Schizophrenia, antipsychotics and CBT

There was a study published in The Lancet last week looking at CBT compared with treatment as usual in those not receiving antipsychotic treatment (Free to access! Good stuff Lancet).

I took a look at this study on Wednesday evening but have been too busy with ethics hearings and being ill to string much together, by way of coherent thought, before now.

This study was reported in an excellent review on The Mental Elf website and has generated a lot of debate, as is to be expected. I thought I would add my own, somewhat belated, thoughts on the study.

Antipsychotics and Schizophrenia

For the purpose of this blog I’m going to accept Schizophrenia as a concept, albeit one I’m very uncomfortable with.

The primary treatment modality offered to people with a diagnosis of Schizophrenia is antipsychotic drugs. NICE guidance also suggests that CBT be offered to all.

It is the biomedical model of ‘treatment’ that, sadly in my opinion, predominates and while NICE suggests that antipsychotic treatment be ‘offered’ refusal of treatment is often met with resistance from care teams and attendant deterioration in therapeutic relationship. This situation is grossly complicated by the ‘elephant in the room’ of Mental Health Law.

I have concerns about, what I perceive as, an over reliance on antipsychotic treatment and a too reductionist approach to the psychosocial experience of mental distress. That being said I do not wish to disparage anyone taking, or cautiously prescribing, antipsychotic medication. I prescribe them and believe them to be of great benefit, when used appropriately, in relieving the acute distress associated with trauma and crisis – their maintained use I am far less certain about.

Cognitive Behavioural Therapy

CBT practitioners work with clients to try and understand the interconnection of Thoughts, Feelings and Behaviours. Formulations are developed collaboratively between practitioner, for example, feelings of anxiety are present during social interaction but reduced by self-isolation, voluntary isolation reinforces fear of social situations and increases the strength of the anxiety provoking thought, a cycle develops in which anxiety and isolation are mutually reinforcing.

Through identifying these reinforcing cycles they are ‘challenged’, for example through ‘behavioural experiments’.

Research into the outcomes of psychotherapy research studies has demonstrated that author affiliation correlates with study outcome. Therefore, before providing my thoughts on the CBT in antipsychotic free individuals study I will state my own bias:

Although I recognise a role for CBT and believe it has benefitted many, I personally find it to be a little too superficial! addressing only surface manifestations of deeper rooted problems. I have some clinical experience of working through a CBT based model – this had some positive outcomes but was I felt too limiting in scope and ran the risk of dismissing strong unconscious conflict. This is my view and I welcome opinion and debate with more experienced CBT practitioners than myself.

Cognitive therapy for people with schizophrenia spectrum disorders not taking antipsychotic drugs: a single-blind randomised controlled trial

As I said in the introduction this trial has been reviewed elsewhere so I will try to be brief and present only my conclusions. I’m happy to share my ‘working out’ in more detail on request…

Risk of bias
As I’ve said – author affiliation correlates with study outcome. All the authors in this study are proponents of CBT models.

I’m concerned that the ‘treatment as usual’ control arm of the trial received a pretty poor standard of care, this may represent alienation from clinical teams owing to ‘treatment refusal’.

There was no ‘placebo’ or active control, for example through befriending offered. The gap between intervention and treatment as usual is therefore quite substantial and a large non-specific placebo response can be expected to bias results towards the CBT intervention.

There was, in my opinion, quite a high loss of participants to follow up throughout the trial. I accept this is a difficult to reach group of people, but it raises some concerns for me about the participant experience in both trial arms. The resulting low numbers of participants makes the results difficult to interpret in my opinion.

Results
In terms of symptom severity those participants receiving CBT tended to have lower scores then the control group. However there was no single point of measurement during the trial when a statistically significant difference was found between groups. This means that when intervention stopped (9 months) and follow up stopped (18 months) it is not possible to distinguish between groups. Also described here.

The authors’ quote overall differences averaged over every measurement point. This represents how you would do ‘overall’ if assigned to CBT or control group. By this measurement a difference (statistically significant) was demonstrated in favour of CBT. I find this a somewhat unusual representation of the data. Does it not make sense to present discrete findings from end of intervention and end follow up?

By my interpretation the overall treatment effect was approximately 17% change at 9 months and 20% at 18 months. This change from a ‘moderately severe’ baseline score on the PANSS has been described as ‘minimal clinical improvement’.

A common measure of ‘good response’ is a 50% reduction in PANSS from baseline. The authors state that 7 of those receiving CBT reached this point versus 3 in the control group. They do not comment on the statistical significance of this finding. However by my ‘back of an envelope’ calculation I did not find this to be statistically significant.

The secondary outcome measures from the trial included a personal rating of ‘recovery’ using service-user defined measures. No difference between groups was developed on this measure.

There was a statistically significant difference between groups on a ‘personal and social performance scale’ (sounds terrifying…). This difference was ~5 points on a 100 point scale. I am not familiar enough with the rating scale to accurately comment on the clinical significance of the finding.

Concluding remarks…

This paper has generated, rightly, quite a lot of media interest. Of interest BBC have toned down their interpretation over the weekend.

I think the authors need to be congratulated on conducting the research in this neglected group of people. I think their findings have been over represented – but that only highlights a need for more research.

Some people have been commenting on this as a CBT vs antipsychotics finding. You can’t make any such conclusion from this trial – there was no form of true control group, and no head-head comparison can be made.

There were 2 deaths during the 18 month duration of the trial. The authors state

two deaths, both of which were deemed unrelated to trial participation or mental health

Two deaths out of 74 participants over 18 months with an average age of 30 years. To me that seems high but I haven’t found suitable data to compare with this population.

While I’ve described myself as critical of CBT I want to emphasise that, in keeping with my comment above! I am not claiming superiority for any other treatment modality. As I have said I have concerns regarding antipsychotics as well, particularly some of the ‘evidence base’ that supports their use.

Advertisements

2 thoughts on “Schizophrenia, antipsychotics and CBT

  1. “There was no ‘placebo’ or active control, for example through befriending offered. The gap between intervention and treatment as usual is therefore quite substantial and a large non-specific placebo response can be expected to bias results towards the CBT intervention.”

    Yes, and it wasn’t a multi-centre trial with N=250. But a first step surely?

    “There was, in my opinion, quite a high loss of participants to follow up throughout the trial. I accept this is a difficult to reach group of people, but it raises some concerns for me about the participant experience in both trial arms. The resulting low numbers of participants makes the results difficult to interpret in my opinion”

    The trial report states the drop-out was pre-planned, so this doesn’t derandomise the sample. My understanding is that the main consequence is a reduction in power, but since all data from every time-point used, this is not a problem.

    I’m concerned that (a) James Coyne is deleting from his blog replies from people who disagree with him and (b) no-one in the relevant twitter sphere is raising this.

    • Thanks for the comments. I agree this trial is a first step in an under-supported group and therefore to be lauded. Agree impact on power but, likely because I’m no statistician, I’m concerned that use of data at each time point adds a level of complexity versus comparison at discrete time points?

      I think we need more process based research with clear qualitative and service user involvement to better inform care planning.

      I’m sorry you feel others are deleting replies – I obviously can’t comment on this. This post represented my interpretation of the study – and is obviously therefore open to disagreement.

      Thanks for taking the time to read and comment. Disagreement and argument welcome.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s