diet and physical activity). Also, the possible mediation effect of psychological eating behavior in the stress-diet relationship will be tested.

To reduce multicollinearity and complexity, principal component analysis will be executed on the stress questionnaires and lifestyle related factors. Finally, the association between stress and obesity will be tested both cross-sectionally and longitudinally (with repeated measures analysis). For this purpose, regression analyses will be executed correcting for the lifestyle related factors. Moreover, using mediation analyses the effect of lifestyle related factors on the stress versus body composition relationship will be examined. For a mediation effect, the relationship between the independent variable and the dependent variable should be significantly reduced after controlling for the mediator. This indirect effect will be formally tested with bootstrapping in SPSS [59] or in the more advanced Mplus software.

In all above-mentioned analyses, body composition parameters will be examined as continuous variables, such as body fat percentages, fat free mass percentages, BMI z-scores etc., instead of comparing obese versus non-obese groups. Results Participation rate Table 1 presents, the consent- and drop-out-percentage, as well as the number of cases valid for data-analysis for each measurement module that took place at baseline and first follow-up. In the baseline and first follow-up survey, participation proportions of 68.7% (N=523/761) and 65.8% (N=418/635) were obtained, respectively. Willingness to participate from baseline to first follow-up (i.e. consent percentage) was 71.3% (N=453/635). Table 1 Participation numbers, separately for baseline survey (2010) and first follow-up (2011) of ChiBS in Aalter, Belgium Socio-demographic characteristics Table 2 shows the children��s and parental socio-demographic characteristics of participants with a baseline Anacetrapib assessment of the stress questionnaires (ChiBS participants) compared to non-participating children in the ChiBS study.