Pilot study suggests that CBT may be a viable alternative to antipsychotics for people with schizophrenia, or does it?


People with schizophrenia stop taking their antipsychotics for a wide range of reasons (e.g. debilitating side effects or a belief that they will not help them), but when they do health professionals often find it extremely difficult to care for these patients, because the alternative treatment options available to them are very limited.

Of course, patients not taking antipsychotics often disappear from mental health services because they are not engaged in treatment and sometimes live in fear of being sectioned and forced to restart their medication.

We know from the biggest trial to date comparing antipsychotics, that as many as three quarters of patients with schizophrenia discontinued their drugs over 18 months (Lieberman et al, 2005), so this is a widespread problem and alternative treatments are badly needed.

To date, the research looking at combining antipsychotics with CBT for schizophrenia has brought back mixed results.  At least one meta-analysis supports the idea that the two treatments can work safely and effectively alongside each other, whilst various other reviews and analyses are less encouraging.

A new trial published today in The Lancet offers a chink of light for people with schizophrenia who do not take antipsychotics as (according to Oliver Howes who has written an attached commentary) it provides:

proof of concept that cognitive therapy is an alternative to antipsychotic treatment.


The research team led by Prof Anthony Morrison from the University of Manchester conducted a single-blind, pilot randomised trial that recruited people with schizophrenia who had decided not to take antipsychotics.

74 people were randomised to one of two groups:

  1. Treatment as usual (TAU)
  2. Cognitive therapy plus TAU

The primary outcome was total score on the Positive And Negative Syndrome Scale (PANSS), which was measured at baseline and then every 3 months until 18 months.  The PANSS is a clinician-administered semi-structured interview assessing positive symptoms (e.g. hallucinations, delusions, suspiciousness, paranoia), negative symptoms (e.g. lack of initiative, social withdrawal, lack of expression, emotional withdrawal), and general psychopathology (e.g. depression, anxiety, poor insight, guilt).

Data were analysed according to a predetermined plan which meant that investigators could not ‘cherry pick’ good outcomes in favour of their treatment; all amendments to the study plan were documented in the report; assessments of outcome were taken by assessors who did not know which therapy participants were receiving (rater blindness).

Patients could not be blinded in this study. Indeed, it’s worth noting that there are many factors that make this trial very difficult to carry out, but overall the authors should be commended for publishing a well conducted piece of research.


The trial reports that:

  • Compared to TAU, cognitive therapy is associated with important treatment signals including a reduction in psychotic experiences such as hearing voices and paranoia and an improvement in day to day social functioning
  • Mean PANSS total scores were consistently lower in the cognitive therapy group than in the TAU group, with an estimated between-group treatment effect size of -6·52 (95% CI -10·79 to -2·25; p=0·003)
    • This equates to a standardised effect size (Cohen’s d) of 0·46


  • Analysis of the effect sizes for CBT vs controls at 9 months reveals no significant differences on PANSS total, PANSS positive or PANSS negative. This finding is consistent with the results of the recent meta-analysis published in the British Journal of Psychiatry (Jauhar et al, 2014)
  • Furthermore, the baseline and 9 months PANSS data for the unmedicated TAU group show a symptom improvement that is just as large as the CBT group at 18 months.  In other words, the end of study finding for CBT is no bigger than the natural levels of fluctuation (improvement) that can be seen in the data for the unmedicated patients
  • CBT did not reduce distress at all (for delusions or hallucinations), self-rated recovery, depression or anxiety
Can we be sure that the trial results represent more than just natural fluctuation?

Can we be sure that the trial results represent more than just natural fluctuation?


The lead researchers summarised their findings quite simply by concluding that:

Cognitive therapy significantly reduced psychiatric symptoms and seems to be a safe and acceptable alternative for people with schizophrenia spectrum disorders who have chosen not to take antipsychotic drugs. 

Joint lead author Douglas Turkington, Professor of Psychiatry at Newcastle University, said:

One of our most interesting findings was that when given the option, most patients were agreeable to trying cognitive therapy. If someone is on antipsychotics they should not just suddenly stop taking them as there is a major risk of relapse. Medical advice should always be sought if you are considering stopping your medication.

The Mental Elf team believes that this is an overstatement of the evidence as reported in this article, for the reasons set out below.


  • The patients in this trial were a highly selected group who were willing to try CBT. Clinicians will tell you that schizophrenia patients who won’t take antipsychotics are often difficult to engage in any form of therapy.
  • The study did not include an active comparison group such as befriending or supportive therapy.
  • Arguably to establish effectiveness of CBT a third placebo arm would have been ideal. This would be along the lines of supportive/non-specific counselling in addition to TAU, so as to show that it was CBT alone (and not the regular meeting with a professional to discuss symptoms) that led to the overall change in PANSS.
  • The issue of blinding is problematic, and cases where blinding was broken continued to be included in the study; albeit with attempts made to mask this unblinding. Patients could not be blinded to their group, and biases will have been introduced by this.
  • Not all patients were accounted for at the conclusion of the trial.  This crucial issue was difficult to appraise because different numbers of patients appeared to be included for different outcomes.  Furthermore, different numbers of patients were included for different time periods.
  • However, for all outcomes, drop out was high by the end of the study:  30% in the CBT group and 32% in the TAU group.  High enough that we’d expect it to affect our assessment of the effectiveness of the treatments.
  • Serious adverse events occurred in more than 10% of patients, mostly in the TAU group, which suggests that lack of medication is not an acceptable ‘treatment’.
  • A significant number of the patients in both groups were given antipsychotics during and after the trial, and this confounder does not seem to have been clearly addressed. Presumably these subjects should have been removed from the trial as it was specifically looking at those who were unmedicated.
  • The trial did not include any information on the costs of treatment, which is a significant omission considering that patients who completed the 18 months of therapy would have received 26 hours of very costly CBT.

