Purchase this article with an account.
William H Swanson, Mitchell W Dul, Douglas G Horner, Victor E Malinovsky; Evaluating Sources of Test-retest Variablity in Glaucomatous Visual Field Defects. Invest. Ophthalmol. Vis. Sci. 2014;55(13):5643.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate effects of stimulus range and stimulus size on perimetric test-retest variability in patients with glaucoma.
: One eye each of 52 patients with glaucoma and 63 age-similar controls was tested with multiple forms of perimetry for repeated visits. Conventional automated perimetry (CAP) was conducted with 24-2 SITA-Standard, frequency-doubling perimetry (FDP) was conducted with the 24-2 pattern, and “contrast sensitivity perimetry 2” (CSP-2) was conducted on custom perimetric testing stations (Horner et al, Optom Vis Sci 2013 90:466-474), using Gabor sines scaled with stimulus location (0.14 - 0.5 cycle/degree). A subset of each group also made repeated visits for perimetry with two sets of two-dimensional Gaussian blobs: one set with the same scaling as CSP-2 and another set with a fixed standard deviation of 0.5°, which is twice the radius of the CAP stimulus and is 1/10th the width of the FDP stimulus. Temporal modulation was 5Hz counterphase flicker for 600 msec with CSP-2, 18 Hz flicker for 500 msec with FDP, and a luminance increment for 200 msec with CAP and the Gaussian blobs. Data were expressed as contrast sensitivity (CS) using Weber contrast, and as defect depth (DD) using log difference from mean normal. Stimulus range was set by choice of ceiling and floor: any value greater than the ceiling was set equal to the ceiling, and any value lower than the floor was set equal to the floor.
Figures show results of bootstrap analyses: symbols are means for 14 bins, error bars are 95% confidence limits, and dashed lines show limits imposed by stimulus range. Figure 1 shows test-retest variability versus contrast sensitivity for CAP with three different stimulus ranges: peak variability was halved as stimulus range decreased from 3.5 log units to 1.3 log units (t = 1.68, p < 0.05). Figure 2 shows test-retest variability versus defect depth for five different forms of perimetry. Peak variability was greatest for CAP and lowest for CSP-2 (t > 4, p < 0.0005). At all levels of defect, variability was similar for FDP, CSP-2, and perimetry with Gaussian blobs (t < 0.5, p > 0.50).
Test-retest variability in glaucomatous defects was reduced by restricting the stimulus range, and by modest increases in stimulus size.
This PDF is available to Subscribers Only