Quality of Large-Scale Video Head Impulse Test Data: The Coronary Artery Risk Development in Young Adults (CARDIA) Study.

Abstract

Objective: To evaluate the quality and agreement of video head impulse test (vHIT) data collected by technicians without extensive vestibular training compared with audiologists.

Study design: Review of data collected at a single time point (year 35) as part of a larger observational cohort study known as the Coronary Artery Risk Development in Young Adults (CARDIA) study.

Setting: Birmingham, AL; Chicago, IL; Minneapolis, MN; and Oakland, CA.

Participants: 464 participants (928 ears) enrolled in the CARDIA Study, 5 novice technicians who completed vHIT, and 3 audiologists with extensive vHIT training.

Intervention: Each technician completed a 2-hour, virtual vHIT training. Technicians recorded average head velocity, gain, presence/absence of saccades, and a data quality rating. Three audiologists with extensive vestibular experience re-reviewed all data for comparison. Data categorized as noisy and/or having saccades were reanalyzed to determine whether data cleaning improved interpretation.

Main outcomes measures: vHIT gain, quality of vHIT results, presence of saccades.

Results: A total of 823 ears (88.7%) were classified as normal, 10 ears (1.1%) were abnormal, 13 ears (1.4%) had low gain without saccades, and 82 ears (9%) had high gain (>1.2). The most frequent artifact was high gain. Data cleaning increased the percentage of saccades and changed the interpretation of eight participants (1.7%).

Conclusions: Novice examiners can administer vHIT with online training and an initial practice period. Additional training on how to identify artifacts and clean data can improve population-based vHIT testing.