If you read enough press about Cognitive Behavioural Therapy for psychosis, you will see two pictures of the same evidence base. In one, CBTp is a transformative treatment that everyone with schizophrenia should be offered. In the other, it is a modest intervention whose effects shrink to nothing in the most rigorous trials. Both descriptions are misleading. The actual evidence is interesting, contested, and worth understanding directly — particularly for patients and families weighing whether to invest the time CBTp requires.
CBTp produces small-to-moderate improvements in positive symptoms, larger improvements in symptom-related distress, and modest gains in negative symptoms and functioning. Effects are smaller in the most rigorously blinded trials, but consistent enough that NICE, the APA, and SAMHSA all recommend it.
Where CBTp came from
Aaron Beck described cognitive therapy for a delusional patient in 1952. The approach lay mostly dormant until the early 1990s, when several British groups — Kingdon and Turkington in Newcastle, Garety, Kuipers, Fowler and Bebbington in London, and Tarrier in Manchester — independently developed manualised protocols and ran the first controlled trials. Within a decade these were combined into the modern CBTp tradition. The first major British government endorsement came in the original NICE schizophrenia guideline in 2002, updated most recently as NICE CG178.
What the major reviews show
Wykes and colleagues, 2008
This influential meta-analysis (Schizophrenia Bulletin) pooled 34 trials and reported moderate effects on positive symptoms. It was one of the foundations for the 2009 NICE update.
Jauhar and colleagues, 2014
Published in the British Journal of Psychiatry, this much-cited critique re-analysed 50 trials and found that when only blinded outcome trials were included, effects on positive symptoms shrank to small. Critics called this the "death of CBTp"; defenders pointed out that blinding patients to a talking therapy is impossible and that "blinded outcome" criteria therefore over-correct.
Mehl, Werner, and Lincoln, 2015
A targeted meta-analysis on delusions in Frontiers in Psychology showed small-to-moderate effects, larger when the therapy was specifically focused on delusions.
Bighelli and colleagues, 2018
A network meta-analysis in Lancet Psychiatry compared multiple psychotherapies for positive symptoms and found CBTp consistently among the more effective interventions.
NICE 2014 (CG178)
The current NICE psychosis guideline reviewed the cumulative evidence and recommended that CBTp be offered to all people with schizophrenia. NICE updates have not changed that recommendation since.
How big are the effects, really?
Effect sizes (Cohen's d) cluster roughly as follows across the major reviews:
- Positive symptoms: 0.2 to 0.4 (small to moderate)
- Distress associated with symptoms: 0.4 to 0.6 (moderate)
- Negative symptoms: 0.1 to 0.3 (small)
- Functioning: 0.2 to 0.3 (small)
- Hospitalisation rates: small reductions in some trials
For comparison, antipsychotic medications produce effect sizes of roughly 0.4 on positive symptoms in the highest-quality trials (Leucht et al., Lancet 2017). CBTp is not as strong as medication for positive symptoms — but it adds on top of medication, addresses different outcomes, and does not have the side effects.
The blinding problem
The biggest methodological issue in the CBTp literature is blinding. In drug trials you can give one group an active pill and the other an identical placebo, and neither patient nor doctor knows which is which. In a psychotherapy trial, the patient obviously knows whether they are in the talking-therapy group. This means that any subjective rating they provide — including most symptom measures — is potentially inflated by expectancy. Trials that use raters blinded to which arm the patient was in (rather than self-report) consistently show smaller effects.
This is not a unique problem for CBTp; it applies to every psychotherapy ever studied. It does mean that the headline numbers should be taken as upper bounds, and the more rigorously-blinded numbers as lower bounds. The truth almost certainly sits in between.
Specific populations
First-episode psychosis
The 2018 EDIE-2 trial and others have studied CBTp in early-intervention services. Effects are roughly comparable to chronic populations; benefits include functioning and reduced relapse.
Treatment-resistant schizophrenia
The MRC-funded FOCUS trial (Morrison et al., Lancet Psychiatry 2018) tested CBT in patients who declined or could not tolerate antipsychotic medication, and found significant reductions in symptoms and distress. This is one of the few datasets in this difficult population.
At-risk mental states
CBT in people at clinically high risk for psychosis (Morrison et al., BMJ 2012; van der Gaag et al., 2012) shows reduced rates of transition to full psychosis at 12 to 18 months, though the long-term picture is less clear.
Where the field is honest about its limits
- Effects on hallucinations per se are smaller than effects on related distress
- Negative symptom benefits are real but small
- Long-term durability beyond 18 months is under-studied
- Effects are variable across trials and depend heavily on therapist training
- Access to qualified CBTp therapists is poor in most countries
What the major guidelines say
- NICE (UK): offer CBTp to all people with psychosis or schizophrenia (CG178)
- APA (US): 2020 schizophrenia guideline recommends CBTp as part of integrated care
- SAMHSA: includes CBTp in its evidence-based practices for first-episode psychosis (samhsa.gov)
- WHO: includes psychological interventions in its mhGAP intervention guide for psychosis (who.int)
How to read this as a patient or family
The evidence is good enough to justify trying CBTp. It is not so strong that you should expect a transformation. The right framing is the one used for most chronic-illness adjuncts: this is one of several tools that, layered together, give the best chance of a good long-term course. It works best alongside medication, not instead of it; alongside lifestyle support, not as a replacement; with a properly trained therapist, not someone offering generic CBT.
This article is for educational purposes only and is not medical advice, diagnosis, or treatment. Always consult a qualified mental health professional. If you or someone you know is in crisis, call or text 988 in the US, or your local emergency number.