Accuracy and Reproducibility of Virtual Reality Perimetry in Children

Abstract

Purpose: To determine normal threshold values for the 24-2 program of the VisuALL-K (Olleyes, Inc), a portable static automated virtual reality perimeter that uses a video game–based testing modality on an updated version of the algorithm and to repeat the test to assess retest variability. Methods: This prospective cohort study included 19 participants (aged 8 to 17 years) with healthy afferent systems. Participants performed two 24-2 automated perimetry tests on the virtual reality perimeter device. Results: Mean age was 12.7 ± 2.5 years; 53% were boys and 47% were girls. Mean test location threshold was 30.5 ± 4.0 dB for test 1 and 31.3 ± 2.8 dB for test 2. Sensitivity decreased with eccentricity as expected (-0.17 dB per degree). Bland-Altman analysis showed a repeatability coefficient of ±5.1 dB for mean deviation, meaning that 95% of retest values were within ±5 dB of the threshold value for the first test. Higher thresholds for the first test regressed toward the mean values in the second test, indicating a learning effect. Intraclass correlation coefficient for mean deviation between test 1 and test 2 was 0.70 (0.55 to 0.80, P