Methods of synthesis Trial effects were weighted by the inverse of their variance, and synthesised using a random-effects meta-analysis. Separate analyses were conducted for the outcomes of anxiety, depression, fatigue, pain and quality of life; overall, and by assessment after treatment, or assessment at follow-up. Heterogeneity was investigated through the Q statistic and Ι². Publication bias was assessed in funnel plots and statistically. Primary effect moderators were included in a weighted,