Purpose: To determine the reproducibility of the assessment for glaucomatous change in serial optic disc stereo-slides.
Design: Masked interobserver variability study.
Participants: Serial optic disc stereo-slides from 40 patients.
Methods: Three independent ophthalmologists evaluated for change a set of two serial 20 degrees optic disc color stereo-slides of 40 patients. This test set was not from European Glaucoma Prevention Study (EGPS) patients. Each observer performed two evaluations at least 30 days apart and was masked from the temporal sequence of the slides and his or her previous evaluation. Each patient was graded as changed or stable by two-out-of-three agreement. A kappa statistic was used to calculate the intra- and interobserver reproducibility as well as the assignment reproducibility (first consensus versus second consensus). The same procedure was followed to test the reproducibility when another experienced ophthalmologist was added to one of the three reading centers.
Main outcome measures: Reproducibility in evaluating glaucomatous optic disc change.
Results: The intraobserver reproducibility (95% confidence interval [CI]) in the evaluation of change ranged between 0.79 (0.45-1.14) and 1.00 (0.69-1.31). The interobserver reproducibility (95% CI) in the evaluation of change ranged between 0.45 (0.15-0.75) and 0.75 (0.44-1.06). The assignment reproducibility (first consensus versus second consensus in the evaluation of change) between the senior EGPS readers was 0.94 (0.63-1.25). The assignment reproducibility when another experienced ophthalmologist replaced one of the readers was 0.94 (0.63-1.25).
Conclusions: The assignment reproducibility of three expert readers looking for glaucomatous change in serial optic disc stereo-slides was excellent. It remained so when one of the three experts was replaced by another experienced reader.