Six trials were included (1047 patients).
Inspection of individual trials suggested the following potential heterogeneity differences: use of differing patient inclusion criteria; inclusion of one trial with mean dose exceeding the reviews inclusion criteria; use of different rating scales; differing baseline rates of anticholinergic medications; differing mean number of previous hospitalizations; and differing percentages of males.
Clinical efficacy: the clinical response rate was significantly higher in patients receiving risperidone. Mean difference 13.9% (95% CI: 5.6%, 22.3%; P < 0.05). No evidence of statistical heterogeneity was found.
Rate of anticholinergic medication: the rate of anticholinergic medication prescription was significantly lower in patients receiving risperidone. Mean difference 17.7% (95% CI: 9.4%, 25.9%; P < 0.05). No evidence of statistical heterogeneity was found.
Drop-out rate: the drop-out rate was significantly lower in patients receiving risperidone. Mean difference 12.7% (95% CI: 4.3%, 21.2%; P < 0.05). No evidence of statistical heterogeneity was found.
Sensitivity analyses:
Limiting analyses to trials using a 10 mg haloperidol dose: results were similar although point estimates for the difference were reduced. Mean difference in clinical response = 8.1% (95% CI: 1.3%, 14.8%). Mean difference in anticholinergic use = 12.2% (95% CI: 6.0%, 18.3%). Mean difference in drop-outs = 7.9% (95% CI: 1.8%, 14.0).
Including 3 trials excluded from the initial analyses: the authors report that results still favoured risperidone, though some fail to reach statistical significance. Mean difference in clinical efficacy = 8.3% (95% CI: -0.7%, 17.2%). Mean difference in anticholinergic medication = 16.6% (95% CI: 10.0%, 23.6%). Mean difference in drop-outs = 7.8% (95% CI: 0.9%, 14.8%).