Iris Kanera

164 Chapter 7 survival rate (97%) were very well presented (The Netherlands Cancer Registry, 2016). The overrepresentation of breast cancer survivors may also result from the well-organized breast cancer care in Dutch hospitals, which could have made access to this group relatively easier. Moreover, our sample was relatively young. Younger people may be more interested in and having more computer skills to use online cancer aftercare than older survivors (Makai et al., 2014; Statistics Netherlands, 2016). Besides, more women than men participate in web- based programs (Bantum et al., 2014; Kohl et al., 2013; Owen, Bantum, Gorlick, & Stanton, 2015). Hence, the interpretation of our results cannot simply be generalized to male cancer survivors, older survivors, survivors of other types of cancer than breast cancer, and survivors with more psychosocial and lifestyle risks, including a low level of PA. Preferences of less represented groups regarding web-based interventions should be explored in future research. Dropout In our RCT, the dropout (missing follow-up measurements) during the study period was small, with 11.5% after six months and 17.5% after 12 months, which reinforces the internal validity of the intervention (Eysenbach, 2005). Reasons for the low dropout might be the automated e-mail and telephone reminders to fill in the measurements. Comparable low dropout rates were also found in other web-based interventions designed for cancer survivors (Bantum et al., 2014; van den Berg et al., 2015). Missing follow-up measurements within web-based interventions may also be due to factors such as spam filters and e-mail address changes (Baker et al., 2010). Unfortunately, such data was not collected in our study. Although overall dropout was low, the RCT was subject of selective dropout, which might have threatened the internal validity. In Chapters 5 and 6 was found that participants of the intervention condition were more likely to drop out after six months and after 12 months compared to the control condition. This is not uncommon in web-based lifestyle interventions (Reinwand et al., 2015a; Walthouwer et al., 2015c). Reasons for the intervention group for dropping out might be that using the intervention required time and intensive cognitive effort. Moreover, after 12 months, the intervention group had nothing more to expect and they possibly did not realize the importance of the follow-up measurement, while the control group was awaiting intervention participation after the last follow-up measurement. Besides, participants of the entire study sample were more likely to drop out when being male, having lower modal/high income, lower vegetable intake, higher fruit intake, and received other cancer treatment than surgery, chemotherapy and/or radiation therapy. In the statistical analysis was corrected for these predictors of dropout to keep the influence of selective dropout as limited as possible, and intention-to-treat analyses were conducted to verify our results.