Scanning laser polarimetry in a selected group of patients with glaucoma and normal controls

PURPOSE: To evaluate the ability of scanning laser polarimetry to discriminate between subjects with glaucoma with specific patterns of visual field defect and normal controls. METHODS: This cross-sectional, prospective study in a glaucoma practice, focused on subjects with glaucoma with predefined types of visual field defect, including advanced (group A, n = 14), localized (group L, n = 46), or mixed (diffuse and localized) defects (group M, n = 22) and normal controls (n = 32). Scanning laser polarimetry was performed in one study eye per subject. Two methods of analysis were used: a subjective analysis, in which examination printouts with the image of the optic disk manually blocked were classified by two observers masked to the diagnosis, and a logistic regression analysis of the retardation parameters included in the printouts. RESULTS: The observers correctly identified 97% of the controls and 68% of subjects with glaucoma (overall correct classification of 77%), with 93%, 70%, and 47% of patients from groups A, L, and M, respectively, being correctly identified. The best discrimination obtained with the logistic regression correctly identified 69% of controls and 94% of glaucoma subjects (overall correct classification of 87%). The performance was only slightly better for cases from group A compared with L and M. CONCLUSIONS: Subjective assessment of the scanning laser polarimetry standard printout of single eyes might not be sensitive enough to detect cases of glaucoma with localized or milder mixed types of visual field defect. The discriminating ability of scanning laser polarimetry improves slightly when logistic regression analysis is employed.