Abstract
Purpose::
To evaluate the agreement between three criteria for glaucomatous visual field (VF) progression using an event-based model and pointwise linear regression analysis (PLR).
Methods::
A retrospective analysis of serial VF data in glaucoma and ocular hypertensive patients followed annually with at least 5 years follow-up and 5 reliable 24-2 VF (≤ 33% fixation loss, false positive, false negative responses) was performed. Progression was identified using three methods: an event-based model using Glaucoma Progression Analysis (GPATM, Carl-Zeiss Meditec) defined as at least three progressing locations on the pattern deviation map on 3 consecutive VF, and PLR (PROGRESSORTM, Moorfields Eye Hospital) defined as a negative slope > 1.0 decibel per year with a significance p < 0.05 for at least one test location (PLR1) or at least three test locations (PLR2). One eye was randomly selected. The level of agreement for identification of progressing fields, and progressing test locations, was evaluated using a weighted kappa statistic.
Results::
Fifty-three eyes of 53 patients (42 glaucoma, 11 suspects) were enrolled (mean age 75.3±8.8 years). All eyes with glaucoma had associated VF abnormalities (average baseline mean deviation -3.9± 4.4 dB). The mean number of VF examinations was 6.4±2.5 and average length of follow-up was 5.4±0.8 years. Progression was identified in 8/53 (15%) eyes using GPA, 31/53 (58.5%) eyes using PLR1, and 12/53 (22.6%) eyes using PLR2. Significantly (p=0.016) greater agreement regarding the presence or absence of progression was observed between GPA and PLR2 (43/53, 81.1%, weighted kappa = 0.81) compared with GPA and PLR1 (30/53, 56.6%, weighted kappa = 0.63). The mean number of progressing test locations was similar (p = 0.11, McNemar test) using PLR (2±2) compared with GPA (1±2). Amongst the entire study population, 42 progressing locations were identified using GPA, 61 progressing locations were identified using PLR, and 26 progressing locations were identified using both GPA and PLR.
Conclusions::
Identification of glaucomatous VF progression varies markedly depending upon the criteria employed. In the present study, good agreement was observed between GPA and PLR2.