Trial of a Behavior Change Intervention to Increase Aerobic and Resistance Exercise and Quality of Life in Older Prostate and Breast Cancer Survivors

Recruitment Potential participants will be contacted by the Newfoundland and Labrador Cancer registry. They will be given a brief explanation of the study will be asked to consent to a mail-out questionnaire from the registry. The questionnaire will contain a cover letter explaining the study in more detail along with informed consent. If the participant wishes to continue in the study, they will return the questionnaire to the research team, who will contact them from this point. Participants will also be recruited through presentations at local support groups as well as advertisements in local papers and posters at local hospitals and recreation centers. Power Analysis According to G\*Power a sample of 40 prostate and breast cancer survivors per group (n=120) is needed to detect a medium effect size (d= .50) on our primary outcome (i.e., PA) with a power of 0.80, at a p-level of 0.05. We expect a 20% attrition rate, based on previous research (McGowan, North, \& Courneya, 2013; Vallance, Courneya, Plotnikoff, Yasui, \& Mackey, 2007) therefore we will attempt to recruit a sample of 150 breast and prostate cancer survivors, which would allow for 50 participants per group. Data Analysis Analyses of covariance will be used to examine group differences on our primary outcome (i.e., PA minute/week) at month 1 and month 3, and secondary outcomes (e.g., QoL) at month 3. Additional analyses of covariance will be run to explore the group differences on sedentary behaviour, and light, moderate and vigorous minutes of activity/week at month 1 and month 3. Medical (e.g., comorbidities, months since diagnosis) and demographic (e.g., marital status, age, BMI) will be tested as possible moderators of PA behaviour change. If we experience a large proportion of missing data multiple imputation techniques in SPSS will be used to replace missing values. Otherwise, a last outcome carried forward approach will be used. The relationships between self-reported PA and QoL, and objectively measured PA and QoL in older breast and prostate cancer survivors will be explored. To examine this, differences in QoL between participants in the two PA categories (i.e., meeting PA guidelines and not meeting PA guidelines) will be tested using analysis of variance (ANOVA). These analyses will be repeated using analysis of covariance (ANCOVA) to control for the demographic and medical variables that had statistically significant associations with the QoL. The relationship between objectively measured PA and self-reported PA will be explored using a two-way mixed intraclass correlation coefficient to calculate the level of absolute agreement between the two types of measurements.