Quantification of Visual Field Variability in Glaucoma: Implications for Visual Field Prediction and Modeling

Area of Science:

• Ophthalmology
• Glaucoma Research
• Visual Field Analysis

Background:

• Visual field (VF) variability is a critical factor in glaucoma diagnosis and progression monitoring.
• Understanding VF variability's relationship with sensitivity and location is essential for accurate modeling.
• Existing regression models may not fully capture the complex nature of VF changes over time.

Purpose of the Study:

• To quantify visual field (VF) variability based on threshold sensitivity and location in glaucoma patients.
• To compare the data fit and prediction capabilities of weighted pointwise linear regression (PLR), unweighted PLR, and pointwise exponential regression (PER).

Main Methods:

• Retrospective analysis of two glaucoma patient datasets with extensive visual field testing and follow-up.
• Characterization of VF variability using standard deviation of residuals from PER.
• Comparison of unweighted PLR, weighted PLR, and PER for model fit and prediction accuracy.

Main Results:

• VF variability exhibited a zenith at 33 dB and a nadir at 11 dB of threshold sensitivity.
• Variability decreased with eccentricity, being significantly higher in the central 10° (P < 0.001).
• Pointwise exponential regression (PER) demonstrated superior ability in predicting future visual field sensitivity in the mid-to-long term.

Conclusions:

• Glaucomatous visual field variability is influenced by disease severity and retinal location.
• Weighted linear regression offers no significant advantage over unweighted models for VF data fit or prediction.
• PER's superior predictive performance is attributed to its ability to model the nonlinear decay characteristic of long-term glaucomatous perimetric progression.