PURPOSE: To determine the accuracy of glaucoma detection by frequency-doubl
ing perimetry.
METHODS: Stereoview optic nerve photographs, visual field examination, intr
aocular pressure measurements, medical and ocular history, and a screening
and full threshold frequency-doubling perimetry examination were performed
in a prospective study of consecutive subjects, Inclusion criteria included
age of 45 years or older, absence of ocular disease other than glaucoma, c
ataract, or mild drusen, and Snellen visual acuity of 20/60 or better. A to
tal of 125 eyes in 102 glaucoma subjects and 95 eyes of 95 normal subjects
were included. Each eye was classified as "normal," "glaucoma," Or "uncerta
in" by each of three ophthalmologists on the basis of all available clinica
l information with the exception of frequency-doubling perimetry results. T
hose in the glaucoma group were subclassified as having early (n = 51), mod
erate (n = 42), or severe (n = 32) glaucoma on the basis of automated Humph
rey visual field criteria. In the glaucoma group, two eyes from a subject w
ere allowed to be included (23 of 102 subjects) if they differed in level o
f damage because they were never analyzed within the same statistical analy
sis.
RESULTS: Several diagnostic algorithms were evaluated. Algorithms based on
the most depressed single point, pair of adjacent points, and cluster of th
ree points performed nearly identically. For the screening test, if any abn
ormality was identified, specificity was 95%, whereas sensitivity was 39%,
86%, and 100% for early, moderate, and severe glaucoma, respectively. For t
he full threshold test, with at least one point depressed to the P < 0.5% l
evel, specificity measured 91%, whereas sensitivity was 35%, 88%, and 100%
for early, moderate, and severe glaucoma, respectively, The two global indi
ces, mean deviation and pattern standard deviation, were also evaluated and
were generally less accurate.
CONCLUSION: Frequency-doubling perimetry, which is rapid and easily adminis
tered, is effective at detecting moderate and severe disease and appears we
ll suited for glaucoma screening. (C) 2000 by Elsevier Science Inc. All rig
hts reserved.