In the course of a favourable review of cognitive therapy in schizophrenia, Thornicroft & Susser (Reference Thornicroft and Susser2001) cite the recent trial by Sensky et al (Reference Sensky, Turkington and Kingdon2000), but fail to mention that it had negative results. This 90-patient, 9-month randomised controlled trial, carried out under blind conditions, compared this form of treatment with a control intervention (befriending) and found no significant difference between the two. It is true that differences emerged 9 months after completion of treatment, but this latter part of the study was uncontrolled.

Of the other trials of cognitive therapy cited in their article, that of Drury et al (Reference Drury, Birchwood and Cochrane1996) did not use blind evaluations, and that of Kuipers et al (Reference Kuipers, Garety and Fowler1997) employed neither blind evaluations nor a condition to control for the non-specific effects of intervention (the Hawthorne effect). Only one other published study (Reference Tarrier, Yusupoff and KinneyTarrier et al, 1998) incorporated both these design features; this found a non-significant advantage of cognitive therapy over supportive counselling (Reference CurtisCurtis, 1999).

Rather than being ready for an assessment of its effectiveness and cost-effectiveness in non-experimental settings, as Thornicroft & Susser argue, cognitive therapy may be in the process of meeting the fate of an earlier treatment for schizophrenia where advocacy preceded rigorous evaluation — insulin coma.