Diagnostic ability of glaucoma probability score to discriminate between healthy individuals and glaucoma suspects

Abstract

Purpose To evaluate the diagnostic performance of the contour‐line independent glaucoma probability score (GPS) classification obtained with the Heidelberg retina tomograph (HRT) 3 to differentiate between healthy and suspected glaucoma subjects (SG). Methods 62 normal eyes and 48 SG (normal standard automated perimetry, intraocular pressure higher than 21 mmHg, and retinal nerve fiber layer defects measured with Stratus optical coherence tomography or scanning laser polarimetry GDx VCC) were included. The receiver operating characteristic (ROC) curves were plotted for the GPS numerical values and compared with HRT3 stereometric parameters. Results GPS nasal and the FSM discriminant function had the best ability to differentiate between healthy and suspected glaucoma eyes. The areas under the ROC curve were 0.830 and 0.818, respectively. Sensitivity was 47.9% for GPS nasal at a fixed specificity of 90%. Conclusion GPS had a good diagnostic ability to discriminate between normal and SG. There are structural changes at early stages of the disease that can be quantified by means of GPS.