McKenna criticises the basis on which Bebbington included clozapine in his conclusion that the newer neuroleptics had little therapeutic advantage over their older counterparts. While he may be right to conclude that clozapine is especially effective, our metaregression (Geddes et al, 2000) did appear to apply equally to all atypicals.

Part of the problem with a correct evaluation of the effectiveness of clozapine is that it is largely based on studies involving patients known to be resistant to treatment with conventional neuroleptics. Greater effectiveness in this context may merely be a reflection of effectiveness in a different client group. Equally, conventional neuroleptics would be almost bound to do better in a group of patients unresponsive to clozapine. Furthermore, there is now evidence that clozapine has no advantage over conventional neuroleptics in unselected patients with first episodes (Reference Lieberman, Phillips and KongLieberman et al, 2001).

This is not to say that the clozapine lacks utility in people unresponsive to other medication, but it does bear on the claim that it should be a first-use drug.