Screening Questions to Identify Visual Impairment and Legal Blindness

Abstract

Purpose : This study examined the efficacy of using simple Yes/No questions to classify individuals as visually impaired or legally blind, offering an alternative to standard visual acuity (VA) tests.

Methods : We developed a visual acuity survey comprising 100 yes/no questions, assessing participants’ ability to recognize objects at typical distances. The items were selected to challenge a wide range of VA. All participants had their VAs measured using letter acuity charts. Using Item Response Theory, we estimated the vision ability of each participant and the difficulty of each item. The linear relationship between estimated vision abilities and measured VAs was used to derive VA thresholds for each item. Items with VA thresholds near 20/40 (0.3 logMAR) and 20/200 (1.0 logMAR) were analyzed for their diagnostic odds ratio (DOR, see Fig.), sensitivity, and specificity.

Results : 385 participants from four U.S. low-vision clinics and research labs were recruited. The average age was 56.6 years (18 to 93), and the mean VA was 0.58 logMAR (-0.3 to 2.0, SD = 0.45). Two items selected for VA screening at the 0.3 and 1 logMAR thresholds were "When holding a newspaper at a regular viewing distance of 16 inches, are you able to read the printed text in the articles? (VA threshold: 0.31 logMAR)" and "In daylight, when waiting to cross at an intersection, are you able to see whether the stoplight is red, green or yellow? (VA threshold: 0.93 logMAR)". 91% (269/295) of participants with VA poorer than 0.3 logMAR responded No to the newspaper item, showing 91% sensitivity. 73% (66/90) with better VA said Yes, indicating 73% of specificity. The DOR of 28.5 for the newspaper item implies that a 'No' response is 28.5 times more likely from visually impaired individuals than from normally sighted individuals. The stoplight item showed 52% sensitivity (38/73) and 91% specificity (283/312) in detecting legal blindness, with a DOR of 10.6. However, its efficacy might be affected by factors beyond VA, such as color vision and memory. Reduced sensitivity could stem from the fewer legally blind participants, where the error in the estimate could notably affect sensitivity.

Conclusions : The findings suggest the newspaper item is effective in identifying visual impairment, whereas the stoplight item is more useful in identifying those who are not legally blind. This simple tool could prove beneficial for telehealth services and community surveys.

This abstract was presented at the 2024 ARVO Annual Meeting, held in Seattle, WA, May 5-9, 2024.