Statistical Analysis

Analyses to compare baseline characteristics between groups included descriptive statistics. To evaluate the effect of the intervention, group differences in change over the 11-week study period for each continuous outcome were determined using analysis of covariance adjusted for baseline values. Due to the small number of men exceeding the cutoff for anxiety or depression at baseline, these outcomes were analyzed using a continuous score. We assessed interactions between group assignment and baseline values by entering the cross product of these variables in each model. Wilcoxon rank sum tests were applied to categorical outcomes. Subjects with missing values or implausible values were excluded from the main analyses. In analyses with CRP as the outcome, participants with levels of CRP greater than 10 mg/L at baseline or at the end of follow-up were excluded.

We conducted further analyses to evaluate effects of potential confounding by variables that were imbalanced between groups at baseline. Variables considered to be potential confounders were added to each model to determine change in parameter estimates and statistical significance. Due to the small sample size, we evaluated the effect of adjustment for each potential confounder separately. To account for potential weight loss due to advanced disease, we also examined change in body composition excluding subjects with stage T3 prostate cancer. All tests were two-sided and an α-level of 0.05 was applied to evaluate statistical significance. Data analysis was performed using SAS version 9.4.