Faisal A. Almobarak
King Saud University, Saudi Arabia
Title: Glaucomatous Optic Nerve Head Assessment in residency program using Discus software
Biography
Biography: Faisal A. Almobarak
Abstract
Objective & Purpose: To validate DISCUS software and establish reference data for the performance and reproducibility of residents evaluating optic nerve head damage.
Materials & Methods: 54 residents tested themselves 3 times using the freely available online discus software displaying 100-disc photographs: 80 non-stereo-photographs of VF negative optic nerves and 20 of VF positive glaucomatous discs. 26 photographs were repeated (2 glaucomatous and 20 healthy) to test the reproducibility of answers. At the end of the test, online submission and analysis of results is done where, each resident will repeat the test other day and will finally submit the file to the residency office where new identification numbers will be created. Online feedback was given to each resident. Interclass correlation, Bland-Altman test, average of diagnostic performance and ROC curves were calculated.
Results: The mean response latency was 4.91 seconds with no difference in the response between all residents (p=0.28, Wilcoxon test). The interclass correlation showed high degree of reliability of the 3 tests (0.76). The Bland-Altman test showed good agreement and reproducibility of 3 tests among all residents with more convergence in senior residents and only two repeated tests were more than one standard deviation but within two standard deviations. Senior residents had better performance that junior residents where the AUROC was 0.68, 0.68, 0.69 and 0.71 for the first, second, third and fourth year residents respectively (p=0.04, Kruskal-Wallis test). There was no correlation between the performance and the response latency (ρ=- 0.03, p=0.89). There was a moderate correlation with experts, 0.59 (±0.1), for all level, but fourth years residents had the highest correlation among others [r=0.64 (±0.08)] which was significantly different when compared to junior residents (p=0.02).
Conclusion: Senior residents had better performance and less variability on repeated tests compared to junior residents and had the highest correlation with experts. Discus allows a rapid and efficient assessment of resident’s performance of different levels in disc damage, which can help improving training defects whenever present.