Interobserver and intraobserver variability in the detection of glaucomatous progression of the optic disc

To evaluate the potential value of obtaining follow-up stereoscopic photographs on glaucoma suspects in identifying progressive optic nerve damage.

Nineteen sets of stereoscopic optic disc photographs, reflecting one eye from each of 19 patients at two time points, were selected from the records of subjects enrolled in the Glaucoma Screening Study. By consensus, three experts judged 13 of these eyes to have progressive glaucomatous optic nerve damage. Four other ophthalmologists who were masked to the expert panel evaluation then assessed glaucomatous progression in the same eyes. They were asked to evaluate glaucomatous progression in three ways: first, by drawing the optic nerve head appearance from initial stereoscopic photographs and later comparing their own drawings to follow-up stereoscopic photographs; second, by comparing serial stereoscopic photographs directly; and third, by comparing drawings of the optic nerve head made by another examiner to the follow-up photographs.

Neither sensitivity nor specificity was consistently better for serial stereoscopic photographs than for drawings. Individual ophthalmologist agreement rates with the expert panel's determinations of progression were 23-62% when examiners compared their own drawings to follow-up photographs, 54-71% when examiners compared serial stereoscopic photographs, and 38-85% when comparing another ophthalmologist's drawings to follow-up photographs.

Baseline stereoscopic photographs of the optic nerve head did not substantially improve recognition of progressive glaucomatous optic nerve damage when compared with the use of baseline drawings of the optic nerve head made from photographs in subjects who developed visual loss in the interim.