Retest Variability and the Gradient of the Visual Field: Blind Spot vs. Scotoma

Abstract

Purpose: : We have shown previously that perimetric test-retest variability is correlated with the gradient of visual field sensitivity in glaucomatous defects (Vis. Res. 2007) and also near the normal blind spot (ARVO 2007). We suggested that small fixational eye movements may underlie the relationship. The 2 deg test-location spacing used in that work may be too coarse to characterize sharp edges adequately; therefore, we have studied scotoma and blind spot edges with 1 deg test spacing.

Methods: : Six normal subjects ("NS") had one eye tested on a Humphrey HFA-2 with custom test patterns consisting of 4 rows (a row = 5 locations, 1 deg apart) along the 4 principal meridians of the subject’s blind spot. Six nasal-field locations were tested as controls. Four patients with glaucoma ("GP"), with steep-edged scotomas, had one eye tested similarly, but GP test patterns consisted of 3-4 rows crossing scotoma edges. 3 or 4 relatively healthy locations were tested as controls. Test sessions. NS: 3 sessions, each location once per session. GP: 2 sessions, each location twice per session. Added repeats performed automatically by HFA software were included. Mean number of tests per location: 5.2 (NS) and 6.1 (GP). HFA gaze-tracking records were quantified by photoediting and measuring with ImageJ.

Results: : Gradients were steep: for NS and GP combined, the average max gradient along test rows = 18.2 dB/deg (max = 29.5, min = 7.8). For both groups, mean |fixation offset| ~ 1 ± 0.5 deg. Plots of pointwise variability vs. sensitivity showed similar "inverted-U" relationships for NS and GP; plots of pointwise variability vs. gradient showed similar noisy linear relationships for NS and GP. The value of r² for variability vs. gradient was 0.14 (NS) and 0.20 (GP). For individual eyes, r² varied from 0.03 to 0.37 (NS) and from 0.01 to 0.69 (GP). For GP eyes, correlations were stronger for eyes with less stable fixation. For test locations with sensitivities 10-20 dB, GP and NS variability did not differ significantly (p ~ 0.11). At these locations, NS variability = 4.3 dB (range 0.9-10.0); GP variability = 6.0 dB (range 1.2-13.2).

Conclusions: : Similarities between NS and GP data are consistent with a similar mechanism for variability. We propose that normal small eye movements are an important factor in test-retest variability when edges are steep, and that they can be very steep.

Keywords: perimetry • neuro-ophthalmology: diagnosis • visual fields