Meta-analysis of the effects of psychosocial interventions on survival time in cancer patients

Objective: To provide a quantitative summary of effects of psychosocial interventions on cancer survival, and to present an overview of methodological and reporting aspects of the studies. Method: Electronic searches and manual searches of reference lists from review articles and retrieved papers. Two coders independently coded study, participant, treatment, and outcome characteristics of the studies meeting selection criteria. Results: Thirteen journal articles published between 1989 and 2003 reporting results from 14 controlled intervention studies were included. Results are based on data obtained from 2626 subjects. Effect sizes [hazard ratios (HR)] were heterogeneous and random effects models were used in the analyses. The total mean inverse-variance-weighted HR was 0.85 (95% CI: 0.65 - 1.11). Randomized studies (n = 8) showed nooverall treatment effect (HR: 0.77, 95% CI: 0.56-1.06), neither did the nonrandomized studies (HR: 1.00, 95% CI: 0.61-1.62). Interventions using individual treatment (n = 3) were, however, found to be effective (HR: 0.55, 95% CI: 0.43-0.70) but interventions using group treatment (n = 9) were ineffective (HR: 0.97, 95% CI: 0.73-1.27). Group treatments of breast cancer (n=6) were ineffective (HR: 0.95, 95% CI: 0.69-1.31). Conclusion: A definite conclusion about whether psychosocial interventions prolong cancer survival seems premature. Future studies should use randomization to avoid self-selection of patients with poor prognosis. Interventions should focus on a single diagnosis, take into account known risk factors, and describe their interventions thoroughly. (C) 2004 Elsevier Inc. All rights reserved.