Professor David highlights important points in relation to the trial of cognitive therapy for the prevention of psychosis. He asks for clarification regarding the exclusion of two participants for having been psychotic at inception, but only reporting this on second contact with an assessor. This was certainly what happened, and following this the research assistants were instructed to ask all participants about this. This was not in the original protocol for the study, as such an event was unexpected (although, with hindsight, maybe it should not have been). It did seem reasonable to exclude these participants, especially given that the study is the first of its kind (clearly future studies should address this issue in the protocol).

He also raises the issue of randomisation. The procedure for randomisation is very clearly outlined within the original paper and the difference in gender rates was due to chance. It is true that this method resulted in more of the treatment group being female, which is an indicator of better outcome for such a population; however, the method also resulted in the treatment group having a higher proportion of people who were unemployed and a significantly higher level of baseline positive symptoms, both of which would predict poorer outcomes for the treatment group. It is also worth noting that gender was utilised as a covariate in all analyses regarding transition that were reported.

Professor David has identified two important issues that can only be clarified by replication of the results of this study with a more rigorous protocol and an alternative method of randomisation; we would agree that such research is required to determine whether the preventive effects of cognitive therapy with people at ultra-high risk of developing psychosis are generalisable and robust.