Sir: Bebbington's conclusion that the new atypical antipsychotics are no more effective in reducing psychotic symptoms than their older counterparts (Psychiatric Bulletin, August 2001, 25, 284-286) may not apply to one of these drugs, clozapine.

Clozapine was re-introduced in 1989 on the basis of repeated indications of therapeutic superiority, which culminated in the Kane et al trial (Reference McKenna and BaileyMcKenna & Bailey, 1993), not, as Bebbington suggests, as part of a strategy to develop drugs without extrapyramidal side-effects. Supporting this, the meta-analysis of Geddes et al (Reference Geddes, Freemantle and Harrison2000) found the effect size for clozapine's effectiveness over conventional neuroleptics to be 0.68, which falls between the values of 0.5 and 0.8 proposed by Cohen for ‘moderate’ and ‘large’, respectively. This is difficult to reconcile with Bebbington's statement that “the meta-analysis indicated that some of the atypical antipsychotics had slightly better efficacy”. Geddes et al (Reference Geddes, Freemantle and Harrison2000) argued that the apparent superiority of atypical neuroleptics was owing to the high dose of comparison drug used in many of the studies. However, clozapine was the atypical neuroleptic in only 12 of the 30 studies included in their two meta-regressions. When the Cochrane Collaboration (Reference Wahlbeck, Cheine and EssaliWahlbeck et al, 1999) compared clozapine trials using low doses and standard doses of the comparison drug, no difference in clinical improvement, relapse rate or drop-outs was found.