Purchase this article with an account.
S. Liu, G. Lai, L. Liu, C. Ye, C. Cheung, D. Lam, C. Pang, C. Leung; Comparison of Trend and Event Based Analyses for Detection of Visual Field Progression in Glaucoma. Invest. Ophthalmol. Vis. Sci. 2010;51(13):4013.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Objective: To evaluate the performance and agreement of event and trend based analyses of visual field (VF) progression in glaucoma.
147 eyes from 84 glaucoma patients were included. Each eye had a minimum of 3 reliable VF tests (Humphrey visual field analyzer, Carl Zeiss Meditec, Dublin CA) with the first and the last tests separated by at least 3 years. Event based VF progression was defined by the AGIS criteria, CIGTS criteria and Guided Progression Analysis (GPA). Trend based VF progression was analyzed by performing linear regression between visual field index (VFI) / MD and age (years). Statistically significant (p<0.05) negative trend was identified as progression. Agreement between visual field tests was evaluated with kappa statistics.
A total of 1476 visual field tests were analyzed. 75 eyes were found to have progression by at least one VF progression criteria whereas 3 eyes showed progression by all criteria. Trend analyses with VFI and MD detected more eyes as progressing (35 (23.8%) and 59 (40.1%) eyes, respectively) compared to event analyses by AGIS criteria, CIGTS criteria and GPA (8(5.4%), 14 (9.5%), and 21(14.3%) eyes, respectively). Agreement between trend and event analyses was poor (kappa ranged between 0.133 and 0.285). The best agreement was found between AGIS and CIGTS criteria (kappa = 0.596) and between VFI and MD trend analyses (kappa = 0.423). Progressing eyes detected by VFI trend analysis generally had more severe visual field defects than those detected by GPA. The mean MD (SD) of progressing eyes identified by AGIS criteria, CIGTS criteria, GPA, MD trend analysis, and VFI trend analysis were -6.5±4.2 dB, -5.1±3.7 dB, -4.7±3.3 dB, -5.7±6.4, -8.8±8.0 dB, respectively.
The agreement between trend and event VF progression analyses was poor. Trend analysis detected more eyes as progressing than event analysis. The selection of analysis approaches has significant impact on interpretation of VF progression.
This PDF is available to Subscribers Only