is the evidence that cbt works for schizophrenia poor?
“If I look at the literature on CBT, for example, and its uses in cases of schizophrenia, the evidence is pretty poor that it works.”
a guest, historian of psychiatry (Andrew Scull) said on Lex Fridman Podcast, 2026-09-20, at 17:19. hear it in context → the speaker cited: the literature on CBT.
first, the part that holds
the speaker is right on the main point. pooled trials of cbt for schizophrenia find small effects, the effects shrink toward zero when outcome raters are blinded, and independent reviews rate the certainty of the evidence low or very low.
what we found when we went and looked
we read four meta-analyses and one umbrella review of meta-analyses. a 2014 meta-analysis of about 34 randomised trials found pooled effect sizes of -0.33 for overall symptoms, -0.25 for positive symptoms and -0.13 for negative symptoms. in trials where raters knew who got cbt, the effect on positive symptoms was -0.57. in trials where raters were blinded, it was -0.08. a 2024 umbrella review of 16 meta-analyses (70 comparisons) found "no convincing or highly suggestive evidence" for cbt, said 57.1% of comparisons were non-significant, and asked guideline bodies to reconsider. what the speaker's phrasing leaves out: "poor evidence" is not the same as evidence of no effect. a 2018 network meta-analysis found cbt added to antipsychotics reduced positive symptoms more than usual care (smd -0.30, moderate confidence), so a small benefit is reported, not proven absent.
assertion by assertion
| assertion | verdict | what the record says |
|---|---|---|
| the evidence that cbt works for schizophrenia is poor | confirmed | pooled effects are small (overall -0.33, positive -0.25, negative -0.13, about 34 rcts) and fall to -0.15 (overall) and -0.08 (positive) in rater-blinded trials. the umbrella review rated no outcome above 'suggestive' and found 57.1% of 70 comparisons non-significant. one limit: 'poor' is a judgement about the strength of the evidence. it does not say the effect is zero, and one estimate against usual care carries moderate confidence. [1][2][4] |
| a popular reading of the remark: cbt has been shown not to work for schizophrenia | overstated | the trials do not show zero effect. against usual care, cbt reduced positive symptoms (smd -0.30, 95% ci -0.45 to -0.14, moderate confidence). against supportive therapy the smd was -0.47 (low confidence). the effect is small and fragile, but it has been measured. [2] |
| apparent benefits shrink when trials control for bias | confirmed | blinding changed the result significantly. on positive symptoms, unblinded trials showed -0.57 and blinded trials showed -0.08. on overall symptoms, -0.62 against -0.15. a separate analysis found higher response rates in trials whose researchers had an allegiance to the therapy. [1][3] |
| cbt does little for negative symptoms or day-to-day functioning | confirmed | the umbrella review found weak (es -0.13) or non-significant evidence on negative symptoms and called it 'no clinically relevant effect'. a 2022 meta-analysis of 30 rcts found a functioning benefit of smd -0.26 but rated confidence in that estimate 'very low' because of bias, heterogeneity and possible publication bias. [4][5] |
| how many patients improve meaningfully with cbt | confirmed | in 33 rcts (1,142 people on cbt added to medication), 44.5% reached a 20% improvement in overall symptoms, which the authors call 'minimal'. only 13.2% reached a 50% improvement. these are single-arm response rates with no comparison group, so they do not isolate the effect of cbt. [3] |
both directions
for the speaker: the 2014 meta-analysis puts every symptom effect in the 'small' range, and the positive-symptom effect nearly disappears (-0.08) in rater-blinded trials. the 2024 umbrella review of 16 meta-analyses found no convincing or highly suggestive evidence for any outcome. more than half of its 70 comparisons were non-significant, effects on hallucinations and functioning weakened at longer follow-up, and it asked guideline bodies to reconsider their recommendations. a response-rate analysis found that only 13.2% of people getting cbt plus medication reached a 50% improvement in overall symptoms. the 2022 functioning meta-analysis rated its own estimate's certainty very low.
against: small and uncertain is not the same as nothing. the 2018 network meta-analysis of 53 trials and 4,068 participants found cbt added to antipsychotics reduced positive symptoms more than usual care (smd -0.30), with moderate confidence. it also beat supportive therapy and inactive controls, both at low confidence. the 2022 meta-analysis found a functioning benefit (smd -0.26). in the 2014 meta-analysis, the kind of control condition used did not change the effect, and there was little sign of publication bias. listeners who hear the remark as 'cbt has been shown not to work' would be overstating the record.
