Seven studies (n=1,746) met the inclusion criteria, of which three also included placebo comparison groups.
Treatment failed in 50% of the risperidone-treated patients, compared with 66% of those treated with haloperidol and 83% of those treated with placebo. It would be necessary to treat 11 patients with risperidone to prevent one treatment failure in those patients treated with haloperidol; the odds ratio (OR) was 0.74 (95% confidence interval, CI: 0.58, 0.94, P=0.02). The three multicentre trials that included placebo as a treatment arm were pooled. The pooled results showed that one in three patients treated with risperidone (4 to 16 mg/day) and one in six treated with haloperidol (10 to 20 mg/day) would derive significant benefit; the ORs were 0.22 (95% CI: 0.13, 0.39, P<0.00001) and 0.44 (95% CI: 0.22, 0.84, P=0.02), respectively. Moreover, there was a highly significant greater need for anticholinergic medication due to extrapyramidal symptoms in the haloperidol-treated patients, compared with risperidone (OR 0.54, 95% CI: 0.42, 0.70, P<0.0001).