Moderators of the effect of psychosocial interventions on fatigue in women with breast cancer and men with prostate cancer: Individual patient data meta-analyses

Harriët J.G. Abrahams, Hans Knoop, Maartje Schreurs, Neil K. Aaronson, Paul B. Jacobsen, Robert U. Newton, Kerry S. Courneya, Joanne F. Aitken, Cecilia Arving, Yvonne Brandberg, Suzanne K. Chambers, Marieke F.M. Gielissen, Bengt Glimelius, Martine M. Goedendorp, Kristi D. Graves, Sue P. Heiney, Rob Horne, Myra S. Hunter, Birgitta Johansson, Laurel L. NorthouseHester S.A. Oldenburg, Judith B. Prins, Josée Savard, Marc van Beurden, Sanne W. van den Berg, Irma M. Verdonck-de Leeuw, Laurien M. Buffart*

*Corresponding author for this work

Research output: Contribution to journalReview articleAcademicpeer-review


Objective: Psychosocial interventions can reduce cancer-related fatigue effectively. However, it is still unclear if intervention effects differ across subgroups of patients. These meta-analyses aimed at evaluating moderator effects of (a) sociodemographic characteristics, (b) clinical characteristics, (c) baseline levels of fatigue and other symptoms, and (d) intervention-related characteristics on the effect of psychosocial interventions on cancer-related fatigue in patients with non-metastatic breast and prostate cancer. Methods: Data were retrieved from the Predicting OptimaL cAncer RehabIlitation and Supportive care (POLARIS) consortium. Potential moderators were studied with meta-analyses of pooled individual patient data from 14 randomized controlled trials through linear mixed-effects models with interaction tests. The analyses were conducted separately in patients with breast (n = 1091) and prostate cancer (n = 1008). Results: Statistically significant, small overall effects of psychosocial interventions on fatigue were found (breast cancer: β = −0.19 [95% confidence interval (95%CI) = −0.30; −0.08]; prostate cancer: β = −0.11 [95%CI = −0.21; −0.00]). In both patient groups, intervention effects did not differ significantly by sociodemographic or clinical characteristics, nor by baseline levels of fatigue or pain. For intervention-related moderators (only tested among women with breast cancer), statistically significant larger effects were found for cognitive behavioral therapy as intervention strategy (β = −0.27 [95%CI = −0.40; −0.15]), fatigue-specific interventions (β = −0.48 [95%CI = −0.79; −0.18]), and interventions that only targeted patients with clinically relevant fatigue (β = −0.85 [95%CI = −1.40; −0.30]). Conclusions: Our findings did not provide evidence that any selected demographic or clinical characteristic, or baseline levels of fatigue or pain, moderated effects of psychosocial interventions on fatigue. A specific focus on decreasing fatigue seems beneficial for patients with breast cancer with clinically relevant fatigue.

Original languageEnglish
Pages (from-to)1772-1785
Number of pages14
Issue number11
Publication statusPublished - Nov 2020

Cite this