Promoting physical activity among cancer survivors: Meta-analysis and meta-CART analysis of randomized controlled trials.

Objective: We conducted a meta-analysis of physical activity interventions among cancer survivors to (a) quantify the magnitude of intervention effects on physical activity and (b) determine what combination of intervention strategies maximizes behavior change. Method: Out of 32,626 records that were located using computerized searches, 138 independent tests (N = 13,050) met the inclusion criteria for the review. We developed a bespoke taxonomy of 34 categories of techniques designed to promote psychological change, and categorized sample, intervention, and methodological characteristics. Random effects meta-analysis and metaregressions were conducted; effect size data were also submitted to meta-analysis with classification and regression trees (i.e., meta-CART). Results: The sample-weighted average effect size for physical activity interventions was d(+) = .35, equivalent to an increase of 1,149 steps per day. Effect sizes exhibited both publication bias and small sample bias but remained significantly different from zero, albeit of smaller magnitude (d(+) >= .20), after correction for bias. Meta-CART indicated that the major difference in effectiveness was attributable to supervised versus unsupervised programs (d(+) = .49 vs. .26). Greater contact time was associated with larger effects in supervised programs. For unsupervised programs, establishing outcome expectations, greater contact time, and targeting overweight or sedentary participants each predicted greater program effectiveness, whereas prompting barrier identification and providing workbooks were associated with smaller effect sizes. Conclusion: The present review indicates that interventions have a small but significant effect on physical activity among cancer survivors and offers insights into how the effectiveness of future interventions might be improved.