Clearly, there is a risk of bias in these results and they need to be replicated in a bigger trial before changes are made in clinical practice. Professor Morrison and colleagues are about to commence such a study in Manchester to compare cognitive therapy alone with antipsychotic medication alone and with a combined treatment in people with schizophrenia spectrum disorders.

Moreover, perhaps the next study should be conducted by another research group? After all, the authors of this study are the top cognitive therapists for psychosis in the country, who have published extensively on the subject.  It might also be sensible to compare CBT with other, more diverse approaches such as social support, nurse visits or education in any future studies. Surely we would then be better placed to know the true value of CBT for this population?

The results of this trial are not as clear cut as the press release would have you believe

The results of this trial are not as clear cut as the press release would have you believe

Comments from experts in the field

As you would expect, this new RCT has generated a great deal of interest from the mental health community.

Prof Til Wykes, Professor of Clinical Psychology and Rehabilitation, Institute of Psychiatry, King’s College London, said:

The Morrison study is particularly interesting as it offers CBT to people who are engaged with mental health services, but have chosen not to take any medication. Most people who do not take medication usually stay as far away from mental health services as possible in case they ‘get sectioned’ and are made to take the drugs, so their sample is narrowly defined and it may not generalise to people who completely drop out of contact.

Prof Shitij Kapur, Professor of Schizophrenia, Imaging and Therapeutics, and Head of School at the Institute of Psychiatry, King’s College London, said:

This is an important trial that shows that CBT works – and works for a very difficult to treat population – the patients who do not take drugs. This is important as clinicians can often get nihilistic about patients who are “non-compliant”. What the paper tells us is that we can engage psychotic patients who do not take antipsychotics and deliver clinical benefits.

Prof Robin Murray, Professor of Psychiatric Research, Institute of Psychiatry, King’s College London, said:

This approach is certainly worth exploring further. This is especially so because we are realising that long-term antipsychotic treatment may cause receptor changes that make it difficult to stop the drugs, and consequently psychiatrists may become more cautious about the long-term prescription of antipsychotics.

Prof Keith Laws, Department of Psychology, University of Hertfordshire, said:

This study by Morrison and colleagues provides novel insights, but not necessarily those alluded to by the authors. Crucially, the symptom change they attribute to CBT is no greater than the level of ‘natural’ symptom  fluctuation observable in the controls (e.g. between baseline and 9 months) – so it’s quite possible that what Morrison et al are documenting is simply the ‘natural fluctuation’ of symptoms that occurs in unmedicated patients and nothing at all to do with CBT.


Morrison, AP. et al Cognitive therapy for people with schizophrenia spectrum disorders not taking antipsychotic drugs: a single-blind randomised controlled trial (PDF). The Lancet, Published online February 6, 2014 http://dx.doi.org/10.1016/S0140-6736(13)62246-1/ [Commentary PDF]

Lieberman JA, Stroup TS, McEvoy JP, et al, and the Clinical Antipsychotic Trials of Intervention Effectiveness (CATIE) Investigators. Effectiveness of antipsychotic drugs in patients with chronic schizophrenia. N Engl J Med 2005; 353: 1209–23.

Jauhar S, McKenna PJ, Radua J, Fung E, Salvador R, Laws KR. Cognitive-behavioural therapy for the symptoms of schizophrenia: systematic review and meta-analysis with examination of potential bias. Br J Psychiatry. 2014 Jan;204:20-9. doi: 10.1192/bjp.bp.112.116285. [PubMed abstract]

Naeem F, Farooq S, Kingdon D. Cognitive behaviour therapy (brief versus standard duration) for schizophrenia. Cochrane Database of Systematic Reviews 2013, Issue 7. Art. No.: CD010646. DOI: 10.1002/14651858.CD010646.

Buckley LA, Pettit TACL, Maayan N, Soares-Weiser K, Adams CE. Supportive therapy for schizophrenia. Cochrane Database of Systematic Reviews 2007, Issue 3. Art. No.: CD004716. DOI: 10.1002/14651858.CD004716.pub3. [Update soon to be published].


Thanks to my colleagues who contributed to this blog. It’s been very much a group effort with key contributions from Clive Adams, Douglas BadenochAlex LangfordKeith Laws, Kirsten Lawson, Chris Pell, Andrew Watson and Stephen Wood.

Share on Facebook Tweet this on Twitter Share on LinkedIn Share on Google+
Mark as read
Create a personal elf note about this blog
Profile photo of Andre Tomlin

Andre Tomlin

André Tomlin is an Information Scientist with 20 years experience working in evidence-based healthcare. He's worked in the NHS, for Oxford University and since 2002 as Managing Director of Minervation Ltd, a consultancy company who do clever digital stuff for charities, universities and the public sector. Most recently André has been the driving force behind the Mental Elf and the National Elf Service; an innovative digital platform that helps professionals keep up to date with simple, clear and engaging summaries of evidence-based research. André is a Trustee at the Centre for Mental Health and an Honorary Research Fellow at University College London Division of Psychiatry. He lives in Bristol, surrounded by dogs, elflings and lots of woodland!

More posts - Website

Follow me here –

Profile photo of Douglas Badenoch

Douglas Badenoch

I am an information scientist with an interest in making knowledge from systematic research more accessible to people who need it. This means you. I've been attempting this in the area of Evidence-Based Health Care since 1995. So far the results have been mixed. For some reason we expected busy clinicians to search databases and appraise papers instead of seeing patients. We also expected publishers to make the research freely available to the people who paid for it.. Ha! Hence The National Elf service.

More posts - Website

Follow me here –