Reliability parameters in computerized perimetry

Abstract

The distribution of results of perimetric reliability tests, false positive and false negative answers, fixation losses and short-term fluctuation, was studied in a material of 84 healthy normal subjects, randomly selected from the population, and in 45 patients with glaucoma. All subjects were tested with the 30–2 programme of the Humphrey Field Analyzer. We further investigated the influence of perimetric reliability on the measured visual field, as expressed by the visual field indices Mean Deviation and Pattern Standard Deviation. Sample multiple correlation coefficients were calculated for a number of combinations of reliability parameters and visual field indices.

False negative answer were more common and short-term fluctuation higher in glaucomatous than in normal subjects. False positive answer and fixation losses did not differ between groups. Among glaucoma patients false negative answers were significantly more common than false positive answers. No such difference was found in the group of normals. The results showed no age-dependence. A large part of inter-individual field variability, particularly among normals, could be explained by perimetric reliability. false negative answers and short-term fluctuation had the largest predictive value.

Normal inter-individual variation decreased considerably when stricter criteria for minimun results at reliability testing were applied. Naturally the number of fields meeting these criteria decreased at the same time. therefore, when establishing limits for normality, a compromise must be made resulting in reasonably narrow prediction limits without excluding a large percentage of normal and pathological fields from analysis.