Our study was designed specifically to answer the question raised by Jones et al (Reference Jones, Cormac and Mota1999) of whether the benefits achieved by expert therapists in research settings could be replicated by non-expert therapists working in community mental health teams. An end-of-therapy comparison was therefore necessary with one of the methodologically robust studies quoted in the above review. Kuipers et al (Reference Kuipers, Garety and Fowler1997) was chosen because a similar, good clinical outcome analysis on overall symptoms had been reported at end of therapy. The appropriate end-of-therapy comparison is 14/28 (50%) for cognitive—behavioural therapy (CBT) as measured at the level of 20% improvement in overall symptoms in the original Kuipers et al (Reference Kuipers, Garety and Fowler1997) paper compared with 112/257 (44%) as measured at the level of a 25% improvement in our study. These results show a comparable effect size for CBT in the two studies, considering that our study had to satisfy a more stringent criterion for a good clinical outcome. The difference in the numbers needed to treat is solely due to an improved performance in our treatment as usual group compared with standard care.

It is certainly correct to state that the two study populations were different by definition. However, consideration of the demographics as reported in the two papers shows that there was little difference in those who actually ended up being enrolled in the two studies. The mean number of admissions in Kuipers et al (Reference Kuipers, Garety and Fowler1997) was 5.2 for the CBT group and 4.3 for standard care and in our study 4.71 for CBT and 5.18 for treatment as usual. We ended up enrolling a more treatment-resistant group because of the fact that patients with schizophrenia whose symptoms were well controlled with medication often did not see the need to enter the study when it was offered to them.

It is certainly true that the CBT delivered by Kuipers and colleagues was of 20 sessions' duration with a more sophisticated treatment manual. This makes the result of our brief CBT intervention as delivered by psychiatric nurses all the more impressive. We await the analysis of our short-term follow-up results to see whether the impressive durability results reported above can be equalled. If CBT is to make a real impact in terms of the management of schizophrenia, it will need to be delivered by non-expert therapists in community mental health teams. The real issues for expert cognitive therapists are to organise training courses, provide supervision and to deliver more complex CBT for those patients with schizophrenia who are more psychologically difficult or who have comorbidity such as post-traumatic stress disorder, alcohol dependence and social phobia. There is therefore a potential role for both expert and non-expert therapists in the management of every patient with schizophrenia.