Diagnostic Accuracy of Lateral Flow Blood Tests to Detect Large Vessel Occlusion Stroke

BACKGROUND: A rapid and accurate tool to identify large vessel occlusion stroke (LVO) for use by prehospital emergency medical services responders could support direct access to time-critical thrombectomy at regional Comprehensive Stroke Centers. This study evaluated the accuracy of D-dimer and GFAP (glial fibrillary acidic protein) detected by lateral flow tests (LFT) to identify LVO stroke. METHODS: This hospital-based prospective observational cohort study recruited adults within 6 hours of onset of at least 1 face, arm, or speech test (FAST) symptom. The LFTs use fingerprick capillary blood and give an overall positive (ie, LVO stroke predicted) or negative LFT outcome. Independent adjudication of brain imaging and clinical data established LVO or non-LVO. Specificity, sensitivity, negative and positive predictive values were calculated for (1) LFT outcome for participants with >1 FAST symptoms (primary analysis population); (2) a combined decision rule where participants with 1 FAST symptom were rule negative, patients with >1 FAST symptoms rule positive or negative according to LFT outcome; (3) FAST symptoms alone split by 1 symptom versus >1 symptom. RESULTS: The study involved 382 participants (age [mean], 72.4 years; female, 50.8%; National Institutes of Health Stroke Scale score [median], 6; time from onset to LFT [median], 130 minutes). Of 278 participants who met the primary analysis population, 202 of 278 had both LFT outcome and LVO/non-LVO assigned. Analysis gave specificity 79% (95% CI, 72%-85%), sensitivity 53% (95% CI, 39%-67%), negative predictive value 86% (95% CI, 79%-90%), and positive predictive value 42% (95% CI, 30%-55%). For 382 participants with 1 to 3 FAST symptoms, 290 of 382 had both a combined decision rule result and LVO/non-LVO assigned giving specificity 86% (95% CI, 82%-90%), sensitivity 51% (95% CI, 37%-65%), negative predictive value 90% (95% CI, 86%-93%), positive predictive value 42% (95% CI, 30%-55%). FAST symptoms alone (n=318) gave specificity 33% (95% CI, 27%-38%), sensitivity 96% (95% CI, 87%-99%), negative predictive value 98% (95% CI, 92%-99%), positive predictive value 23% (95% CI, 18%-29%). CONCLUSIONS: LFT outcome combined with FAST symptoms gave high specificity for LVO identification. This approach could be considered for facilitation of direct access to thrombectomy providers. However, further research is required to evaluate test performance in the prehospital setting and consider technological improvements to boost sensitivity without compromising specificity. REGISTRATION: URL: https://www.isrctn.com/; Unique identifier: ISRCTN12414986.