Iris Kanera

146 Chapter 6 (median = 57 years) confirmed this by showing that the intervention was effective among participants aged younger than 57 years after six months ( B = 141.819, SE = 69.126 [95% CI = 6.34; 277.30], p = .040, d = .38, f 2 = .008) and after 12 months ( B = 256.549, SE = 70.941 [95% CI = 117.51; 395.59], p = .000, d = .59, f 2 = .076), while it was not effective among participants aged 57 years and older ( B 6 month = 38.343, SE = 65.579 [95% CI = -90.19; 166.88], p = .559; B 12 month = -16.047, SE = 68.337 [95% CI = -149.98; .50], p = .814). Among the younger participants, the increase in moderate PA at six months from baseline was 197.2 min p/w in the IC vs 36.1 min p/w in the UC, and after 12 months, the increase of moderate PA from baseline was 226.5 min p/w in the IC vs -49.6 in the UC (i.e. a decrease in PA). Among the older participants (≥ 57 years of age), the increase in moderate PA at six months from baseline was 102.4 min p/w in the IC vs 102.5 min p/w in the UC, and after 12 months, there was a decrease of moderate PA from baseline of -71.8 min p/w in the IC vs an increase of 33.8 min p/w in the UC. Additional tests ruled out age as a confounder. The PA module was used by n = 46 (28.1%) of the IC participants who completed the 12-month follow-up questionnaire. Using the PA module did not moderate the intervention effect on the PA outcome after six months (β = 47.692, SE = 38.714 [95% CI = -28.19; 123.57], p = .218) and after 12 months (β = -27.736, SE = 39.302 [95% CI = ‑104.77; 49.29], p = .480). Furthermore, no significant moderation effects were found for gender (β = 109.065, SE = 104.309 [95% CI = -113.51; 95.38], p = .296) and education (β medium = -8.275, SE = 94.673 [95% CI = -193.83; 177.28], p = .930; β high = ‑87.163, SE = 93.508 [95% CI = -270.44; 69.11], p = .351). DISCUSSION The present study evaluated the 12-month effects of the web-based KNW on moderate PA and vegetable consumption. Complete cases, as well as ITT analyses, revealed that the fully automated KWN was effective in increasing and maintaining moderate PA in the long term among early cancer survivors. However, the 6-month intervention benefits on vegetable consumption did not persist in the long term. It is very positive and promising that the effect sizes of the KNW effect on moderate PA increased over time. In contrast, Kohl et al. (2013), reported in their review of reviews of Internet-delivered behavior change interventions among the general population, that four out of six reviews on PA-only interventions showed that PA-only interventions were effective to increase PA. The reported effect sizes were modest and decreased during follow- up. The KNW effect on moderate PA was moderated by age. Respondents of the IC, aged younger than 57 years significantly increased moderate PA, while older participants (≥ 57 years) did not significantly increase moderate PA over time. The magnitude of this stratum- specific difference is of clinical importance. Higher effect sizes were found in a subgroup of participants younger than 57 years of age compared to the overall effect on moderate PA.