Refinement of pointwise linear regression criteria for determining glaucoma progression

Abstract

Purpose: A variety of pointwise linear regression (PLR) criteria have been proposed for determining glaucomatous visual field progression. However, alternative PLR criteria have only been assessed on a limited basis. The purpose of this study was to evaluate a range of PLR slope and significance criteria to define a clinically useful progression decision rule for longitudinal visual field examinations.

Methods: Visual field data for each of 140 eyes (one per participant among 96 cases and 44 controls) were evaluated using the Humphrey Field Analyzer II program 24-2 Swedish interactive thresholding algorithm standard test strategy and Goldmann size III stimuli. The pointwise linear regression A2 (PLRA2) method was used to analyze the data, which included nine visual field examinations performed every 6 months for 4 years. Data from the Ocular Hypertension Treatment Study (OHTS) were used to validate the decision rule.

Results: Several slope criteria produced specificities of 0.90 or higher, particularly slope criteria of less than -1.2 dB/y. The use of the slope criterion less than -1.2 dB/y at a significance level of P < 0.04 for classification resulted in a hit rate of 0.38, more than a 2-fold increase compared with a commonly used standard slope criterion of less than -1.0 dB/y at a significance level of P < 0.01. A similar increase in the hit rate was shown for a slope of less than -1.2 dB/y and P < 0.04 compared with the standard criterion in the independent OHTS validation data.

Conclusions: When systematically evaluating criteria for detecting glaucoma progression, PLR criteria can be refined by requiring a stricter slope criterion such as less than -1.2 dB/y and relaxing the significance criterion to P < 0.04. Increasing the hit rate of PLR will be useful for early detection and treatment of glaucoma.

Keywords: glaucoma progression; perimetry; pointwise linear regression; trend analysis; visual field.

Figures

Figure 1.

Visual field location flagged as progressing by PLR analysis. The first progression found by PLR is at time point seven. The PLRA2 then fits a regression model incorporating visual field examinations eight and then nine to confirm the determination of progression at time point seven.

Figure 2.

Locations satisfying the criterion of slope less than −1.2 dB/y and P < 0.04 but not satisfying the criterion of slope less than −1.0 dB/y and P < 0.01. The PLR with the refined criterion would classify these individuals as progressing, whereas PLR with the standard criterion would not classify progression. (A) and (B) are two locations from the same patient with glaucoma, and (C) and (D) are two locations from the same control.