Evolution is part of the therapeutic process of therapeutic communities

EDITOR—Kisely bases his ideas of what comprises a therapeutic community on a model that has undergone considerable evolution and development over 50 years.1 The Henderson Hospital and its offshoots are based on a specific type of therapeutic community, which started the social psychiatry movement after the second world war. Although it still provides a robust and effective treatment, several other types of therapeutic community are now used in the treatment of severe personality disorder.

Perhaps the most radical departure is to run the programmes as day services, as occurs in Reading. Francis Dixon Lodge in Leicester does not use small group therapy, and the Cassell Hospital is unusual in incorporating individual psychotherapy in the programme. Research is clearly needed, but the methodological difficulties are formidable. The treatment is complex and volatile and is delivered in very different ways; patient drop out is often part of the clinical process, long term follow up is essential, and measuring robust and useful personality change is difficult.

Randomised controlled trials are being considered, for day units as well as the new Henderson units, but they will not answer the question “Is a different type of therapeutic community better than this one?” or “What is going on here that is important?” As each therapeutic community differs considerably from every other, findings from randomised controlled trials will be difficult to generalise. And as evolution in response to the administrative environment is part of the therapeutic process of these communities, they could not be considered therapeutic communities if they were exactly defined and not allowed to change.

A project funded by the National Lottery and covering 10 centres is under way to isolate the “active ingredients” across a range of therapeutic communities for severe personality disorder. It is being coordinated by the Association of Therapeutic Communities and should help to define meaningful outcomes and identify relevant processes. The methodology is path analytic structural equation modelling. This is much more complex than a randomised controlled trial and, in the hierarchy of acceptable evidence, is not far below it. The protocol is available at www.pettarchiv.org.uk/atc-protocol.htm.

In his commentary on the article Pelosi writes in a belittling way of people with personality disorder. They have nearly always suffered extreme disturbance of emotional development (usually neglect, trauma, abuse, or loss in early life) and find it difficult to establish and sustain a normal life. To describe deeply disturbed people in the way that he does is surely unprofessional.

Article did not do justice to available research data

EDITOR—Psychotherapy for severe personality disorder has certainly been neglected. Unfortunately, Kisely's article does not do justice to available research data and consequently comes to unwarranted conclusions about commissioning.1

In contrasting the outcome of research in therapeutic communities (inpatient) with dialectical behaviour therapy (outpatient) Kisely does not compare like with like. There are no adequate data on which to compare studies, and there are none on patients randomly assigned to treatment in therapeutic communities, in day hospitals, and as outpatients. Criticism is made of the Patuxent study on reoffending rates, but this is a blunt outcome measure with complex determinants which may not be appropriate in terms of health economics or psychosocial variables.

Kisely states that the best evidence is for commissioning dialectical behaviour therapy, but such therapy was not in fact a recommendation in the strategic review.2 Compared with his comments about therapeutic communities, his criticisms of the unreplicated studies are curiously muted. All subjects were female. Men may be harder to treat. He fails to mention that there were no differences on some measures and that differences disappeared at follow up. New findings from Seattle suggest that if treatment with selective serotonin reuptake inhibitors is considered, some of the early differences between groups disappear (H Heard, personal communication).

We have conducted a randomised trial of 38 patients with borderline personality disorder who were treated with a psychoanalytically informed intervention.3 The results show that treatment in a specialist setting is superior to treatment as usual. We doubt, however, that this will (or should) lead to a call for commissioning of psychoanalytically informed treatments. Commissioning decisions must not depend solely on unreplicated studies of a small number of patients simply because they are randomised trials.

We performed a systematic review of the effectiveness of psychotherapeutic treatment of personality disorders, taking into account case identification, comorbidity, randomisation, specificity and context of treatment, and meaningful outcome measurement.4 We concluded that the evidence neither suggests superiority of one therapy over another nor indicates which subgroup of patients should be offered psychotherapy as inpatients, day patients, or outpatients. Moderately effective treatments tend to be well structured, devote effort to enhancing compliance, have a clear focus, be theoretically coherent to both therapist and patient, be long term, encourage a powerful attachment relationship, and be well integrated with other services. Both treatment in a therapeutic community and dialectical behaviour therapy meet these criteria.

Effective commissioning for personality disorder remains difficult. The real problem lies in an unbiased assessment of the literature, identifying adequate funding for further research, and developing appropriate protocols measuring meaningful outcome.

