Abstract
Purpose: :
To evaluate the diagnostic accuracy of Moorfields Regression Analysis (MRA) and Glaucoma Probability Score (GPS) in primary open-angle glaucoma (POAG) and measure the level of agreement in classifying eyes as normal or abnormal.
Methods: :
We prospectively selected 184 healthy subjects (123 women, 61 men) and 158 subjects with POAG (98 women, 60 men) in the study. All subjects underwent an ophthalmological examination, visual field analysis with Humphrey Visual Field Analyzer (30-2 SITA Standard) and imaging of the optic nerve head with Heidelberg Retina Tomograph II, version 3.0. The diagnostic accuracies of the 2 classifications were measured when the borderline was taken as normal (highest specificity criteria) or abnormal (highest sensitivity criteria) and the agreement between them was defined with kappa (Κ) value.
Results: :
MRA classified healthy eyes as normal in 75%, borderline in 20.1% and abnormal in 4.5% of the subjects, whereas GPS classified healthy eyes as normal in 57.6%, borderline in 30.4% and abnormal in 12%. MRA classified glaucomatous eyes as normal in 19%, borderline in %13.3 and abnormal in 67.7% of the eyes, whereas GPS classified glaucomatous eyes as normal in 10.8%, borderline in 18.4% and abnormal in 70.9%. The Κ value was measured as 0.51 and agreement ratio was 68%. According to highest specificity criteria; MRA had a sensitivity of 67.6% and a specificity of 95.1%, whereas GPS had a sensitivity of 70.9% and a specificity of 88%. According to highest sensitivity criteria; MRA had a sensitivity of 75% and a specificity of 81%, whereas GPS had a sensitivity of 89.2% and a specificity of 57.6%.
Conclusions: :
The diagnostic performances of MRA and GPS were found to be similar and the level of agreement in classification of the eyes is moderate.
Keywords: imaging/image analysis: clinical • intraocular pressure • optic nerve