Skip to main content

Table 5 Main findings on accuracy of vision screening tests

From: Vision screening in newborns and early childhood

Screening tests for visual impairment Included studies Main findings
Visual acuity tests (assessment of picture identification tests (LEA Symbols chart) and HOTV eye test) 6 studies [6] When screening test cut-offs were set to achieve specificities of 90%:
• Positive LR: 6.1 (95% CI 4.8 to 7.6); ‘an abnormal result moderately increased the likelihood of amblyopia, amblyopia risk factors (strabismus, astigmatism, hyperopia, myopia, anisometropia), or significant nonamblyogenic refractive error’
• Negative LR: 0.43 (95% CI 0.38 to 0.50); ‘a normal result indicated a small decrease in the likelihood’
Ocular alignment tests
(Cover-uncover test)
1 study [6] (n = 3121) Sensitivity to detect strabismus was 60% for specificity set to 90%
• Positive LR: 7.9 (95% CI 4.6 to 14.0)
• Negative LR: 0.73 (95% CI 0.15 to 0.85)
  1 study [5] For detecting strabismus in children at 37 months:
• Sensitivity of 75% (95% CI: 57.7 to 89.9%)
• Specificity of 100%
Stereoacuity tests 4 studies [6] (n = 7801) • Positive LRs: range from 3.6 to 4.9
• Negative LRs: in the minimal range for detecting amblyopia risk factors or significant nonamblyogenic refractive error and in the moderate range for detecting refractive error or strabismus
Combination of clinical tests
(visual acuity, ocular alignment and stereoacuity tests)
4 studies [4,5,6] (n = 1854) • Positive LRs: median of 14; range from 12 to 17 (3 studies); 4.8 (95% CI 2.8 to 8.4; 1 study; n = 141)
• Negative LRs: median of 0.28; range from 0.10 to 0.91
• In one study, the cover-uncover test performed by professionals or a stereoacuity test showed an increased detection of strabismus when combined with visual acuity tests.
Autorefractors 16 studies [6] (n = 16,712) (5 studies recruited children < 3 years of age) • Positive LRs: ‘most studies reported moderate positive LRs’, ‘some studies reported large positive LRs’
• Negative LRs: ‘most studies reported small negative LRs’, ‘some studies reported large negative LRs’
Photo-screeners 11 studies [6] (n = 6187) • Positive LRs: ‘most studies reported moderate positive LRs’
• Negative LRs: ‘most studies reported small negative LRs’
  1 study [16] Prospective study that evaluated the diagnostic accuracy of a digital photo-screener in a school screening programme in Canada [16, 22]. Among the 335 recruited children (98% were 4 or 5 years of age), 271 completed both the screening test and the reference standard test (ophthalmic examination by a physician who was blinded to the photo-screening findings). Results from both tests agreed in 94% of cases.
• Sensitivity and specificity in detecting amblyopia risk factors were 83 and 95% respectively
• Positive and negative predictive values were 73 and 97%, respectively.
When looking more specifically at detection of strabismus, sensitivity and specificity were 46 and 97% respectively, with 13 children detected by reference standard test. The estimated prevalence of strabismus in the population was 4.8%.
Retinal birefringence scanning
(assessment of the Pediatric Vision Scanner (REBIScan))
1 study [6] (n = 102) • Positive LR: 10.4 (95% CI 5.6 to 19.4)
• Negative LR: 0.0
  1. Abbreviations: CI confidence interval, LR likelihood ratio