AUTOMATED SUPRATHRESHOLD SCREENING FOR GLAUCOMA - THE BALTIMORE EYE SURVEY

Purpose. To evaluate automated suprathreshold perimetric screening for glaucoma in a population-based survey of ocular disorders in east Bal timore, Maryland. Methods. A population-based sample of persons greate r-than-or-equal-to 40 years of age residing in 16 clusters was selecte d for an ocular screening examination that included automated suprathr eshold testing with the Full Field 120 program of the Humphrey Field A nalyzer. Subjects who failed the test underwent manual testing to conf irm the defect. Subjects were referred for definitive examination by a n ophthalmologist if they had an abnormal field, visual acuity worse t han 20/30, intraocular pressure > 21 mm Hg, optic disc damage, a histo ry of glaucoma, or shallow angles. The sensitivity and specificity of the automated visual field testing for identifying glaucoma was estima ted and compared with other methods to screen for glaucoma. Results. O f 5,341 subjects greater-than-or-equal-to 40 years of age who under-we nt a screening eye examination at neighborhood centers, 4,735 (89%) co mpleted the automated field test. The median test time was 7.25 minute s per eye. Screening test results were abnormal in one or both eyes in 1,234 (26%) of the subjects. Kinetic perimetry was performed on 95% o f these subjects, and defects were confirmed for 448 (36%) of them. He nce, 9.5% of the 4,735 subjects who completed the automated test were referred for definitive examination because the defect on automated pe rimetry was confirmed on manual testing. For a specificity of 90%, the sensitivity of the screening visual field test to detect glaucoma was 52% for 17 or more relative or absolute defects, higher than that of intraocular pressure at 39% for a cut-off of 20.5 mm Hg, vertical cup- to-disc ratio at 45% for a cut-off of 0.53, narrowest remaining rim wi dth at 42% for a cut-off of 0. 1 6, and was comparable to a combinatio n of these and other nonfield parameters. Conclusion. Suprathreshold t esting performed better than nonperimetry-based screening tests for gl aucoma. However, a number of logistical weaknesses of this visual fiel d screening method were identified.