Purpose-To compare the diagnostic accuracy of the Moorfields Regression Analysis (MRA), parameters, and Glaucoma Probability Score (GPS) from Heidelberg Retinal Tomograph HRT-3 with MRA and parameters from HRT-II in discriminating glaucomatous and healthy eyes in subjects of African (AA) and European ancestry (EA).
Design-case-control, institutional setting.Methods-78 glaucoma patients (AA=44, EA=34) and 89 age-matched controls (AA=46, EA=33), defined by visual fields and self-reported race were included. Imaging was obtained with HRT-II and data were exported to a computer with the HRT-3 software using the same contour line. Area under Receiver-operating Characteristic [ROC] curves [AUC], sensitivity and specificity were evaluated for whole group, AA and EA separately. Mean disc area was compared between correctly and incorrectly diagnosed eyes by each technique.Results-Disc, cup and rim areas from HRT3 were lower than HRT-II (P<0.0001). AUC (sensitivity at 95%-specificity) was 0.85 (54%) for vertical cup-to-disc ratio (VCDR) HRT3, 0.84 (45%) for VCDR and 0.81 (44%) for GPS score at temporal sector. MRA-HRT3 showed greater sensitivity but lower specificity than HRT-II for whole group, AA and EA. GPS classification had lowest specificity. Glaucomatous eyes incorrectly classified by GPS had smaller