Abstract
Purpose :
Precision of visual field (VF) progression estimates is important as more precise estimates allow for earlier intervention. Early identification of patients whose VF testing has poor value in guiding treatment may allow for earlier discontinuation of VF testing or earlier intervention to improve VF performance. This retrospective clinical study aims to characterize factors that may influence precision of VF progression rate estimates.
Methods :
VFs were collected from patients at a tertiary care academic glaucoma practice between 2002 and 2011. VFs were performed using the Swedish Interactive Threshold Algorithm 24-2 pattern. Participants were excluded if baseline mean deviation (MD) was < -20dB, first 2 VFs (baseline VFs) were > 14 months apart, or < 5 VFs were available. Precision of the VF progression rate was defined as the width of the 95% confidence interval (CI) of the MD progression rate for the first 5 fields. The association between the 95% CI of MD progression rates and age, baseline severity, and the difference in MD between the first 2 VFs was calculated using linear mixed effects models, which account for intra-person and intra-eye effects. Subjects were stratified by baseline disease severity -- early: MD ≥ -6 dB; moderate: MD < -6 dB & ≥ -12 dB; advanced: MD < -12 dB & ≥ -20 dB.
Results :
A total of 12,580 VFs spanning a mean of 3.5 (1.30) years from 2,516 eyes were included. Mean (standard deviation) age was 63.5 (14.0) years and baseline MD was -4.19 (4.8) dB. The mean MD slope was -0.21 (1.48) dB/year, and the mean width of the 95% CI of the MD progression rate was 4.02 (6.72) dB/year. Greater difference in MD between the first 2 VFs was significantly associated with decreased certainty of MD progression (P<0.001). For every 1dB difference between the 2 baseline VFs, the 95% CI of the MD trendline was 1.17 times wider (95% CI 1.16, 1.19; P<0.001). Moderate and advanced disease decreased certainty of VF progression as compared to early disease (β: 1.36, β: 1.34; P<0.001 for both, early disease as reference group). No association was found with age (P=0.11).
Conclusions :
Estimation of VF progression is less precise when the first 2 VFs are dissimilar and in moderate or advanced glaucoma. In these situations, consider coaching of patients to improve VF results, or prioritize other diagnostic modalities, such as optical coherence tomography.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.