Evaluation of decision rules for frequency-doubling technology screening tests

Frequency-doubling technology (FDT) perimetry has shown promise as a screening test for glaucoma. This study investigates different possible decision rules for FDT screening by applying them to groups of normal and glaucoma subjects.

Within three centers, 218 subjects (aged 15-88 years; 78 with glaucoma, 140 without ocular disease) were each tested twice with the screening program of the FDT perimeter. The subjects consisted of 140 normal subjects with no evidence of glaucoma or other ocular disease likely to affect the visual field and 78 subjects with a diagnosis of glaucoma and no other ocular disease. Fifteen decision rules were applied to the data to compare their sensitivity and specificity.

Estimated specificities of the different decision rules ranged from 78% to 99%, although with this sample size, the confidence intervals for these estimates are quite large. Estimated sensitivities ranged from 40% to 72%. Suggested criteria for distinguishing normal subjects from those with glaucoma seem to be either a cluster of two or more adjacent locations abnormal at the p < 2% level with at least one location confirmed or a single location very abnormal (p < 1%) and confirmed.

Specificity was clearly improved by confirming an apparently abnormal test result by repeating the screening test outweighing the resultant small loss in sensitivity. These findings provide useful information for making an informed choice of decision rules for FDT screening results.