Comparing Telephone Survey Responses to Best-Corrected Visual Acuity to Estimate the Accuracy of Identifying Vision Loss: Validation Study

**Background:** Vision loss affects approximately 7.08 million Americans and costs the US economy $134 billion per year. Surveillance of vision loss relies on either examination studies (gold standard BCVA) or self-reported survey questions. Self-reported data are cheaper and allow larger samples but their validity for predicting prevalence and disparities in visual acuity has not been established. This study aimed to estimate the diagnostic accuracy of self-reported vision loss measures compared to BCVA at both individual and population levels.

**Methods:** The study was a retrospective chart review and concurrent telephone interview among patients from University of Washington ophthalmology and optometry clinics. From a patient-level electronic health record file (visits May 2018–April 2020), 1200 patients were sampled, oversampling for age-related macular degeneration (20%), diabetic retinopathy (20%), glaucoma (20%), and two additional cohorts (20% each) with or without visual acuity loss. Between summer 2020 and spring 2021, 669 patients consented, and 438 (65.5%) completed the telephone survey. Chart abstraction captured BCVA in each eye, assigned a single logMAR value based on the better-seeing eye. The telephone survey included three questions from national surveys: Q1 ("Are you blind or do you have serious difficulty seeing, even when wearing glasses?"), Q2 ("At the present time, would you say your eyesight, with glasses or contact lenses if you wear them, is excellent, good, fair, poor, or very poor"), Q3 ("Are you blind or unable to see at all?"), plus two additional questions (Q4: "Have you ever been told by a doctor that you have visual impairment?" and Q5: "...are blind?"). Diagnostic accuracy was assessed using AUC, sensitivity, and specificity. Population-level accuracy was assessed by comparing prevalence rates overall and by subgroup.

**Key Results:** The sample was 54.1% female, 70.5% non-Hispanic White, 57.3% aged 65-84 years. BCVA distribution: 64.2% normal (>20/40), 17.8% mild impairment (20/40 to <20/80), 4.8% moderate impairment (20/80 to <20/200), 13.2% blind (≤20/200). For any vision loss (BCVA ≤20/40), Q2 (responses "fair," "poor," or "very poor") had the highest AUC of 0.716. Q4 had the highest sensitivity (0.79), Q3 had 100% specificity. For blindness (BCVA ≤20/200), Q1 had the highest AUC of 0.797, with sensitivity 0.43 and specificity 0.930. Q4 had the highest sensitivity (0.91), Q3 had 0.995 specificity. At the population level, prevalence rates from Q1 and Q2 were highly correlated with BCVA prevalence across most demographic groups. Q2 responses of "fair," "poor," or "very poor" had an overall prevalence of 32.9% vs. BCVA vision loss of 35.8%. Q1 "yes" responses had 21.9% prevalence vs. BCVA blindness of 13.2%. The relative relationships were stable across subgroups except those with small sample sizes (e.g., "Other race" subgroup with only 12 patients).

**Clinical Implications:** Self-reported survey questions are not sufficiently accurate to replace clinical vision evaluation at the individual level (AUC <0.80 for most comparisons). However, at the population level, Q1 and Q2 provide a stable and correlated signal of vision loss prevalence across demographic groups, supporting their use in national surveillance systems like the CDC's Vision and Eye Health Surveillance System (VEHSS) and WHO Global Burden of Disease. The findings validate the use of self-reported data to predict variation in objectively measured acuity loss among populations where examination data are unavailable. Limitations include a clinic-based sample not representative of the general population, oversampling for vision loss, retrospective chart review, and lack of other vision measures (e.g., visual field, contrast sensitivity).