Background: We aimed to determine the proportion of patients who had suffered a stroke and compare this to those patients with suspected stroke, and the range of differential diagnosis for suspected stroke.
Methods: We searched for prospective studies of suspected stroke in electronic databases and our personal files. We undertook a meta-analysis of these studies, aimed at determining the proportions of patients with confirmed stroke in different settings.
Results: We identified 29 studies involving 8,839 patients: 13 studies were from emergency departments, five from stroke units or transient ischaemic attack (TIA) clinics, three from primary care, three from ambulance services and five were unspecified. About three-quarters (74% [95% confidence interval (CI): 66 to 83%]) of patients had a diagnosis of stroke, though there was significant heterogeneity in this estimate. The five most frequent non-stroke diagnoses were seizure, syncope, sepsis, migraine and brain tumours.
Conclusion: Patients who had not had a stroke accounted for a significant proportion of people referred to stroke services. Expertise in the differential diagnoses of stroke is needed in order to manage the patients at the point of referral.
Keywords: Stroke; diagnosis; epidemiology; systematic review.