Abstract
PURPOSE: Visual screening is important in public health, but visual acuity (VA) testing is costly at scale. We used items from a validated VA survey with 100 yes/no questions to identify screening questions for low vision (VA ≤ 0.3 logMAR or 20/40) and legal blindness as defined by the VA criterion (VA ≤ 1 logMAR or 20/200). METHODS: We analyzed data from 385 adults with measured VA who completed the survey. Participants were classified positive if they answered "no" to selected single, pair, and triplet items for each threshold. We computed sensitivity, specificity, and area under the curve (AUC). RESULTS: For low vision, "In a fast-food restaurant, can you read the posted menus?" yielded AUC = 0.85 (sensitivity = 0.89; specificity = 0.81). For legal blindness, "When seated in front of a computer, are you able to read the letters on the keyboard keys?" produced AUC = 0.80 (sensitivity = 0.89; specificity = 0.71). For both screening thresholds, adding a second item increased AUC by ≤0.05; a third item gave no further improvement. CONCLUSIONS: One targeted question can screen for low vision or acuity-defined legal blindness with good discrimination; a two-item pair offers a compact option when higher specificity is desired. These brief, functional screens are practical when in-person VA testing is impractical.