what this cannot say
'pretty poor' is a qualitative judgement, so we checked it against how the reviews rated their own certainty, not against a single number. all of these reviews pool trials of cbt added to antipsychotics, mostly in moderately ill adults. they say little about cbt as a stand-alone treatment, about first-episode care, or about any one person. standardised mean differences from rating scales have no obvious clinical meaning. two primary records could not be fetched in this session: the cochrane review comparing cbt with other psychosocial therapies (the site returned 403) and the nice psychosis guideline (also 403). we have not cited either, even though both bear directly on the question. the reviews we did read overlap heavily in the trials they pool, so they are not independent confirmations of each other.
sources, each one fetched and read
- Jauhar S, et al. Cognitive-behavioural therapy for the symptoms of schizophrenia: systematic review and meta-analysis with examination of potential bias. British Journal of Psychiatry, 2014;204(1):20-29. doi:10.1192/bjp.bp.112.116285 What we read there: pooled effect sizes -0.33 overall (34 studies), -0.25 positive (33), -0.13 negative (34). masking significantly moderated effect size: unmasked vs masked -0.62 vs -0.15 overall and -0.57 vs -0.08 positive. 'cbt has a therapeutic effect on schizophrenic symptoms in the small range. this reduces further when sources of bias, particularly masking, are controlled for.' Fetched 2026-10-21 (HTTP 200). https://www.cambridge.org/core/journals/the-british-journal-of-psychiatry/article/cognitivebehavioural-therapy-for-the-symptoms-of-schizophrenia-systematic-review-and-metaanalysis-with-examination-of-potential-bias/10B5A4BB22C8A82FE55E578A15975753
- Bighelli I, et al. Psychological interventions to reduce positive symptoms in schizophrenia: systematic review and network meta-analysis. World Psychiatry, 2018;17(3):316-329. doi:10.1002/wps.20577 What we read there: 53 rcts, 4,068 participants, psychological treatment added to antipsychotics. on positive symptoms, cbt vs inactive control smd -0.29 (95% ci -0.55 to -0.03, low confidence), vs treatment as usual -0.30 (-0.45 to -0.14, moderate confidence), vs supportive therapy -0.47 (-0.91 to -0.03, low confidence). effect sizes in the lower to medium range. Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC6127754/
- Bighelli I, et al. Response rates in patients with schizophrenia and positive symptoms receiving cognitive behavioural therapy: a systematic review and single-group meta-analysis. BMC Psychiatry, 2018 What we read there: 33 rcts, 1,142 people receiving cbt. 44.5% reached a 20% improvement and 13.2% a 50% improvement in overall symptoms. 'adding cognitive behavioural therapy to pharmacotherapy brings about a minimal improvement in overall symptoms among 44.5% of its recipients.' higher response with expert therapists and researcher allegiance. Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC6280425/
- Berendsen S, et al. Cognitive behavioural therapy for the treatment of schizophrenia spectrum disorders: an umbrella review of meta-analyses of randomised controlled trials. eClinicalMedicine, 2024. doi:10.1016/j.eclinm.2023.102392 What we read there: 16 meta-analyses, 70 comparisons. 'no convincing or highly suggestive evidence' for cbt. suggestive evidence for general psychopathology (es -0.33) and hallucinations (es 0.32) at end of treatment, weaker at follow-up. weak or non-significant evidence on negative symptoms. 57.1% of comparisons non-significant. guidelines 'should reconsider their recommendations.' Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC10809079/
- Bighelli I, et al. Effects of psychological treatments on functioning in people with schizophrenia: a systematic review and meta-analysis of randomized controlled trials. European Archives of Psychiatry and Clinical Neuroscience, 2022 What we read there: 30 rcts, 2,657 participants. cbt improved functioning, smd -0.26 (95% ci -0.39 to -0.12), with substantial heterogeneity. 'the confidence in the estimate was evaluated as very low due to risk of bias, heterogeneity and possible publication bias.' Fetched 2026-10-21 (HTTP 200). https://pmc.ncbi.nlm.nih.gov/articles/PMC10238355/
speakers are described by role, not named. one sentence is quoted under fair use, with attribution and a link to the original at the timestamp; we do not re-host audio or video. nothing on this page is a reason to start, stop or change a medication. if you have the file behind anything marked unverifiable, the research board is where people can answer. adam