When does a therapeutic intervention become an accepted part of standard clinical practice? Is it when there is sufficient research evidence? But what constitutes ‘sufficient’? What about available resources and acceptability to patients? Do we have to wait until the National Institute for Clinical Excellence pronounces? A convincing evidence base for family work in schizophrenia (Reference KuipersKuipers, 2000) has existed for many years but has been poorly implemented (Reference Anderson and AdamsAnderson & Adams, 1996). Will cognitive–behavioural therapy (CBT) for psychosis suffer the same fate? Which professional group will champion such an implementation? The evidence for other psychological treatments is less robust. Psychoeducation may prolong time to relapse and improve insight but at the cost of increasing suicidal ideation (Reference Carroll, Fattah and ClydeCarroll et al, 1998). Personal therapy (Reference Hogarty, Kornblith and GreenwaldHogarty et al, 1997) may be of value but is contra-indicated for patients who are living alone in the community. Psychodynamic approaches are advocated (Reference Mace and MargisonMace & Margison, 1997) but most psychiatrists do not support their use in practice, owing to lack of evidence of efficacy.

Over the past decade CBT for schizophrenia has received a great deal of research attention. Recent reviews of its efficacy have been positive (Reference Rector and BeckRector & Beck, 2001; Reference Pilling, Bebbington and KuipersPilling et al, 2002), with the effect size (ES) in residual positive and negative symptoms being large at the end of therapy (ES=0.65) and with more gains over time (ES=0.93) (Reference Gould, Mueser and BoltonGould et al, 2001). The positive results in research settings are cost-effective (Reference Healy, Knapp and AstinHealy et al, 1998) and also appear to translate to clinical practice (Reference Turkington, Kingdon and TurnerTurkington et al, 2002). Benefits have been demonstrated predominantly for residual symptoms in adult community patients receiving an average of 20 sessions over 9 months, with follow-up 3–9 months later. The evidence does not yet support the implementation of CBT in relation to prodromes, first-episode schizophrenia, acute relapse, forensic patients with psychosis or those with comorbidity such as substance misuse, personality disorder or learning disability, nor for psychosis in adolescence and old age.

A SPECIFIC EFFECT OF CBT?

Does CBT really provide added value when compared with other non-specific psychological interventions? Dickerson (Reference Dickerson2000) examined 20 studies of CBT in schizophrenia and concluded that the effects of CBT were less apparent when compared with a psychological treatment that controlled for therapy time. A major criticism of CBT research in this area to date has been that the studies have all employed inadequately delivered psychological treatments for comparison. Investigator allegiance is also an issue in that the supportive therapy and befriending was not delivered by experts in that particular therapy modality but by cognitive therapists (Reference Paley and ShapiroPaley & Shapiro, 2002). In this way the non-specific factors were well controlled for but the quality of the control intervention could be improved further. A clearer understanding of the relative efficacy of CBT in schizophrenia will be apparent when it is compared with expert delivery of, for example, interpersonal therapy. It does seem that by short-term follow-up a clear and specific effect is apparent with CBT.

CONSIDERATIONS FOR FURTHER RESEARCH

Research needs to be explicit about patients' presentations and histories, the specific CBT intervention used, the expertise of the therapists and, most importantly, comparator group characteristics and interventions. Schizophrenia as a diagnosis pulls together a group of people who vary considerably in symptom presentation and course, number of hospital admissions, self-esteem, depression, suicidality, social/family support, response to medication, etc. Cognitive–behavioural therapy as a descriptor is similarly imprecise. At present the label is applied to a wide range of interventions, ranging from a few sessions of support and psychoeducation, through specific interventions for individual symptoms, to therapies that last 20–50 sessions and claim to be weakening not only the full range of positive and negative symptoms but also enduring models of self (schemata). Therapy might be delivered individually or in groups. Then there are therapists themselves: trials are not always using therapists who are accredited cognitive–behavioural therapists, let alone those who are accredited and practised with psychosis. What this points to is that there is no one CBT for schizophrenia. Rather, there are a range of CBT therapies, some with different underlying theoretical models, offered to a diverse group of patients by therapists who vary in ability and experience. Beyond this, one way forward is for researchers to be concerned not only with the effectiveness of ‘CBT for schizophrenia’ but also with addressing more specific questions. What are the common, defining hallmarks of any CBT for psychosis? What are the active ingredients? Case formulation, for example, is frequently described as crucial to CBT for psychosis, but there is no empirical evidence to support this claim (Reference Chadwick, Williams and MackenzieChadwick et al, 2003). Encouragingly, research is beginning to address applications of exciting new cognitive theoretical models (Reference Gumley, White and PowerGumley et al, 1999; Reference Garety, Kuipers and FowlerGarety et al, 2001). Pragmatic trials with process measures will allow further clarification of the crucial elements of CBT for psychosis in clinical practice. Also, attempts are being made to clearly describe patients and to use qualitative methods to assess more of the patient's own experience of psychosis and of therapy.

IS PRACTICE RUNNING AHEAD OF EVIDENCE?

Cognitive–behavioural therapy for schizophrenia is increasingly being included in the basic training of psychiatrists, psychologists and nurses. Certainly, policy has not been slow to endorse it with inclusion in the National Health Service Psychotherapy Review (NHS Executive, 1996) and the National Service Framework for Mental Health (Department of Health, 1999). It has now been taken up internationally, with studies and reviews published from Malaysia, Holland, Scandinavia and North America, but is this enough to lead to its formal acceptance? A recent update by the Cochrane Collaboration of their review of CBT in schizophrenia (Reference Cormac, Jones and CampbellCormac et al, 2002) concluded that it was a promising but underresearched intervention and not yet recommended for clinical practice (although one of the authors declared that he uses it in his clinical practice). The American Psychiatric Association has yet to update its schizophrenia guidelines that concluded similarly, although this was prior to the recent publication of more definitive studies. Protocols for research need to be tightened significantly in relation to valid comparator treatments, therapy definition and fidelity, protection of blinding of raters and long-term follow-up. Despite these provisos, there is now strong evidence and clinical support for the implementation of CBT as part of the standard management of patients with residual symptoms of schizophrenia. It is time for it to be used to begin to fill a therapeutic vacuum that is currently causing patients to experience unnecessary distress and disability.

DECLARATION OF INTEREST

D.T. has accepted hospitality from Pfizer, Lilly, Janssen and Organon. He has undertaken a consultancy for Pfizer, has sat on two advisory boards and has acted as principal investigator on a Pfizer-funded trial into the efficacy of CBT in schizophrenia in community settings. D.K. has received hospitality from Pfizer, Lilly and Janssen and has sat on an advisory board for Pfizer.