Psychological treatment outcomes for cancer patients: what do meta-analyses tell us about distress reduction?

OBJECTIVE: The effectiveness of psychological treatment for distress reduction in cancer patients has been frequently studied and reviewed in systematic reviews but reviewer conclusions vary considerably. Clear and consistent evidence is needed to assist clinicians and administrators with their decision-making. We hypothesized that uneven handling of confounding methodological features are at least partly the reason for disagreements and reviewed the literature in this light.
METHOD: A systematic review of 14 published meta-analyses was conducted to determine whether due consideration of moderating variables in psycho-oncological treatments permits clearer recommendations. Quality of the reviews, treatment type, dosage, therapist qualities, outcomes at follow-up, and screening versus not screening for elevated distress were examined as moderator variables.
RESULTS: Treatment effects are consistently positive but also vary greatly in magnitude. There is lacking evidence for many important questions, in particular, differential treatment effects for different cancer types and stages. Regarding moderators of outcome, quality of review had no impact on results for depression but including lower quality reviews actually lead to underestimation of treatment effects for anxiety. The most potent negative moderator variable, however, is a floor effect that arises when patients are recruited for treatment studies without being selected for high levels of distress. Such indiscriminate recruitment is very frequent in psycho-oncology and leads to small reported treatment effects; when, however, patients are first screened for elevated distress, the ratio of observed treatment effects sizes is roughly three times greater.
CONCLUSION: Sweeping judgments about the effectiveness of psycho-oncological treatments for distress reduction are somewhat misleading and counter-productive. Among moderator variables, floor effects are particularly pervasive and have a large suppressor effect on observed outcomes.