Abstract
Purpose: :
Determining the causes of test–retest variability of visual field measurements is very desirable. In cases of glaucomatous loss, variability is moderately correlated with amount of loss, for sensitivities down to about 10–15 dB. Below this, variability falls off. Visual field inhomogeneities have been shown to be associated with increased variability, but this has only been studied on a point–by–point basis. We studied the role of visual field structure by determining the gradient of visual field sensitivity over a substantial area of retina.
Methods: :
10–2 data were obtained from 12 eyes of 12 patients with glaucoma at the Glaucoma Institute of SUNY. Testing on each patient took place in 21 days or less. Sensitivity of a given eye was taken to be the average over repeats at each test location. Variability was taken to be the SD of sensitivity at each location. The gradient was determined at each test location in the central 6 x 6 array of test locations. Pairs of arrays from an eye were compared by determining the matrix correlation: C [A, B] is a number (–1 ≤ C ≤ 1) that describes the similarity of a pair of matrices or arrays, A and B. For each eye, we determined C [variability, sensitivity] and C [variability, |gradient| ]. Retrospective data from 24–2 or 30–2 tests of 8 of the 12 eyes were collected from the patients' records.
Results: :
For the 10–2 data, correlations with gradient were significant and often graphically striking; correlations with sensitivity were not. Matrix correlation values (mean ± SD) were: C [variability, sensitivity] = –0.02 ± 0.47 (not sig diff from 0), and C [variability, |gradient| ] = 0.44 ± 0.18 (sig diff from 0).
Scatterplots of variability vs. sensitivity for individual test locations showed the familiar pattern with a maximum around 15 dB sensitivity. For scatterplots of variability vs. |gradient|, variability increased approximately linearly with gradient, with a regression slope of 0.48 dB/(dB/deg) or 0.48 deg.
For the 24–2 and 30–2 data, C [variability, sensitivity] = 0.08 ± 0.56 (not sig diff from 0), and C [variability, |gradient| ] = 0.20 ± 0.12 (marginally different from 0).
Conclusions: :
For 10–2 data, correlations between variability and |gradient| were substantial. This suggests that a large part of the test–retest variability in the present 10–2 data could be accounted for by minor fixational eye movements, with an eye–position SD of 0.5 deg parallel to the gradient. Poorer correlations for 24–2 data may result from 24–2 grid spacing preventing accurate characterization of the gradient.
Keywords: visual fields • neuro-ophthalmology: diagnosis