Accuracy and precision of visual acuity tests: A systematic review and meta-analysis across testing strategies
- 1UC Irvine
Abstract
Purpose: Visual acuity (VA) testing has expanded from examiner-guided to examinee self-guided assessments, yet measurement variability may limit clinical utility. This meta-analysis investigated VA-test's intertest and test-retest accuracy and precision across different testing strategies. Methods: Following PRISMA guidelines, we searched PubMed, Google Scholar, and Scopus for VA validation studies published between 1 January 1983 to 02 December 2024. Eligible studies reported intertest or test-retest Bland-Altman mean bias and limits of agreement (LoA); objective VA assessments and correlation coefficient only studies were excluded. Testing strategies were classified by test site, optotype sampling methods, response input methods, and level of supervision. Accuracy outcomes were reported as mean biases. Precision outcomes were reported as intertest LoA half-width and test-retest coefficient of repeatability (CoR) for test-retest. Multilevel random-effects models accounted for multiple estimates per study. Risk of bias was assessed using QUADAS-2. Results: Across 82 included studies (10,873 participants), pooled intertest and test-retest mean biases were both 0.01 logMAR. Intertest LoA half-width was 0.19 logMAR and test-retest CoR was 0.14 logMAR. Variability showed marginal difference by optotype sampling (adaptive: LoA half-width 0.20 logMAR, CoR 0.15 logMAR; fixed: 0.18, 0.13) or response input method (examiner: 0.19, 0.13; self-run: 0.19, 0.18). Only few studies reported results for test site and level of supervision, hence only intertest was compared (test site, clinic: LoA half-width 0.19 logMAR vs remote 0.27 logMAR and level of supervision, supervised: 0.19 logMAR vs unsupervised 0.26 logMAR). QUADAS-2 identified patient selection as the domain of greatest bias concern. Conclusion: Across testing strategies, VA assessments showed minimal systematic bias but substantial variability. Lack of precision is more notable in remote and unsupervised testing while no difference within optotype sampling and response-input methods. Future development should prioritize on improving precision of visual acuity tests with finer adaptive optotype resolution and remote/unsupervised testings through repeatability studies.
Related
- AI-Assisted Clinical Decision Support Versus Standard Teleconsultation in Adult Medicine: A Systematic Review of Diagnostic Accuracy, Referral and Testing Decisions, and Clinician Confidence.
- Outcomes and organ protection effects of perioperative precision venous access management strategies in patients undergoing complex hepatobiliary, pancreatic, splenic and abdominal great vascular surgery: a systematic review and meta-analysis.
- Accuracy of Visual Estimation of Active Shoulder Elevation Among Orthopedic Physicians: An Exploratory Video-Based Study
- Comparison of 24-2 and 24-2C Visual Field Testing in Eyes with Ocular Hypertension
- Safety and efficiency of the HEART-GP strategy with point-of-care high-sensitivity troponin testing for acute chest pain in out-of-hours primary care: a prospective multicentre diagnostic accuracy study
BibTeX
@misc{chen2026accuracy,
title = {{Accuracy and precision of visual acuity tests: A systematic review and meta-analysis across testing strategies}},
author = {C. Chen and A. W. Browne and J. Skerswetat},
howpublished = {medRxiv},
year = {2026},
month = nov,
doi = {10.64898/2026.11.04.26366580},
url = {https://www.medrxiv.org/content/10.64898/2026.11.04.26366580v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Accuracy and precision of visual acuity tests: A systematic review and meta-analysis across testing strategies
AU - C. Chen
AU - A. W. Browne
AU - J. Skerswetat
PY - 2026
DA - 2026/11/04
PB - medRxiv
DO - 10.64898/2026.11.04.26366580
UR - https://www.medrxiv.org/content/10.64898/2026.11.04.26366580v1
AB - Purpose: Visual acuity (VA) testing has expanded from examiner-guided to examinee self-guided assessments, yet measurement variability may limit clinical utility. This meta-analysis investigated VA-test's intertest and test-retest accuracy and precision across different testing strategies. Methods: Following PRISMA guidelines, we searched PubMed, Google Scholar, and Scopus for VA validation studies published between 1 January 1983 to 02 December 2024. Eligible studies reported intertest or test-retest Bland-Altman mean bias and limits of agreement (LoA); objective VA assessments and correlation coefficient only studies were excluded. Testing strategies were classified by test site, optotype sampling methods, response input methods, and level of supervision. Accuracy outcomes were reported as mean biases. Precision outcomes were reported as intertest LoA half-width and test-retest coefficient of repeatability (CoR) for test-retest. Multilevel random-effects models accounted for multiple estimates per study. Risk of bias was assessed using QUADAS-2. Results: Across 82 included studies (10,873 participants), pooled intertest and test-retest mean biases were both 0.01 logMAR. Intertest LoA half-width was 0.19 logMAR and test-retest CoR was 0.14 logMAR. Variability showed marginal difference by optotype sampling (adaptive: LoA half-width 0.20 logMAR, CoR 0.15 logMAR; fixed: 0.18, 0.13) or response input method (examiner: 0.19, 0.13; self-run: 0.19, 0.18). Only few studies reported results for test site and level of supervision, hence only intertest was compared (test site, clinic: LoA half-width 0.19 logMAR vs remote 0.27 logMAR and level of supervision, supervised: 0.19 logMAR vs unsupervised 0.26 logMAR). QUADAS-2 identified patient selection as the domain of greatest bias concern. Conclusion: Across testing strategies, VA assessments showed minimal systematic bias but substantial variability. Lack of precision is more notable in remote and unsupervised testing while no difference within optotype sampling and response-input methods. Future development should prioritize on improving precision of visual acuity tests with finer adaptive optotype resolution and remote/unsupervised testings through repeatability studies.
ER -