Video Head Impulse Testing for Acute Dizziness in Stroke Protocols

BACKGROUND AND OBJECTIVES: Dizziness is a common emergency department (ED) concern, yet posterior-circulation strokes can be difficult to detect with standard clinical tools and neuroimaging alone. Quantitative video head impulse test (vHIT) provides an objective bedside measure of vestibulo-ocular reflex function and can help differentiate central from peripheral causes of acute dizziness. METHODS: We conducted a prospective, single-center feasibility and diagnostic-accuracy study of adults (18-90 years) presenting on an ED stroke pathway for acute dizziness. vHIT was performed within 48 hours of presentation, after completion of standard stroke evaluation. Two vestibular-trained clinicians, blinded to clinical and imaging data, classified vHIT as peripheral, central, equivocal, or poor quality. The reference standard (stroke vs nonstroke) was the final diagnosis adjudicated by stroke neurologists using all clinical information, noncontrast CT, and MRI when available. Diagnostic metrics for stroke were calculated for interpretable vHIT studies and in an intention-to-diagnose analysis. RESULTS: Thirty patients underwent vHIT; 11 (36.7%) had acute ischemic stroke and 19 (63.3%) had nonstroke diagnoses. Interpretable vHIT results (central or peripheral pattern) were obtained in 23 of 30 patients (76.7%). Overall, vHIT patterns were central in 19 (63.3%), peripheral in 4 (13.3%), equivocal in 3 (10.0%), and poor quality in 4 (13.3%). Among strokes, 9 of 11 (81.8%) had a central pattern and none had a peripheral pattern. In the primary analysis of interpretable studies, vHIT sensitivity and negative predictive value for stroke were both 100%, while specificity was 28.6% and overall accuracy was 56.5%. In the intention-to-diagnose analysis including all patients, sensitivity and negative predictive value remained 100%, with specificity at 21.1% and accuracy at 50.0%. DISCUSSION: Our findings suggest that vHIT was feasible in most ED stroke-pathway patients with acute dizziness and demonstrated high sensitivity and negative predictive value when a clearly peripheral pattern was present, supporting its potential role as an adjunctive rule-out tool. However, specificity was low, and nearly one-quarter of studies were equivocal or poor quality, limiting its use as a standalone diagnostic test. With further validation and larger studies, vHIT may help support more targeted use of hospital admissions, diagnostic workups, and ED stroke resources in patients without evidence of central pathology.