Abstract
Abstract: :
Purpose: To compare the performance characteristics of 7 criteria for analyzing glaucomatous visual field progression, using a combination of real patient data and computer simulation. Methods: Initial and final Humphrey 30-2 visual fields (separated by 7 years) of 76 patients with open-angle glaucoma were entered into a computer simulation program, which generated 14 interim semi-annual fields under conditions of high, moderate and no variability. Progression was analyzed using the methods of the Advanced Glaucoma Intervention Study (AGIS), the Collaborative Initial Glaucoma Treatment Study (CIGTS), 3 criteria (varying number of significant "black triangles") based on the Glaucoma Change Probability (GCP) analysis, and 2 criteria (number of significant test locations) based on pointwise linear regression analysis (PLRA). Specificities were calculated after the simulation was performed with the same visual field of each patient as both the initial and final field (no progression) under conditions of moderate and high variability. Results: Under the condition of no variability, progression rates were 18% for AGIS, 36% for CIGTS, 47% to 62% for the 3 GCP criteria, and 67% and 72% for the 2 PLRA criteria. With increasing variability, progression rates decreased for all criteria, except for the GCP criteria for which there was an increase. The time to detect confirmed progression was longest for the PLRA criteria and shortest for the CIGTS and GCP criteria. Under the moderate variability condition, all methods had high specificity (≷91%), while for high variability, all but two of the GCP criteria attained specificity ≷98%. Conclusion: AGIS and CIGTS criteria had high specificity, but classified fewer cases of progression than the other criteria. The GCP criteria had the shortest follow-up times to confirmed progression, but were not as specific. Criteria based on PLRA were specific but follow-up times to confirmed progression were the longest.
Keywords: 624 visual fields • 511 perimetry • 359 clinical research methodology