Petersen J
Universitäts-Augenklinik Göttingen.
Ophthalmologe. 1993 Oct;90(5):533-8.
The experimentally determined psychometric function for Landolt rings (8 orientations, 183 eyes) is used to calculate the statistics of visual acuity results according to DIN 58,220 and their bias due to deviations from DIN 58,220. Test procedure fulfilling DIN 58,220, criterion 3/5: With a second test the first result is reproduced in 33.7%. In 47% the difference is 1 step, in 16.2% 2 steps, and in 2.8% 3 steps. Selecting the highest acuity value out of several tests yields a bias of +0.44 steps (2 tests), +0.66 steps (3), +0.81 steps (4 tests). Deviations from DIN 58,220, criterion 3/5: Linearly instead of log-scaled optotypes: above acuity 0.2 the effect is minimal (< 0.1 steps). For acuity 0.1 the bias is -0.9 steps. Old criterion 3/4: bias -0.55 steps; Repetition of a failed optotype: bias +1.11; Repetition of a failed acuity step: +0.61; Continued test despite fulfilled break condition: bias +0.3 steps; Weakened criterion of 2/5: bias +0.89 steps.
Visual acuity values can contain considerable errors. There is a need for more effective algorithms and an unbiased investigator. The solution is programmed automatic testing on a computer screen.
