Assessment of Optic Disc Stereophotographs in the Glaucoma Imaging Longitudinal Study (GILS)

Abstract

Purpose:: To measure inter observer grading of changes in stereophotographs taken during the GILS.

Methods:: Subjects with OAG and visual field loss were studied every 6 months for a mean of 3 years, with automated field testing and 2 methods of fundus imaging. Non-simultaneous, color stereophotographs were taken annually. Three glaucoma specialists evaluated the first and last disc photographs taken during the study without knowing the temporal sequence. They categorized the disks as "changed" or "unchanged" and, if judged to be changed, designated the worse disc. They also noted the presence of disc hemorrhages.

Results:: 7 of 170 pairs of photographs were deemed unreadable by at least 2 of the readers. The median time between the pairs of stereophotographs was 26 months (range 5-50 months). The number of discs judged worsening varied from 13/163 (8%) to 18/163 (11%) to 44/163 (27%), with a Κ among the 3 readers of 0.20. Only 1 out of 163 (0.6%) discs was classified as progressing between the earlier and later photographs by all 3 readers and five of 163 (3.1%) discs by 2 of 3 readers. All 3 readers classified one other disc as progressing, and 2 of 3 readers classified 5 other discs as progressing, even though the photograph of the "progressed" disc was taken earlier. A disc hemorrhage was noted by at least one observer on 7/340 (2.1%) photos, by at least 2 observers on 6 photos, and by all 3 observers on 5 photos.

Conclusions:: The incidence of disc changes detectable on stereophotographs in this cohort was low. Half of the changes seen were in the "wrong" temporal sequence, suggesting that the actual rate of progression on disc photos over this short time interval was even lower. Agreement among expert observers was slight to fair.

Keywords: imaging/image analysis: clinical • optic disc