Evaluating stroke recognition and prehospital quality metrics in emergency dispatch: a retrospective cohort study comparing mimics and confirmed strokes

Objectives Accurate identification of acute stroke by the Emergency Medical Dispatch Centre (EMDC) is essential for timely treatment and efficient resource use. However, stroke mimics—conditions resembling stroke can often lead to overtriage, complicating emergency care. We aimed to quantify EMDC stroke overtriage and compare characteristics of stroke mimics with confirmed stroke cases. Design Retrospective cohort study. Setting Patients with suspected stroke identified by the EMDC in Western Norway between 1 January 2021 and 31 December 2022. Participants A total of 8,261 patients with EMDC suspected stroke were included. Stroke suspicion was determined using the Norwegian Medical Priority Dispatch System, which includes 11 stroke-related criteria. The study population was stratified into stroke mimics (non-stroke diagnosis) and confirmed strokes (transient ischaemic attack, acute ischaemic stroke or intracerebral haemorrhage discharge diagnosis) at hospital discharge. Data collected included dispatch codes, prehospital time metrics and hospital discharge diagnoses. Outcome measures Outcomes of interest were the positive predictive value (PPV) of stroke identification based on EMDC dispatch criteria, the association between specific dispatch criteria and confirmed stroke and differences in prehospital time metrics. Results Of 8,261 patients with an ambulance dispatch due to EMDC suspicion of stroke, only 1181 (14.3%) were confirmed strokes, resulting in a PPV of 14.3% (95% CI 13.6% to 15.1%). Five dispatch criteria—breathing problems, acute vertigo, hemianopsia, ataxia/confusion and headache—had stroke mimic rates exceeding 90%. The FAST (Face-Arm-Speech-Time) criteria were significantly associated with confirmed stroke after adjusting for age and sex (OR 3.55, 95% CI 1.55 to 8.07). EMS response times were similar between groups (median 10 min (IQR 7–14) vs 9 min (IQR 6–14); p=0.47), whereas stroke mimics had significantly longer on-scene times (median 16 min (IQR 9–24) vs 10 min (IQR 7–16); p<0.001). Conclusions EMDC stroke dispatch criteria demonstrated low PPV, reflecting substantial overtriage due to stroke mimics at the earliest stage of emergency assessment. FAST-based criteria were associated with improved discrimination for confirmed stroke. Confirmed stroke cases had shorter EMS on-scene times, suggesting more streamlined prehospital management. These findings highlight the operational impact of stroke mimics on emergency services and the need to refine dispatch strategies. Because sensitivity could not be assessed, overall diagnostic performance remains uncertain, and future studies should include all stroke presentations to evaluate system-level accuracy.