Conversion of visual field size V testing thresholds into size III thresholds for uninterrupted trend analysis

Abstract

Purpose : Mounting evidence suggests that visual fields, at least in patients with advanced visual field defects, should be performed with stimulus size Goldmann V versus the traditional Goldmann III.
Any significant change in examination parameters in visual fields usually prevents direct comparison of the test results. Hence, the previously gained patient history which is relevant for assessing progression is lost and a new baseline, a new series of tests with the new examination parameters must be established before progression can be assessed again.
In this study we derive a transformation function that converts a size V test into the scale of a size III test.

Methods : We evaluate the pointwise pairs of size III and size V thresholds of visual field tests from advanced glaucoma patients from the study and publication of Sood et al (Standard automated perimetry using size III and size V stimuli in advanced stage glaucoma, BMJ Open. 2021) For each size V threshold we determined the median threshold (50^th percentile) of the corresponding size III threshold and the distribution of thresholds. Size III and size V testing was performed with a maximum stimulus intensity of 4000asb (0dB) on the Octopus 900 perimeter..

Results : Our dataset included 2387 pairs of size III and size V thresholds from 77 patients (OD only), after exclusion of test locations that deviated more than 1° in X or Y axes.
Some exemplary sensitivities for size V and corresponding size III (50^th percentile) were: 17dB (V) vs. 4dB (III), 22dB vs. 9dB, 30dB vs. 22dB, 36dB vs. 29dB. Hence, the biggest difference was 13dB at the lower part of the scale and the difference continuously decreased to ~7dB at 29dB size III thresholds. For values of 12dB and lower in size V testing, the 50^th percentile was 0dB respectively “not seen at 0dB” in size III testing.

Conclusions : The transformation function in the range of 14..30dB size III equivalent compares well with the differences known from the normal values and extends the available data down to 0dB. Using this function, it appears feasible to calculate size III thresholds from a size V test, as the dynamic range of size V testing is larger than the size III dynamic range. Thresholds in size V that corresponds to a threshold in size III below 0dB would be converted into a “not seen at 0dB” stimulus.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.