EEG Controlled Triage in the Ambulance for Acute Ischemic Stroke

Overview

Endovascular thrombectomy (EVT) is the standard treatment for patients with a large vessel occlusion (LVO) stroke. Direct presentation of patients with an LVO to a comprehensive stroke center (CSC) reduces onset-to-treatment time by approximately an hour and thereby improves clinical outcome. However, a reliable tool for prehospital LVO-detection is currently not available. Previous electroencephalography (EEG) studies have shown that hemispheric hypoxia quickly results in slowing of the EEG-signal. Dry electrode EEG caps allow reliable EEG measurement in less than five minutes. We hypothesize that dry electrode EEG is an accurate and feasible diagnostic test for LVO in the prehospital setting. ELECTRA-STROKE is a diagnostic pilot study that consists of four phases. In phases 1, 2 and 3, technical and logistical feasibility of performing dry electrode EEGs are tested in different in-hospital settings: the outpatient clinic (sample size: max. 20 patients), Neurology ward (sample size: max. 20 patients) and emergency room (sample size: max. 300 patients), respectively. In the final phase, ambulance paramedics will perform dry electrode EEGs in 386 patients with a suspected stroke. The aim of the ELECTRA-STROKE study is to determine the diagnostic accuracy of dry-electrode EEG for diagnosis of LVO-a stroke when performed by ambulance personnel in patients with a suspected AIS. Sample size calculation is based on an expected specificity of 70% and an incidence of LVO stroke of 5%.

RATIONALE Endovascular thrombectomy (EVT) is standard treatment for acute ischemic stroke (AIS) if there is a large vessel occlusion in the anterior circulation (LVO-a). Because of its complexity, EVT is performed in selected hospitals only. Currently, approximately half of EVT eligible patients are initially admitted to hospitals that do not provide this therapy. This delays initiation of treatment by approximately an hour, which decreases the chance of a good clinical outcome. Direct presentation of all patients with a suspected AIS in EVT capable hospitals is not feasible, since only approximately 7% of these patients are eligible for EVT. Therefore, an advanced triage method that reliably identifies patients with an LVO-a in the ambulance is necessary. Electroencephalography (EEG) may be suitable for this purpose, as preliminary studies suggest that slow EEG activity in the delta frequency range correlates with lesion location on cerebral imaging. Use of dry electrode EEG caps will enable relatively unexperienced paramedics to perform a reliable measurement without the EEG preparation time associated with 'wet' EEGs. Combined with algorithms for automated signal analysis, we expect the time of EEG recording and analysis to eventually be below five minutes, which would make stroke triage in the ambulance by EEG logistically feasible. HYPOTHESIS We hypothesize that EEG accurately identifies the presence of an LVO-a stroke in patients with a suspected AIS when applied in the ambulance. OBJECTIVE To determine the diagnostic accuracy of dry-electrode EEG for diagnosis of LVO-a stroke when performed by ambulance personnel in patients with a suspected AIS. STUDY DESIGN This diagnostic study consists of four phases: Phase 1: Optimization of measurement time and software settings of the dry electrode cap EEG in a non-emergency setting in patients in whom a regular EEG is/will be performed for standard medical care. Sample size: maximum of 20 patients. Phase 2: Optimization of measurement time and software settings of the dry electrode cap EEG in patients close to our target population in a non-emergency setting. Sample size: maximum of 20 patients. Phase 3: Validation of several existing algorithms and development of one or more new algorithms for LVO-a detection, as well as optimization of logistics and software settings of the dry electrode EEG cap in patients close to our target population in an in-hospital emergency setting. Sample size: maximum of 300 patients. Phase 4: Validation of several existing algorithms and algorithms developed in phase 3 for LVO-a detection in patients with a suspected AIS in the ambulance, as well as assessment of technical and logistical feasibility of performing EEG with dry electrode caps in patients with a suspected AIS in the ambulance. Sample size: maximum of 386 patients. STUDY POPULATION Phase 1: Patients in the outpatient clinic of the Clinical Neurophysiology department of the AMC, in whom a regular EEG has been/will be performed for standard medical care. Phase 2: Patients with an AIS admitted to the Neurology ward of the coordinating hospital with an LVO-a (after reperfusion therapy). Phase 3: Patients with a suspected AIS in the emergency room (ER) of the coordinating hospital (before endovascular treatment). Phase 4: Patients with a suspected AIS in the ambulance. INTERVENTION Performing a dry electrode cap EEG (in phase 1 in the outpatient clinic, in phase 2 during hospital admission, in phase 3 in the ER and in phase 4 in the ambulance). MAIN END POINTS Primary end point: the diagnostic accuracy of dry electrode cap EEG to discriminate LVO-a stroke from all other strokes and stroke mimics in the prehospital setting (study phase 4) expressed as the area under the receiver operating characteristics (ROC) curve of the theta/alpha ratio. Secondary end points: * Sensitivity, specificity, PPV and NPV of the theta/alpha ratio, and test characteristics of other existing EEG data based algorithms for LVO-a detection (e.g. Weighted Phase Lag Index, delta/alpha ratio); * Logistical and technical feasibility of paramedics performing dry electrode cap EEG in the ambulance in suspected AIS patients; * Developing one or more novel EEG data based algorithms with an optimal diagnostic accuracy for LVO-a detection in suspected AIS patients with ambulant dry electrode cap EEG.