Purchase this article with an account.
Gustavo V De Moraes, Jeffrey M Liebmann, Robert Ritch, Theodore Krupin, ; Reliability Indices and Rates of Visual Field Change in Glaucoma. Invest. Ophthalmol. Vis. Sci. 2014;55(13):5634.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
We investigated the relationship between reliability indices (false-negative, false-positive, and fixation loss) and rates of visual field (VF) change in a cohort of patients from the Low-pressure Glaucoma Treatment Study (LoGTS).
We analyzed the entire VF database of the LoGTS, a prospective, randomized, double-masked, multicenter clinical trial. Patients were tested with standard automated perimetry using 24-2 Full-Threshold Algorithm. All VF tests were included regardless of their reliability indices. Mixed-effects linear regression models were used to determine the relationship between mean VF sensitivity and reliability indices.
2,945 VF tests of 134 patients (268 eyes) were analyzed. The mean sensitivity significantly changed over time at -0.24 dB/yr (P<0.001). There was a significant relationship between mean sensitivity change and false-negative rates: as VFs deteriorated, false-negatives increased (P<0.001). The relationship between mean sensitivity and false-positive and fixation loss rates was in the opposite direction: these indices decreased as VFs deteriorated over time. When VF tests with false-negative rates >33% were excluded, the mean (SE) rate of sensitivity change was -0.20(0.06) dB/yr; when the same cut-off was applied for false-positive and fixation loss rates, the mean rate was -0.24(0.06) in both cases. Only the change observed when false-negative rates >33% were removed was significantly different from when all fields were included (P<0.001). Selecting only fields with reliability indices ≤33%, the mean rate of sensitivity change was -0.18(0.06) dB/yr, which was significantly different from when all fields were included (P<0.001). By setting different cut-off values for false-negative rates, we observed significantly slower rates of sensitivity change as the criteria became more stringent (P<0.05).
Test-retest variance is higher among eyes with more severe VFs and is associated with more false-negatives. In this exercise with patients followed longitudinally, we observed: 1) a learning effect which leads to decrease in false-positives and fixation losses over time; 2) analogous to what is known with regard to test-retest reliability, as VFs progress, false-negative rates increase; and 3) measured rates of sensitivity change can vary significantly depending on cut-off values applied for reliability indices.1] Bengtsson, Heijl. IOVS 2000;41:2201.
This PDF is available to Subscribers Only