Screening with automated perimetry using a threshold-related three-level algorithm

Eighty eyes were tested using the computer-driven, automated threshold-related, three-level screening strategy of the Humphrey Field Analyzer. Using Goldmann perimetry as the standard, different statistical algorithms were used in an attempt to differentiate eyes with glaucoma (n = 24) from normal controls (n = 56). Seventeen or more absolute and/or relative defects anywhere in the full field test area identified approximately 95% of the glaucoma eyes and screened out 80% of the control eyes. Retesting these criteria on a second group of 25 glaucomatous and 108 control eyes provided comparable results. The number of stimulus presentations can be substantially reduced, without deterioration in either sensitivity or specificity, by using a two-level strategy and limiting testing to the nasal field.