Comparison of Heidelberg Retina Tomograph-3 glaucoma probability score and Moorfields regression analysis of optic nerve head in glaucoma patients and healthy individuals

Abstract

To compare the sensitivity and specificity of Moorfields regression analysis (MRA) and glaucoma probability score (GPS) between healthy and glaucomatous eyes with Heidelberg Retinal Tomograph 3 (HRT-3). The study included 120 eyes of 75 glaucoma patients and 138 eyes of 73 normal subjects, for a total of 258 eyes of 148 individuals. All measurements were performed with the HRT-3. Diagnostic test criteria (sensitivity, specificity, etc.) were used to evaluate how efficiently GPS and MRA algorithms in the HRT-3 discriminated between the glaucoma and control groups. The GPS showed 88 % sensitivity and 66 % specificity, whereas MRA had 71.5 % sensitivity and 82.5 % specificity. There was 71 % agreement between the final results of MRA and GPS in the glaucoma group. Excluding borderline patients from both analyses resulted in 91.6 % agreement. In the control group the level of agreement between MRA and GPS was 64 % including borderline patients and 84.1 % after excluding borderline patients. The accuracy rate is 92 % for MRA and 91 % for GPS in the glaucoma group excluding borderline patients. The difference was nor statistically different. In both cases, agreement was higher between MRA and GPS in the glaucoma group. We found that both sensitivity and specificity increased with disc size for MRA, while the sensitivity increased and specificity decreased with larger disc sizes for GPS. HRT is able to quantify and clearly reveal structural changes in the ONH and RNFL in glaucoma.