Analysis of progressive change in automated visual fields in glaucoma

Purpose. To detect and estimate the rate of progression of visual field loss in subjects with glaucoma who undergo long-term automated perimetric visual field testing. Methods. Automated visual field data were obtained for subjects with glaucomatous visual field loss and a minimum of seven threshold field tests over at least 4.5 years. Univariate linear regression was performed with respect to mean deviation (MD), corrected pattern standard deviation (CPSD), mean thresholds of clusters corresponding to the Glaucoma Hemifield Test (GHT), and thresholds of 52 individual test locations. Subjects were classified as progressive or stable (unchanged or improved) based on the slope and statistical significance of these parameters. Adjusted P values were used to maintain the overall type 1 error at 5%. Results. One hundred ninety-one subjects with a mean follow-up period of 7.1 years (range, 4.5 to 10.5 years) and a mean number of visual field tests of 9.5 (range, 7 to 16) were included. Twenty-four subjects (12.6%) showed progression in MD (mean slope [95% confidence interval], -1.26 [-1.50, -1.01] dB/year), and 27 (14.1%) showed progression in CPSD (mean slope [95% confidence interval], 0.71 [0.58, 0.84] dB/year). Thirty-five subjects (18.3%) had ≥1 progressive GHT cluster. The mean slope in progressive clusters ranged from -1.51 [- 1.82, -1.20] to -2.84 [-3.39, -2.29] dB/year. Thirty-six subjects (18.8%) had ≥ 1 progressive individual test locations. Fifty-two subjects (27.2%) were classified as progressive based on progression of CPSD, ≥ 1 cluster and/or ≥ 1 point. Conclusions. Fewer than 1 in 3 subjects progressed by any one of the criteria for progression over an average of 7.1 years. Rates of progression that could be statistically confirmed were in the range of approximately 1 to 5 dB/year, depending on the number of fields, the variability over time, and the parameter assessed (global indices, GHT clusters, or individual points). No correlation between initial visual field status and the rate of progression was found. A minimum of approximately 5 years of follow-up with annual perimetry would be required to detect significant changes in the visual field by linear regression.