Reading focus
Syllabus: How Does Psychotherapy Produce Change?
Keep control conditions and active-treatment comparisons separate; examine uncertainty and sensitivity analyses.
Why you’re reading it. Apply the outcome framework to a large evidence synthesis without turning a pooled effect into a mechanism.
A question to consider. How could the same collection support benefit relative to control conditions while offering little clear evidence of superiority over other psychotherapies?
Why this reading
An effect belongs to a comparison. This synthesis separates CBT versus control conditions from CBT versus other active treatments, allowing readers to see why evidence of benefit and evidence of superiority require different judgments.
Before you read
A meta-analysis combines results across studies. Hedges’ g expresses a difference in standard-deviation units. A confidence interval describes uncertainty around an estimate; a prediction interval reflects variation in effects across settings represented by the model. Neither describes every individual’s likely response.
Read for
- Compare the control and active-treatment analyses, then inspect how risk-of-bias and publication-bias analyses change the estimates.
- Distinguish directly reported response or remission rates from rates estimated using symptom-score distributions.
Consider afterwards
- How could a treatment help relative to usual care without clearly outperforming another psychotherapy?
- What further evidence would connect an observed benefit to a particular psychological mechanism?
Editorial context
The search covers trials through January 2022. High heterogeneity and risk of bias limit a single pooled summary; some longer-term comparisons contain few studies. Many binary outcomes are estimated. Depression studies across several formats provide a substantial worked example, not evidence that all therapies or diagnoses have interchangeable outcomes.