Abstract

<p><b>Background:</b> The diagnosis of stroke or TIA in the emergency department is difficult, though important for early treatment. Circulating biomarkers might improve upon clinical assessment at admission.</p> <p><b>Methods:</b> We recruited symptomatic patients with suspected stroke or TIA and drew blood soon after admission. Each patient was assessed with the Face Arm Speech Test (FAST). We measured a panel of 15 circulating inflammatory, thrombotic, cardiac, and cerebral tissue damage biomarkers. Improvement in diagnostic performance was assessed by adding biomarkers to the FAST in logistic regression models to predict a final diagnosis of stroke or TIA (verified by expert review and imaging).</p> <p><b>Results:</b> 405 patients had suspected stroke: 285 with TIA or stroke (230 definite or probable ischemic stroke, 40 TIA, 15 hemorrhagic stroke) and 120 with other diagnoses. Only the markers t-PA and NT-proBNP were associated positively and significantly (p < 0.01) with a diagnosis of TIA or stroke. The FAST had a sensitivity of 82% (95% CI 78-87) and specificity of 38% (95% CI 29-46) for the diagnosis of TIA or stroke. No biomarker individually improved the sensitivity or specificity of the FAST. A model containing the FAST, age, systolic blood pressure, NT-proBNP and t-PA had a better sensitivity (88%, p < 0.006) and a better specificity (48%, p = 0.04) than the FAST test alone.</p> <p><b>Conclusions:</b> No single blood marker improved the diagnostic performance of a validated clinical stroke scale. Panels of biomarkers may marginally improve diagnosis, but their practicability is uncertain, and requires further study.</p>