Abstract
PURPOSE: Meeting intervention requirements is crucial in behavioral trials. We examined patterns and predictors of physical activity (PA) adherence and contamination in a 1-year individualized randomized controlled PA behavioral intervention in childhood cancer survivors (CCS). METHODS: CCS aged ≥16 at enrolment, <16 at diagnosis, and ≥5 years in remission were identified from the Swiss Childhood Cancer Registry. We asked participants randomized to the intervention group to perform an additional ≥2.5 h of intense PA/week and controls to continue as usual. Adherence to the intervention was assessed by online diary (adherent if ≥2/3 of individual PA goal reached) and contamination for the control group by pre- and post-questionnaire including PA levels (contaminated if >60 min increase/week in PA). Predictors of adherence/contamination including quality of life (36-Item Short Form Survey) were assessed by questionnaire. We used logistic (control group) and mixed logistic regression models (exercise group) to estimate predictors of study adherence and contamination. RESULTS: One hundred and forty-four survivors (30.4 ± 8.7 years old, 43% females) were included. Adherence was 48% (35/73) in the intervention group, while 17% (12/71) of controls contaminated group allocation. Predictors for PA adherence were female sex (OR 2.35, p = 0.03), higher physical (OR 1.34, p = 0.01) and mental quality of life (OR 1.37, p = 0.001), and week into the intervention (OR 0.98, p < 0.001). Clear differences in PA behavior of adherent and non-adherent participants were seen from week four. No significant predictors for contamination were found for controls. CONCLUSION: Adherence to PA behavior interventions remain challenging in both groups. Further long-term trials should consider intense motivational support within the first month, more detailed data collection for the control group, adjustments to power calculations and other study designs to minimize non-adherence and contamination.