The sensitivity and specificity of direct ophthalmoscopic optic disc assessment in screening for glaucoma: a multivariate analysis.

BACKGROUND: Previous studies have shown that the cup:disc (C/D) ratio has limited sensitivity and specificity. Improved sensitivity and specificity has been found with combinations of disc parameters using stereophotography or disc imaging, but the extent to which these studies reflect what can be achieved by direct assessment is unclear. The purpose of this paper is to report the sensitivities and specificities of a range of disc parameters, both in isolation and in combination, as graded during clinical examination with direct ophthalmoscopy.
METHODS: Sixty-seven patients with newly diagnosed primary open angle glaucoma and 145 non-glaucomatous subjects were examined by direct ophthalmoscopy. The following features of the optic disc were graded: vertical and horizontal C/D ratios (uncorrected for disc size), narrowest rim width, presence/absence of disc haemorrhage, focal pallor of the neuroretinal rim, parapapillary atrophy, steepness of the cup-edge and presence/absence of baring of the lamina cribrosa. Subjects were validated as glaucomatous/non-glaucomatous on the basis of visual fields, independently of the results of optic disc gradings. Multiple logistic regression modelling was used to evaluate the effectiveness of combinations of disc parameters.
RESULTS: Sensitivities and specificities are limited for all features when considered in isolation. The 'cupping' parameters have the highest sensitivities and specificities but no single cut-off criterion for these parameters has both high sensitivity and specificity. Multiple logistic regression modelling indicated that two disc parameters in combination, narrowest rim width and parapapillary atrophy, were independently associated with glaucoma and achieved best prediction. Combining information from these parameters achieved a sensitivity of 81% (95% CI = 69-89%) and a specificity of 90% (95% CI = 84-95%), a level of discrimination that is not significantly better than narrowest rim width alone in terms of area under the receiver operator characteristic (ROC) curve.
CONCLUSION: Disc assessment during clinical examination with direct ophthalmoscopy can achieve comparable sensitivities and specificities with those previously reported for stereophotographic assessment or visual fields screening. However, since the diagnostic accuracy of disc assessment in isolation is inadequate for screening, a combined test strategy is necessary.