Multiple Regression Analysis of Diagnostic Predictors In Optic Nerve Disease

Abstract

Demyelination is assumed to be the cause of the majority of cases of isolated optic neuritis. Because of the importance of establishing the presence of optic nerve dysfunction in patients suspected of having multiple sclerosis several new indices of optic nerve conduction have been reported including the visual evoked potential, the edge-light pupil cycle time, and the Pulfrich test. These measures purport to detect optic nerve dysfunction but with varying degrees of success. This study of 93 patients with clearly documented previous optic neuritis was conducted to determine the statistical relationship between these three measures and other clinical diagnostic indices for detection of previous optic nerve disease and the utility of the diagnostic predictors taken individually, and in combination. The other indices used were the presence of detectable optic atrophy, color vision defect and the presence of a relative afferent pupillary defect. The variables were submitted to linear stepwise multiple regression analysis which indicated that the presence of optic atrophy, defective color vision and prolonged pupil cycle time when used in combination provided the most useful diagnostic prediction of previous optic neuritis in this group of patients. The addition of visual evoked potentials, the Pulfrich test or presence of a relative afferent pupillary defect did not significantly increase predictive reliability.