ABSTRACT
Background: Cognitive behavioural therapy (CBT) forms the standard psychotherapy for schizophrenia spectrum disorders (SSD). We aimed to summarize and evaluate the evidence on the effectiveness of CBT for SSD. Methods: In this umbrella review, we searched PubMed, Embase, Cochrane Database, and PsychInfo, for meta-analyses of randomised controlled trials (RCTs) of CBT in SSD published between database inception up to Aug 18, 2023. Inclusion criteria were RCTs investigating individually provided CBT in a population of patients with SSD, compared to either standard care, treatment as usually, or any other psychosocial therapies. No restrictions concerning follow-up or language were applied. We used the "assessment of multiple systematic reviews" (AMSTAR-2) appraisal checklist for the evaluation of methodological quality of meta-analysis. We extracted summary metrics from eligible studies in duplicate. The strength of evidence was classified by the sample size, p-value, excess significance bias, prediction intervals, significance of largest study, and heterogeneity. The strength of evidence was ranked according to established criteria as: convincing, highly suggestive, suggestive, weak, or not significant. Primary outcomes were general psychopathology, positive and negative symptoms. This study is registered in PROSPERO, CRD42022334671. Findings: We found 26 eligible meta-analyses, of which 16 meta-analyses provided sufficient data. Using the AMSTAR-2, we found limitations in details concerning the selection of study design, quality of the search and reporting of funding in included meta-analyses. A minority of 42.9% of the comparisons showed a significant result in favor of CBT; 57.1% were non-significant with no convincing or highly suggestive evidence. Suggestive evidence was found in favor of CBT for general psychopathology (6.2%, N = 34 RCTs, effect size (ES) = -0.33 (-0.47; -0.19), I2 = 67.93), delusions (16.7%, N = 27, ES = 0.36 (0.22; 0.51), I2 = 50.47), and hallucinations (33.3%, N = 28, ES = 0.32 (0.19; 0.46), I2 = 45.14) at the end of treatment (EoT). Weak (N = 34 RCTs, ES = -0.13 (-0.24; -0.02), I2 = 51.28), or non-significant evidence (N = 28 RCTs, ES = 0.12 (-0.03; 0.27) I2 = 64.63) was found for negative symptoms at EoT. At longer follow-up, evidence became weak or non-significant. Interpretation: Findings suggest that the effectiveness of CBT on general and positive symptoms in SSD at EoT was small to medium, while we found inconsistent evidence for a sustainable effect. CBT has no convincing impact on other relevant outcomes. Guidelines may use these results to specify their recommendations. Funding: None.
