There’s been a lot of chatter here in the woodlands about the role of cognitive behavioural therapy (CBT) in psychosis – what do service users think of it? Can it be used in place of antipsychotics for some people? Outside of the woodlands, CBT for psychosis has also been generating a lot of attention: Does it work at all? Has it been oversold?

When I was a young elf, just beginning my training, I was told that, in terms of evidence base, all psychotherapies were created equal; what mattered were non-specific factors, particularly related to client-therapist relationship. This was called the dodo bird hypothesis:

Everybody has won and all must have prizes

When it comes to psychotherapy and psychosis however it seems that CBT has won overall, as it is now recommended by NICE as a first line intervention in both the treatment and prevention of psychosis. Although family therapy and psychodynamic principles also receive an honourable mention in the guidance (NICE, 2014).

Is there evidence that CBT is better than other forms of psychotherapy? A fairly recent Cochrane review (Jones et al, 2012) suggested that CBT offered no advantage over other therapies, although this review has been criticised for the heterogeneity of its included studies.

To attempt to answer these questions, and finally slay the evil dodo, a new meta-analysis has been published in the American Journal of Psychiatry (Turner et al, 2014) with the aim to:

…improve our understanding of which therapy is most efficacious, and for which particular symptoms.

For the purposes of this review, symptoms of psychosis were divided between positive (e.g. hearing voices, delusional thoughts) and negative (e.g. apathy, slow thoughts, poor attention span).

Methods

Study selection and quality appraisal

• A systematic review was conducted to identify studies assessing psychological therapies against active controls
• Cochrane quality criteria assessed risk of bias

Data analysis

• Treatments were included in the meta-analysis if at least 5 trials were identified
• Control groups from each trial were pooled allowing direct comparison (e.g. CBT vs everything else)
• Between group differences were calculated using Hedges’ g (standardised mean difference)
• Funnel plots were used to measure publication bias
• Direct therapy comparisons were made where possible
• Sensitivity analyses for studies at high, low and no risk of bias were conducted
• The impact of researcher allegiance was assessed

Results

• 48 studies (3,295 participants in total) were included in the analysis
  • 24 studies used an individual therapy
  • 21 used a group therapy
  • 3 used a mixture
• Six psychological modalities were identified

1. Befriending (supportive individual and group meetings)
2. CBT (identification of self-defeating thought and negative behaviours, challenged through a series of cognitive and behavioural experiments, e.g. gathering evidence for and against a conclusion)
3. Cognitive remediation (support and improvement of cognitive deficit through cognitive exercise)
4. Psychoeducation (provision of evidence and information relating to diagnosis)
5. Social skills training (training to improve function in social situations)
6. Supportive counselling (opportunity to discuss problems in a supportive, healing environment)

• Length of follow up ranged from 3 weeks to 2 years

Specific therapies vs all other therapies pooled

• Befriending therapy fared worst
  • All symptoms: g= -0.37 (95% Confidence Interval [CI] -0.60 to -0.13) [suggesting worsening of symptoms]
  • This finding remained statistically significant when controlling for high risk of bias, but not for low or no risk of bias

• CBT was effective against symptoms overall and positive, but not negative symptoms
  • All symptoms: g = 0.16 (95% CI 0.04 to 0.28) [Survived high risk of bias sensitivity analysis, but not low or no risk of bias analysis]
  • Positive symptoms: g = 0.16 (95% CI 0.04 to 0.28) [Survived bias sensitivity analyses]
  • Negative symptoms: g = 0.04 (95% CI -0.09 to 0.16) [Not statistically significant]
• Cognitive remediation was effective vs other pooled therapies if studies at high risk of bias were excluded
  • All symptoms g = 0.20 (95% CI 0.01 to 0.39)

• Social skills training was effective for negative symptoms
  • All symptoms: g = 0.06 (95% CI -0.17 to 0.28) [Not statistically significant, but became so on exclusion of studies at high risk bias]
  • Positive symptoms: g = 0.09 (95% CI -0.23 to 0.41) [Not statistically significant]
  • Negative symptoms: g = 0.27 (95% CI 0.01, 0.53) [Survived test for bias]