Randomised controlled trials may not be best for studies of clinical situations

EDITOR—Kisely acknowledges the limits of evidence based medicine but presents an analysis that exemplifies the difficulties that can arise when this model is used without reference to its limitations.1

The author examines two studies of the outcome of treatment in a therapeutic community, in 235 and 137 patients, and compares them unfavourably with a trial of dialectical behaviour therapy in 39 patients. If it were not for the unavailability of this therapy in the United Kingdom he would like to draw purchasing lessons from this comparison. The most substantial reason for this conclusion is that the smaller study had a randomised controlled design whereas the larger studies had an observational design. It must also be noted that the effectiveness of dialectical behaviour therapy was studied in a sample confined to women aged 18-45 who had borderline personality disorder and were suicidal. Such patients constitute a small subset of all those with severe personality disorder.

The comparison shows some consequences of the rigid application of evidence based medicine's hierarchy of evidence, which grades studies according to design excellence and places randomised controlled trials at the head of the list.2 This approach favours certain interventions, notably brief treatments (which minimise dropout) and novel approaches (as otherwise control subjects can go elsewhere to obtain a desirable treatment to which they have not been randomised). In particular, too, a randomised controlled trial requires that patient variation be minimised in order that before and after comparisons are not too obscured by initial variation within treatment groups.

For these reasons, while the randomised controlled trial is rightly the gold standard for trial design, it has some essential inadequacies as the sole design for studies of clinical situations, a point that has been well made before.3 Alternative trial designs can yield valuable information and should be considered on their merits—something we think Kisely failed to do.

Researchers in psychotherapy are aware of these problems, and well designed studies are becoming available. They need, however, to be interpreted and compared with reference to the context of the interventions under scrutiny. Models and standards are useful but should be handled with care.

Author should have got the facts right

EDITOR—I wish to correct just one of the factual errors in Kisely's article on the limits of evidence based purchasing of psychotherapy for severe personality disorder—an error repeated in Pelosi's commentary on the article.1 The author states (twice) that the NHS Executive's strategic review of psychotherapy services in England recommended dialectical behaviour therapy for the treatment of personality disorder.2

As the lead author of the review report, I can assure readers that it did not. What we did say was that at that time (1996) the strongest evidence from randomised controlled trials of efficacy in treating borderline personality disorder was for dialectical behaviour therapy, and for treating avoidant personality disorder the strongest evidence was for social skills training in combination with cognitive techniques. The report also pointed out that the paucity of controlled treatment trials weakens recommendations, that this is a field in which rapid progress is to be expected over the next few years, and that there will be a lag between new clinical developments and results of trials.

The review emphasised that it is not desirable to purchase only a limited range of treatment packages, targeted at certain diagnostic groups, that meet a hypothetical efficacy criterion. This is because there is no good effectiveness evidence, one way or the other, for many clinical practices. When appropriate controlled treatment research has not yet been undertaken, the absence of evidence of efficacy is not evidence of ineffectiveness. This hardly adds up to a recommendation that dialectical behaviour therapy is the only treatment of choice for personality disorder.

Author's reply

EDITOR—These correspondents seem to have missed the purpose of my article. This was to explore the limits of evidence based purchasing, not to recommend one particular form of psychotherapy over another. The comparison of Henderson-type therapeutic communities with dialectical behavioural therapy was to illustrate the existence of other models for the treatment of severe personality disorders. Others include cognitive analytical therapy and the approaches of the Cassell Hospital or Francis Dixon Lodge.

The article clearly highlighted how none of the interventions for severe personality disorders was entirely satisfactory. No one has disagreed with the methodological issues I raised about the evaluation of the Henderson-type model or the lack of research evidence to provide central funds for further units. I also discussed several of the weaknesses of dialectical behavioural therapy, such as small study numbers and the relative lack of expertise in the United Kingdom. Bateman and Fonagy have added a few more.

The authors of two letters have criticised my article on the basis of anticipated research findings (Haigh) or publications in press (Bateman and Fonagy), making a response difficult. Bateman and Fonagy state that their forthcoming review takes into account case identification, comorbidity, randomisation, specificity of treatment, context of treatment, and meaningful outcome measurement. In that case, have they included the research from the Henderson Hospital in their review? If they have, how have they dealt with the methodological flaws I have highlighted? If they have not, what is the relevance to the issue of central funding for Henderson-type units?

Parry is right to say that the NHS strategic review highlighted how the paucity of controlled treatment trials weakened the recommendations it could make for the treatment of personality disorder. The fact remains that the review concluded that the strongest evidence for efficacy was for dialectical behavioural therapy, and that this was at variance from the Reed report, which favoured therapeutic communities.1 Indeed, therapeutic communities are not mentioned at all in the one paragraph that the strategic review devotes to the efficacy and effectiveness of psychotherapies for personality disorder (paragraph 4.2.5.9).2

Everyone agrees that more research is needed and that a wide range of treatment packages should be purchased. Why then has only the Henderson-type model received central funding? It is more appropriate to acknowledge openly the limits of evidence based medicine than to rely on flawed studies that give the illusion that evidence exists. Let us be honest about how, when, and if we use research evidence.