Older adults with dementia who were prescribed risperidone faced a higher risk of stroke regardless of whether they had pre-existing heart disease, according to a population-based matched cohort study of approximately 165,000 persons published in The British Journal of Psychiatry. The finding eliminates what many clinicians had hoped for: a clearly identifiable low-risk group that could safely take the widely prescribed antipsychotic. With no subgroup spared, the study sharpens a clinical dilemma that has been building since regulators first flagged cerebrovascular dangers more than two decades ago.
Why risperidone’s stroke signal matters for millions of dementia patients
Risperidone is one of the most commonly prescribed antipsychotics for managing agitation, aggression, and psychosis in people living with dementia. Behavioral symptoms affect a large share of dementia patients at some point during their illness, and risperidone often becomes the default pharmacological option when non-drug approaches fall short. The new cohort data, drawn from administrative health records and covering roughly 165,000 matched patients, showed increased stroke risk associated with risperidone overall and across every subgroup examined. That consistency is what sets this analysis apart from smaller trials that left room for the argument that only patients with cardiovascular disease were vulnerable.
One hypothesis worth testing against the evidence is whether risperidone’s stroke risk stems primarily from acute vascular effects, such as blood-pressure swings, rather than from cumulative drug exposure or the progression of neurodegeneration itself. A separate self-controlled case series study offers partial support for the acute-effect theory: rate ratios for stroke were substantially elevated during active antipsychotic treatment periods and were higher in people with dementia than in those without. Because a self-controlled design compares each patient’s risk during exposed and unexposed windows, it controls for stable individual characteristics like genetic predisposition or baseline vascular health. The fact that risk climbed sharply during treatment, rather than accumulating gradually over years of prescriptions, points toward an acute pharmacological trigger. Still, the available studies do not report dose-response curves or treatment-duration breakdowns detailed enough to confirm that mechanism conclusively.
Regulatory warnings and trial data behind the stroke signal
The stroke concern did not emerge in a vacuum. In April 2005, the FDA issued a Public Health Advisory after reviewing 17 short-term randomized placebo-controlled trials involving 5,106 patients. Those trials covered four atypical antipsychotics: olanzapine, aripiprazole, risperidone, and quetiapine. The advisory focused on increased mortality, but the underlying trial data also contained cerebrovascular adverse-event signals that had already prompted label changes in several countries. Health Canada issued parallel warnings around the same period.
Research published in The New England Journal of Medicine extended the safety picture beyond atypical drugs, establishing that conventional antipsychotics carried at least comparable risks in elderly populations. That finding closed off a common clinical workaround: switching patients from newer atypical agents to older conventional ones in the hope of avoiding the cerebrovascular danger. Earlier population-based retrospective cohort work had already documented the stroke association in older adults, showing that the problem predated the large 165,000-person study and was not limited to a single drug class or dataset.
Taken together, the evidence forms a consistent pattern across study designs. Randomized trials flagged the mortality and cerebrovascular signals. Self-controlled case series data isolated the risk to active treatment windows. Retrospective cohorts confirmed the association in real-world prescribing. And the matched cohort study demonstrated that the stroke risk persisted in patients both with and without comorbid cardiovascular disease, removing the last obvious clinical filter for identifying who could be prescribed risperidone more safely.
Unanswered questions about dose, duration, and safer alternatives
Several gaps in the evidence prevent clinicians from translating these findings into precise prescribing rules. The matched cohort study reports increased risk across subgroups but, outside of the abstract, does not provide granular hazard ratios, confidence intervals, or subgroup counts in easily accessible form. Without those details, it is difficult to quantify how much the risk rises for a patient with no cardiovascular history compared with one who has atrial fibrillation or prior transient ischemic attacks. Long-term follow-up outcomes and detailed dose-response data are also limited in the primary publications, leaving open the question of whether very low doses or very short courses carry meaningfully less danger.
No direct patient-level data or raw claims records from the underlying administrative database have been released for independent replication. That lack of open data makes it harder for external researchers to probe potential effect modifiers such as concomitant medications, frailty scores, or markers of small-vessel disease. It also constrains efforts to explore whether certain prescribing patterns-such as rapid dose escalation or frequent drug switching-are especially hazardous.
From a practical standpoint, clinicians are left with a blunt message: for older adults with dementia, there is no clearly identifiable subgroup in which risperidone appears cerebrovascularly safe. Yet they must still balance this risk against the harms of untreated severe behavioral and psychological symptoms, which can include physical injury, caregiver burnout, and loss of placement in long-term care facilities. In the absence of precise risk stratification, guidelines increasingly emphasize non-pharmacologic interventions as first-line management, reserving antipsychotics for situations where symptoms are dangerous, highly distressing, or resistant to other strategies.
Potentially safer pharmacological alternatives remain under active study, but none have been shown conclusively to eliminate cerebrovascular risk in this population. Other atypical antipsychotics share similar warnings, and conventional agents appear no better. Agents outside the antipsychotic class, such as certain antidepressants or anticonvulsants, may help some patients but bring their own adverse-effect profiles and far less robust data on stroke outcomes. For now, the risperidone findings are best interpreted as a class-level caution rather than an isolated problem with a single molecule.
Implications for shared decision-making and policy
For families and caregivers, the message is not that risperidone should never be used, but that any decision to start or continue it deserves a frank discussion about stroke risk and realistic benefit expectations. Clinicians may need to revisit long-standing prescriptions, especially in patients whose behavioral symptoms have stabilized, to consider gradual tapering and closer monitoring. Where antipsychotics are deemed necessary, strategies such as using the lowest effective dose, limiting treatment duration, and reassessing regularly for ongoing need become even more important, even though the evidence base does not yet define exact thresholds.
At a policy level, the absence of a low-risk subgroup strengthens arguments for tighter oversight of antipsychotic use in long-term care settings. Quality metrics that track prescribing rates, coupled with support for staff training in non-pharmacologic behavior management, could help shift default practice away from routine antipsychotic initiation. Regulators may also face renewed pressure to require post-marketing studies that report more detailed cerebrovascular outcomes, including dose and duration effects, and to encourage data sharing that allows independent analysis.
Ultimately, the matched cohort findings do not solve the central dilemma of treating behavioral symptoms in dementia, but they do clarify its boundaries. There is no easy way to identify patients for whom risperidone is free of stroke risk. Instead, clinicians, patients, and caregivers must navigate a landscape where every option carries trade-offs, and where careful, individualized judgment-and ongoing research into safer approaches-remains essential.
More from Morning Overview
*This article was researched with the help of AI, with human editors creating the final content.
