Examiner Agreement of Irido-Corneal Angle Assessment Using an Automated Gonioscopic Device.

Abstract

Purpose : In spite of many technological advances in the area of imaging, gonioscopy is still the clinical gold standard for evaluating the anterior chamber as well as identification of both normal and abnormal irido-corneal angle characteristics The purpose of this study was to evaluate the interobserver agreement for gonioscopy assessment using an automated device taking into account the grade of experience of examiners.

Methods : Twenty five gonioscopic cases were selected by one of the investigators (not participating as a grader) trying to have a diversity of clinical pictures. All images derived from the NGS-1 automated gonioscope (NIDEK Co., Gamagori, Japan). The automated gonio-photos were graded by four independent examiners (2 expert graders - and 2 non-experts) having at least 140 minutes each grader for such purpose. Agreement was determined in five different categories about the gonioscopic findings.

Results : All graders had a complete agreement regarding the quality of visibility of the angle traits. Regarding the number of recognizable angle structures, there was a good positive correlation between the two expert examiners (r=0.62, P=0.024) and fair between both, the two non-expert examiners (r=0.43, P=0.068) and one of the expert graders and the two non-experts ones (r=0.41, P=0.17; r=0.46, P=0.074). In general, one of the expert examiner outperformed the remaining examiner regarding some peculiar findings (neovascularization, anterior synechias, embritoxon). A poor to fair level of agreement (r < 0.5) on iris confirguration was present among all the examiners. A good level of agreement (r > 0.5) was observed to determine if an angle was open or closed.

Conclusions : An automated documentation of the irido-corneal angle characteristics could be very helpful due to the quality of images that can be acquired but the interpretation derived from observed findings requires experience.

This is a 2020 ARVO Annual Meeting abstract.