AUTHOR=Machner Björn , Erber Kira , Choi Jin Hee , Sprenger Andreas , Helmchen Christoph , Trillenberg Peter 

TITLE=A Simple Gain-Based Evaluation of the Video Head Impulse Test Reliably Detects Normal Vestibulo-Ocular Reflex Indicative of Stroke in Patients With Acute Vestibular Syndrome

JOURNAL=Frontiers in Neurology

VOLUME=Volume 12 - 2021

YEAR=2021

URL=https://www.frontiersin.org/journals/neurology/articles/10.3389/fneur.2021.741859

DOI=10.3389/fneur.2021.741859

ISSN=1664-2295

ABSTRACT=Objective: The head impulse test (HIT) assesses the vestibulo-ocular reflex (VOR) and is used to differentiate vestibular neuritis (abnormal VOR) from stroke (normal VOR) in patients presenting with an acute vestibular syndrome (AVS). The video-oculography-based HIT (vHIT) quantifies VOR function and provides information imperceptible for the clinician during clinical bedside HIT. However, the vHIT - like an electrocardiogram - requires experienced interpretation, which is especially difficult in the emergency setting. This calls for a simple, reliable and rater-independent way of analysis.
Methods: We retrospectively collected 171 vHITs performed in patients presenting with AVS to our emergency department. Three neuro-otological experts comprehensively assessed the vHITs including interpretability (artefacts), VOR gain (eye/head velocity ratio), velocity profile (abrupt decline) and corrective saccades (overt/covert). Their consensus rating (abnormal/peripheral vs. normal/central) was compared to a simple algorithm that automatically classified the vHITs based on a single VOR gain cutoff (0.7). 
Results: Inter-rater agreement between experts was high (Fleiss’ kappa=0.74). Five (2.9 %) vHITs were ‘uninterpretable’ according to experts’ consensus, 80 (46.8 %) were rated ‘normal’ and 86 (50.3 %) ‘abnormal’. The algorithm had substantial agreement with the experts’ consensus (Cohen’s kappa=0.75). Importantly, it correctly classified all of the normal/central vHITs denoted by the experts (100% specificity) and at the same time it had sufficient sensitivity (75.6%) in detecting abnormal/peripheral vHITs .
Conclusion: A simple, automated, gain-based evaluation of the vHIT reliably detects normal/central VOR and may be a feasible and effective tool to screen AVS patients for potentially underlying stroke in the emergency setting.