Abstract
Purpose :
We tested the hypothesis that a joint analysis of pointwise progression with visual sensitivities and retinal nerve fiber layer thickness (RNFLT) improves the ability to detect glaucoma progression over each measurement individually.
Methods :
Data consisted of series of 9 visits with visual fields obtained with the 24−2 static automated perimetry HFA II program (Carl Zeiss Meditec, Dublin, CA, USA) and OCT circular scans obtained with the Spectralis OCT (Heidelberg Engineering, Heidelberg, Germany) for 250 eyes of 150 patients with glaucoma selected from the Diagnostic Innovations in Glaucoma Study and the African Descent and Glaucoma Evaluation Study datasets. The mean follow up period was 4.8 years, with the shorter follow up period being 2.9 years and the longest 8.1 years. At baseline, the average mean deviation and RNFLT (and standard deviation) were −2.5 (5.2) dB and 83 (16) microns. The RNFLT for the 52 visual field locations not adjacent to the blind spot were obtained after mapping structure and function with the Jansonius map. Permutation of pointwise linear regression (PoPLR) was applied to series of pointwise visual sensitivities and to series of pointwise RNFLT. Eyes were flagged as progressing according to 4 different criteria: significant progression on pointwise sensitivities alone, on pointwise RNFLT alone, with both pointwise sensitivities and RNFLT (AND criterion), and with either pointwise sensitivities or RNFLT (OR criterion).
Results :
Figure 1 shows the positive rates for series of 9 visits. Positive rates for pointwise RNFLT were similar to those for the OR criterion, and significantly greater than for pointwise sensitivities and the AND criterion. For a significance level of 5%, the positive rates for pointwise RNFLT were about 2 times greater than for pointwise sensitivities and about 1.5 times greater than for the AND criterion.
Conclusions :
Pointwise analysis of RNFLT progression outperformed pointwise visual field progression. Joint analysis did not significantly improve the detection of progression.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.