Purchase this article with an account.
Yanfang Wang, David B. Henson; The Relative Performance Of Visual Field Tests Using Subsets Of The 24-2 Test Pattern At Detecting Early Glaucomatous Field Loss. Invest. Ophthalmol. Vis. Sci. 2011;52(14):5493.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
To establish the relationship between the number of 24-2 test locations (1 to 54) and visual field test performance.
A database of the visual field test results (SITA 24-2 algorithm, Humphrey Visual Field Analyzer, Carl Zeiss Meditec, USA) from 6935 eyes of 3621 patients with suspicious/diagnosed glaucoma was classified into 3 perimetric groups (normal: MD>-2.5dB, 5015 eyes; mild:-2.5~ -6dB, 1086 eyes; moderate/advanced: <6dB, 834 eyes). The positive predictive value (PPV) of each test location was calculated from the ‘normal’ and ‘mild’ groups. The performance (sensitivity and specificity) of optimised test patterns with between 1 and 54 test locations was established with a fail criteria of a single missed stimulus at p<0.01. The optimised test patterns started with a single test point at the maximum PPV location. Eyes missing this location were removed from the sample and the PPV of the residual sample calculated. This process was repeated until all eyes in the mild group had been detected, see figure left.
The PPV of each test location ranged from 0.76 to 0.32 (median 0.57). High PPV values occurred more often in the superior and in the nasal field. With the optimised test patterns all eyes within the ‘mild’ group were detected with 43 test points and 99% with just 39, see figure right. The 11 locations (1 in the BS) that did not contribute anything to the performance were fairly evenly distributed throughout the central field and often (8/11) fell within the arcuate areas (4 superior, 4 inferior). The specificity of the optimised patterns decreased with the number of test locations (99% for 1, 56% for 43).
Subsets of the 24-2 test pattern can reliably detect mild defects seen with the full 24-2 pattern, however, the specificity with the current fail criteria is often low. Subsets could be used to screen patients. Those failing would then undergo the full 24-2 test. This would significantly reduce test times in more than 50% of perimetrically normal patients.
This PDF is available to Subscribers Only