Abstract
Purpose: :
Standard automated perimetry is limited by an exponential rise in variability with decreasing sensitivity and a limited effective dynamic range (EDR). We therefore studied the effect of size on repeatability and EDR using Goldmann stimulus sizes III, V, and VI in glaucoma patients, using a customized Humphrey Field Analyzer modified to produce a size VI stimulus.
Methods: :
We tested 8 glaucoma patients with Humphrey program 24-2 full threshold testing using stimulus sizes III, V and VI each 5 times over a 5 week period. Mean deviation for size III was -13.9 ± 8.2. Point-wise limits of test-retest variability were then established from the empirical 5th and 95th percentiles of the distribution of retest values, stratified by the value of at the first test (Fig). The log differences between test and retest values were linearly regressed onto the averages of the two tests to determine the relationship between variability and sensitivity.
Results: :
The average sensitivities were size III: 14.7 ± 6.6 dB; size V sensitivity: 21.5 ± 8.0 dB and size VI: 25.2 ± 6.9 dB. Using floor effect as an index of effective dynamic range, the numbers of 0 dB trials were: size III - 578, size V - 144 and size VI - 13; this indicates a greater dynamic range for the larger stimuli. There was increasing variability associated with lower visual field sensitivity but the rise in variability was less with the larger stimulus sizes, with size VI having the least rise (Fig). After eliminating the values subject to a floor effect, we found the following correlations: With size III, sensitivity explained 16% of the test-retest variability (r2), while corresponding figures for size V and size VI were 12%, and 10%, respectively.
Conclusions: :
Large sized conventional perimetric stimuli have a diminished increase in variability with decreasing sensitivity and a greater effective dynamic range. These stimuli show promise for use in moderate to severe glaucoma.
Keywords: perimetry • visual fields • clinical research methodology