Sensitivity, specificity, and predictive values of screening tests for eye conditions in a clinic-based population

To assess four commonly available visual function tests to detect visually disabling or vision-threatening eye conditions among new patients of a large, urban, public, general ophthalmology clinic.

Three hundred seventeen patients were tested for contrast sensitivity, Amsler grid abnormalities, and visual acuity at near and at distance. A complete eye evaluation found the prevalence of serious eye diseases, allowing determination of the sensitivity (Sn), specificity (Sp), likelihood ratio (LR), and other characteristics of each test.

Of 317 patients, most were Hispanic (77%), women (60%), and middle-aged (44 +/- 17 years). Normal findings were reported in 18%; refractive error in 43%; cataracts in 16%; glaucoma in 7.3%; and macular degeneration in 4.1%. Near visual acuity of 20/40 or worse (Sn = 0.75; Sp = 0.74; LR = 2.8); and distance visual acuity testing of 20/30 or worse (Sn = 0.74; Sp = 0.73; LR = 2.7) correlated significantly with ocular disease, whereas contrast sensitivity testing (Sn = 0.62; Sp = 0.41; LR = 1.1) and Amsler grid test (Sn = 0.19; Sp = 0.92; LR = 2.4) did not. Test performance decreased when refractive errors were excluded and among those younger than 40 years of age relative to those 40 years of age or older.

Of the four screening tests studied, distance and near threshold visual acuities as defined above were judged to have the best correlations of an abnormal result with ocular disease, both including or excluding refractive error. Different combinations of tests did not result in more accurate detection of ocular disease. More efficient screening tools for detecting ocular disease need to be developed.