Abstract
Purpose :
The 24-2 visual field test has limited ability to detect change in the center of the visual field (VF). One approach to detecting change is to apply the change limits estimated from the central points in repeated 24-2 scans to 10-2 scans. In this study we use short-term reproducibility data to determine the specificity of using the 24-2 inner zone change limits from the HFA guided progression analysis (GPA) to detect change in the 10-2 VF.
Methods :
VFs were acquired from 74 eyes of 74 glaucoma subjects at 5 repeat visits within 4 months, using HFA™ II-i (ZEISS, Dublin, CA). At each visit, SITA Standard 10-2 and 24-2 VFs were acquired. Because the repeated visits occurred over a short period of time, any change observed in this data would by definition be a false positive.
In HFA, GPA change limits in the 24-2 field depend on location as well as baseline mean deviation (MD), and baseline pattern deviation (PD). For this study, we substituted 10-2 MD for the 24-2 MD. We plotted the correlation between the MDs to confirm that this was reasonable.
As in HFA GPA the first two scans in the series were averaged to create a baseline. The first and second follow-ups were compared to the baseline for each test point and the differences compared to the change limits used for the central portion of the 24-2 in the commercial version of GPA. False positive rates (FPR) with 95% confidence intervals (CI) were pooled for all 74 patients and all 68 test points in the 10-2 for both follow-ups. Specificity was defined as 1 – FPR. GPA was not performed on test locations where the MD or PD are too poor(censoring), which would be denoted as "out of range" or "X" on the report and in the analyses. FPR was calculated two ways: pooling all points, and pooling only points not marked “X”.
Results :
Mean age was 63.6 (35.7 to 79.6) years. Mean MD was -3.9 (-18.2 to 1.2) dB. Correlation of 24-2 MD with 10-2 MD showed an R^2 of 0.67, a slope of 0.89 and an offset of 0.19, so it is reasonable to substitute the MD from 10-2 to determine change limits. FPR was 0.47 (CI: 0.43 to 0.52) considering all 10064 pooled points and 0.51 (CI: 0.46 to 0.55) considering 9380 pooled valid points, consistent with specificity of ~95%.
Conclusions :
Specificity was ~95% for detecting progression in 10-2 fields using limits established in the central portion of the 24-2, indicating that a simple GPA for 10-2 can be created using existing data.
This is a 2021 ARVO Annual Meeting abstract.