Iris Kanera

2 49 Prevalence and correlates of lifestyle behaviors Different patterns of correlates For each separate lifestyle behavior we found different prevalence and different patterns of correlates. In accordance with the assumptions of social cognitive theories, we identified proximal variables and intention as strongest correlates in all examined behaviors, although with variations in contribution. Our results confirm theoretical assumptions, that the relative contribution of attitudes, self-efficacy and social influences can differ from one person to another and from one behavior to another (Fishbein & Ajzen, 2010). Regarding the distal factors, we found notably less, but also different patterns of correlations between the lifestyle behaviors. Overall, subscales of the EORTC QLQ-C30 provided the most influential distal factors, although the contribution of all distal factors (socio-demographic, cancer- related, psychological) was considerably lower than the contribution of the proximal factors and intention. It would be interesting to investigate a possible predicting role of the distal factors and possible mediation effects of the proximal factors in longitudinal research. Limitations This study was subject to some limitations. Due to the cross-sectional design, no causal relationships and directions of associations could be determined. Furthermore, the collected data were based on self-report questionnaires. In particular, self-reported outcomes of lifestyle behaviors should be interpreted carefully. In addition, the results of his study might not be generalizable to all cancer survivors, because more than half of the sample has been women with breast cancer. Even though, cancer type and gender had limited correlates in explaining the lifestyle behaviors. In measuring PA using IPAQ short form, possibly over-reporting might have been occurred. This is known as a typical problem in several previous studies using the same questionnaire (Lee, Macfarlane, Lam, & Stewart, 2011). In this study, the cut-off point to achieve the PA recommendations was 600 MET-min/week, which is in accordance with the scorings guideline of the IPAQ questionnaire. However, in guidelines, different cut-off points or ranges were indicated (Garber et al., 2011; Haskell et al., 2007; Nelson et al., 2007). Our cut-off point choice might have affected the outcome of the adherence to PA recommendations. With regard to alcohol consumption, it could be that the results on alcohol are more a reflection of social drinkers and excessive drinkers, because some questions were focused on alcohol consumption, and non-drinkers might have found them to be not applicable to themselves. Although, similar questions were also applied to non-drinkers in prior research (Schulz et al., 2014). There was a probability that significant correlates could have occurred by chance due to multiple testing. However, by applying sequential multiple linear / logistic regression analyses, the chance on Type 1 error was rather small (Tabachnick & Fidell, 2001). Moreover,