Comparison of the diagnostic accuracy of MRA and GPS programs on HRT 3 device in glaucoma patients

COMPARISON OF THE DIAGNOSTIC ACCURACY OF MRA AND GPSPROGRAMS ON HRT 3 DEVICE IN GLAUCOMA PATIENTSPURPOSEThe aim of this study is to evaluate the diagnostic accuracy of Moorfields regressionanalysis (MRA) and the glaucoma probability score (GPS) and to measure theagreement between these two programs in classifying eyes as normal or abnormal.METHODSThe study included 150 glaucoma patients followed in our glaucoma unit and 120control subjects without glaucoma. All subjects underwent total ophthalmologicalexamination, standart achromatic perimetry and imaging of optic nerve head with theHeidelberg Retina Tomograph II (HRT), using HRT III software, Version 3.0. TheHRT parameters were compared between the two groups. The diagnostic accuraciesof the two classifications were measured when the borderline was taken as eithernormal (highest spesificity criteria) or abnormal (highest sensitivity criteria). Theagreement between them was calculated using the kappa (?) coefficient.RESULTSAll HRT parameters except height variation contour in our study showed statisticallydifferences between the control and glaucoma groups(p=0.000). The parameters withthe highest area under curves were GPS global score (0.817), vertical cup/disc ratio(0.816), cup/disc area (0.808) and cup area (0.783). According to the highestspecificity criteria, MRA had a sensitivity of %66 and spesificity of %89.1 and GPShad a sensitivity of %62.7 and spesificity of %81.6. According to the highestsensitivity criteria, MRA had a sensitivity of %85.3 and spesificity of %68.3 but GPS58had a sensitivity of %90 and spesificity of %60.8. A moderate agreement of %65.9(178eyes) with a ? coefficient of 0.47 was found between MRA and GPS.CONCLUSIONSIn conclusion GPS and MRA had a similar sensitivity, but the spesificity of MRA wasfound to be higher than GPS. A moderate agreement was found between theseanalyisi programs.