Background:The possibility that low-dose antipsychotic treatment is associated with increased risk of cerebrovascular events (CVEs) in elderly patients with dementia has been raised. The objective was to determine whether risperidone is associated with an increased risk of CVEs relative to other commonly considered alternative treatments.

Methods:An analysis of Medicaid data from 1999 to 2002, representing approxi-mately 8 million enrollees from multiple states, was conducted. The primary outcome was the incidence of acute inpatient admission for a CVE within 3 months following initiation of treatment with atypical antipsychotics (risperidone, olanzapine, quetiapine, or ziprasidone), haloperidol, or benzo-diazepines.

Results:Descriptive analyses found similar rates of incident CVEs across evaluated agents. Multivariate analyses found no differences in comparisons of risperidone with olanzapine or quetiapine. Risperidone and other antipsychotics as a group were also not associated with a higher odds ratio (OR) of incident CVE than either haloperidol or benzodiazepines. With risperidone as the reference group: olanzapine, OR=1.05, 95% CI 0.63–1.73; quetiapine, OR=0.66, 95% CI 0.23–1.87; haloperidol, OR=1.91, 95% CI 1.02–3.60; benzodiazepines, OR=1.97, 95% CI 1.30–2.98. With benzodiazepines as the reference group, the OR of incident CVE for all antipsychotics as a class was 0.49, 95%CI 0.35–0.69.

Conclusions:This study found no significant difference in the incidence of CVEs between patients taking risperidone and those taking other atypical antipsychotics. Risperidone and all atypical antipsychotics were not associated with higher risk than two common treatment alternatives (haloperidol and benzodiazepines). These findings do not support the conclusion that risperidone is associated with a higher risk of CVE than other available treatment alternatives. The data also suggest that patient characteristics other than antipsychotic use are more significant predictors of CVEs. Given the relatively low rates of incident CVEs, a larger sample of patients with groups closely balanced on a wide spectrum of potential risk factors could provide a more precise assessment of risk.