Purpose: To report the ability of frequency doubling perimetry to detect "neuro-ophthalmic" field defects, characterize them as hemianopic or quadrantanopic, and differentiate glaucomatous from "other" neuro-ophthalmic field defects.
Methods: Sixty eyes of 30 normal subjects, 50 eyes of 29 patients with glaucomatous defects, and 138 eyes of 103 patients with "typical" neuro-ophthalmic field defects underwent automated perimetry using the Swedish Interactive Threshold Algorithm and frequency doubling perimetry. The sensitivity and specificity for identification of a field defect (frequency doubling perimetry 20-5 and 20-1 screening tests), or to characterize hemianopia/quadrantanopia (full threshold test) were determined. Ability to discriminate glaucomatous defects was determined by comparing frequency doubling perimetry full threshold test in glaucoma to pooled results of normal and neuro-ophthalmic groups.
Results: On frequency doubling perimetry, a single point depressed to less than 1% probability had a sensitivity of 97.1% (20-5 test) and 95.7% (20-1 test) for detecting a neuro-ophthalmic visual field defect. The corresponding specificities were 95% using pooled results in normal subjects and patients with glaucoma and "other" neuro-ophthalmic field defects. In 20-5 screening a single abnormal point depressed to less than 2% probability level had a sensitivity of 98.6% (specificity 85%). Two abnormal points in the 20-1 screening depressed to less than 1% probability level had a specificity of 100% (sensitivity 84.8%). In frequency doubling perimetry full threshold, sensitivity and specificity for detection of hemianopia were 86.8% and 83.2%; for quadrantanopia they were 79.2% and 38.6%. The sensitivity and specificity for categorizing a defect as glaucomatous were 86% and 74.7%.
Conclusions: Frequency doubling perimetry is a sensitive and specific test for detecting "neuro-ophthalmic" field defects. The presence of two abnormal points (20-1 screening program) "rules in" the presence of a field defect. A normal 20-5 program (absence of a single abnormal point) almost "rules out" a defect. Frequency doubling perimetry could not accurately categorize hemianopic, quadrantanopic, or glaucomatous defects.