Abstract
Purpose::
To determine the accuracy and repeatability of European ophthalmologists classifying stereoscopic optic disc photographs as healthy or glaucomatous and to compare their accuracy to that of scanning laser polarimetry with variable corneal compensation (SLP-VCC) and confocal scanning laser ophthalmoscopy (CSLO).
Methods::
Simultaneous stereoscopic optic disc photographs of one eye each of 40 healthy subjects, 48 glaucoma patients, and 6 patients with ocular hypertension were obtained. Measurements with SLP-VCC and CSLO were also performed. Sixteen photographs of healthy and glaucomatous eyes were duplicated for assessing intraobserver agreement, expressed as kappa. One-hundred and thirty-six general ophthalmologists from 9 European countries classified the optic disc photographs as normal or glaucomatous. These are interim results of an ongoing study in which approximately 300 opthalmologists from 11 European countries will participate. For SLP, the nerve fiber indicator (NFI) was analyzed. For CSLO, the Moorfields regression analysis (MRA) and the Bathija linear discriminant function (LDF) were used.
Results::
Ophthalmologists had a mean (range across countries) sensitivity, specificity, and overall accuracy of 73.5% (66.3-80.4), 88.9% (74.4-94.6), and 80.5% (73.9-85.9), respectively. The overall accuracy was statistically significantly different between countries (ANOVA, P<0.01). For individual ophthalmologists, the overall accuracy ranged between 58.0% and 94.3%. SLP-VCC had a sensitivity, specificity, and overall accuracy of 91.7%, 95.0%, and 93.2%. CSLO, expressed as Bathija LDF, had a sensitivity, specificity, and overall accuracy of 85.4%, 95.0%, and 89.8%, respectively. When expressed as MRA, CSLO had a sensitivity, specificity, and overall accuracy of 77.1%, 97.5%, and 86.4%, respectively. Intraobserver agreement was good on average (kappa, 0.74; SD, 0.20). Agreement was very good in 58 observers, good in 43, moderate in 24, fair in 5, and poor in 2.
Conclusions::
The accuracy of European ophthalmologists for classifying optic disc photographs in glaucoma varied considerably. We even found it to vary between countries. The intraobserver agreement was moderate to very good. Automated analysis of measurements with SLP-VCC and CSLO had, on average, a higher accuracy than classification of stereoscopic optic disc photographs by ophthalmologists. We think these imaging techniques thus may assist clinicians in diagnosing glaucoma.
Keywords: imaging methods (CT, FA, ICG, MRI, OCT, RTA, SLO, ultrasound) • optic disc