We generally concur with the views of Silver et al and we continue to collaborate with those at Bristol-Myers Squibb and Otsuka to obtain data which were not easily accessible to us when the review was initially conducted. The updated version of this review is much improved by the incorporation of these data (Reference El-Sayeh and MorgantiEl-Sayeh & Morganti, 2006). The original version, however, was submitted to the Journal in June 2004.

We were interested that the review fell short of Professor Crow's expectations. Perhaps he is correct in saying that there is a grumbling background to the whole review but it was peer reviewed and there was no objection to this. Professor Crow was surprised that our searches came up with such a ‘barren yield’ of data. Perhaps his experience in this area is not ours. We asked employees of Bristol-Myers Squibb to check our review. Those that kindly visited us and promised data are explicitly mentioned in widely accessible versions of this review (Reference El-Sayeh and MorgantiEl-Sayeh & Morganti, 2004). Other authors referred us to the company for additional information. Professor Crow goes on to say that it is the duty of systematic reviewers to make data available in comparative form. We have tried to be fair, open and explicit with what we could get. If Professor Crow can get more data we will of course be grateful for those.

Professor Crow draws attention to aripiprazole and mortality as presented in a poster at the Winter Workshop on Schizophrenia Research in February 2006. After the publication of our paper in the Journal we obtained clarification from Bristol-Myers Squibb regarding the eight deaths. This information was forthcoming precisely because of the poster presentation in 2006. Two years earlier we had met with representatives of the company and asked for conformation of our results before publication in the Cochrane Database of Systematic Reviews and a note of this meeting is made (Reference El-Sayeh and MorgantiEl-Sayeh & Morganti, 2004). The offer of clarification and further contact did not materialise until after the poster presentation. Thereafter Bristol-Myers Squibb showed us how we had indeed failed to note how these people had died in the post-randomisation protracted follow-up of the two studies in question, so normalising the seemingly alarming standardised mortality ratio previously presented (Reference El-Sayeh and MorgantiEl-Sayeh & Morganti, 2004). We do not think anyone would say that these data were easy to locate or are clear (Reference Dubitsky, Harris and LarenDubitsky et al, 2002), although Professor Crow may think otherwise.

As mentioned in our review, currently available data do not seem to support the prolific use of aripiprazole. In suggesting otherwise, there may be a danger of giving false hope to clinicians and recipients of care.