Direct comparison between therapies

• This analysis was limited by the number of studies
• CBT was more efficacious than befriending (g = 0.419, p<0.05 [less than 1 in 20 chance of result occurring by chance, i.e. referred to as statistically significant])
• CBT was more efficacious than supportive counselling for positive symptoms (g=0.226, p<0.05)

 Researcher allegiance

• The effect of researcher allegiance was assessed for CBT and social skills training
  • In CBT trials, allegiance assessment led to a loss of statistical significance; but there were only 3 studies with no author allegiance
  • The social skills training findings survived this analysis

Publication bias

• Funnel plots suggested:
  • Inclusion of unpublished studies would render effects of cognitive remediation non-statistically significant
  • There was a suggestion of missing studies for social skills training, but this had no effect on analysis

Conclusions

The authors concluded:

There are small but reliable differences in efficacy between psychological interventions for psychosis, and they occur in a pattern consistent with the specific factors of particular interventions.

Summing up

This study used robust meta-analysis methodologies and a wide-ranging assessment for biases affecting outcomes in the reporting of trials.

So, do the findings from this meta-analysis finally manage to shoot the dodo? The authors are uncertain:

1. They suggest that inter-therapy differences are small with uncertain clinical significance
2. However they also highlight the finding that CBT robustly affects positive symptoms (aside from when author allegiance is taken into account), while social skills training affects negative symptoms. When the therapeutic methods of these two modalities are considered it could be hypothesised that these are the effects that would be anticipated, suggesting the possibility of specific therapy effects beyond general effects

They recognise the limitations of their methodology:

• A low number of studies, which limited the statistical power of the analysis
• Categorising of trials is subjective and a focus on positive, negative and overall symptomatology may bias findings against modalities not directly targeting symptoms. For example, psychoeducation aims to improve compliance with other treatments and would not necessarily therefore be expected to reduce symptomatology itself
• As usual more research is called for, especially well designed trials that consider researcher allegiance

So is the dodo dead?

I think this conclusion will likely be informed by “reader allegiance” as much as anything else.

For me, I think the bird is merely winged and will live to fight another day, although I should reveal my own bias towards more analytic and psychodynamically informed thinking. Unfortunately trying to design clinical trials to address that thorny issue is a whole other can of worms…

Links

Turner DT, van der Gaag M, Karyotaki E, Cuijpers P. Psychological Interventions for Psychosis: A Meta-Analysis of Comparative Outcome Studies. Am J Psychiatry. 2014 Feb 14. doi: 10.1176/appi.ajp.2013.13081159. [Epub ahead of print] [Abstract]

Judge, J. Service user perspectives on individual CBT for psychosis. The Mental Elf, 12 Mar 2014.

Tomlin, A and Badenoch, D. Pilot study suggests that CBT may be a viable alternative to antipsychotics for people with schizophrenia, or does it? The Mental Elf, 6 Feb 2014.

Jauhar, S., McKenna, P. J., Radua, J., Fung, E., Salvador, R., & Laws, K. R. (2014). Cognitive-behavioural therapy for the symptoms of schizophrenia: systematic review and meta-analysis with examination of potential bias. Br-J-Psychiatry, 204(1), 20–29 [PubMed abstract]

McKenna, P., & Kingdon, D. (2014). Has cognitive behavioural therapy for psychosis been oversold? BMJ, 348, g2295. doi:10.1136/bmj.g2295

Shepherd, A. Psychosis and schizophrenia in adults: updated NICE guidance 2014. The Mental Elf, 19 Feb 2014.

Psychosis and schizophrenia in adults: treatment and management (full guideline PDF). NICE, CG178, Feb 2014.

Jones C, Hacker D, Cormac I, Meaden A, Irving CB. Cognitive behaviour therapy versus other psychosocial treatments for schizophrenia. Cochrane Database of Systematic Reviews 2012, Issue 4. Art. No.: CD008712. DOI: 10.1002/14651858.CD008712.pub2.