Abstract Background and aims Delirium is a frequent complication of acute stroke and is associated with poorer outcomes. Despite widespread antipsychotic use for delirium-related agitation in stroke units, stroke-specific guidance remains limited. We aimed to evaluate how antipsychotics are currently used for delirium in acute stroke, and to examine the alignment between available evidence and international guideline recommendations across acute ischaemic stroke (AIS), intracerebral haemorrhage (ICH), and subarachnoid haemorrhage (SAH). Methods A narrative review was conducted via structured PubMed searches (2000-2025) for studies on delirium or agitation in adult AIS, ICH, or SAH patients. Randomised trials, observational studies, and international guidelines were reviewed. Data were extracted on delirium phenotype, antipsychotic indications, safety, and clinical outcomes. Results Stroke-specific evidence supporting antipsychotics as disease-modifying treatment is scarce. Studies are predominantly observational, often using antipsychotic prescription as a marker of agitation severity. Major stroke guidelines (ESO, AHA/ASA) lack explicit pharmacological recommendations for delirium, prioritising non-pharmacological strategies and avoiding oversedation. In the absence of stroke-specific guidance, NICE permits short-term, low-dose haloperidol only for severe distress or safety risks. In neurocritical care, dexmedetomidine or atypical antipsychotics are frequently used for severe agitation under close physiological monitoring. Safety concerns, including QTc prolongation and arrhythmic risk, are critical given high cardiac comorbidity in stroke patients. Benzodiazepines are generally discouraged outside alcohol or sedative withdrawal states. Conclusions Antipsychotics can manage severe agitation in acute stroke despite lacking robust efficacy data. Current recommendations restrict pharmacological treatment to high-risk hyperactive delirium. Prospective, stroke-subtype-stratified studies are required to define the benefit-harm balance across AIS, ICH, and SAH. Conflict of interest Ahmed Abdalla: nothing to disclose, Jonathan Vince: nothing to disclose
Abdalla et al. (2026) studied this question.