Purchase this article with an account.
N. G. Strouthidis, C. Zuniga-Cossio, D. F. Garway-Heath; The Repeatability of the Heidelberg Retina Tomograph Glaucoma Probability Score and Its Usefulness as an Index of Progression. Invest. Ophthalmol. Vis. Sci. 2007;48(13):3329. doi: https://doi.org/.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To evaluate the inter-visit, inter-observer repeatability of the Glaucoma Probability Score (GPS) and to compare its ability to monitor glaucomatous progression with established visual field (VF) progression strategies.
1) Heidelberg Retina Tomograph (HRT) ‘Classic’ inter-visit, inter-observer test-retest data from 74 subjects (Br. J. Ophthalmol. 2005;89;1427-1432) were analysed in HRT-3 Explorer software to generate the global GPS. Variance was related to the GPS, so repeatability coefficients (RCs) were calculated for 3 GPS bins (0-19%, 20-79% and 80-100%).2) 198 OHT and 21 control subjects were examined longitudinally (1993-2001) using regular HRT (‘Classic’) and VF testing. 5 or more HRT and VF tests were available for each subject. GPSs were generated in HRT-3 Explorer software. Linear regression of global GPS over time was performed for each subject, with significance p < 0.05. Criteria for estimates of specificity were: 1) significant deterioration of GPS over time in controls and 2) significant improvement of GPS over time in all subjects. The number of OHT subjects significantly deteriorating was compared to a VF pointwise linear regression strategy with similar specificity.
There was good GPS repeatability for the 0-19% and 80-100% bins, at RC = 0.4% and 4.3%, respectively. However, in the 20-79% bin, repeatability is poor with RC = 13.5%.Estimated specificity (95% confidence intervals) was 95.2% (76.1 - 99.9%) for GPS criterion 1 and 96.8% (93.2 - 98.6%) for GPS criterion 2. 25 OHT subjects were identified as progressing. VF progression was identified in 11 of these subjects. 26 OHT subjects were progressing by VF alone.
GPS scores between 20 and 79% are variable and should be interpreted with caution. Despite this, linear regression of GPS over time achieved a good specificity, with positive detection rates similar to other HRT progression algorithms at the same level of specificity (Invest Ophthalmol Vis Sci. 2006;47:2904-2910). Agreement (proportions progressing by GPS and VF) was also similar to previous reports.
This PDF is available to Subscribers Only