Choosing appropriate tools and referral criteria for vision screening of children aged 4–5 years in Canada: a quantitative analysis

OBJECTIVES: To assess the diagnostic accuracy of five vision screening tools used in a school setting using sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV). DESIGN: We compared the results of the five best evidence-based screening tools available in 2014 to the results of a comprehensive eye exam with cycloplegic refraction by a licenced optometrist. Screening included Cambridge Crowded Acuity Cards, Plusoptix S12 and Spot photoscreeners, Preschool Randot Stereoacuity Test and the Pediatric Vision Scanner (PVS). Referral criteria followed AAPOS (2013) guidelines and published norms. SETTING: A large school in Toronto, Canada, with 25 split classrooms of junior kindergarten (JK: 4 year olds) and senior kindergarten (SK: 5 year olds) children. PARTICIPANTS: Over 2 years, 1132 eligible children were enrolled at the school. After obtaining parental consent, 832 children were screened. Subsequently, 709 children had complete screening and optometry exam data. MAIN OUTCOME MEASURES: The presence/absence of a visual problem based on optometrist’s assessment: amblyopia, amblyopia risk factors (reduced stereoacuity, strabismus and clinically significant refractive errors) and any other ocular problem (eg, nystagmus). RESULTS: Overall, 26.5% of the screened children had a visual problem, including 5.9% with amblyopia. Using all five tools, screening sensitivity=84% (95% CI 78 to 89), specificity=49% (95% CI 44 to 53), PPV=37% (95% CI 33 to 42), and NPV=90% (95% CI 86 to 93). The odds of having a correct screening result in SK (mean age=68.2 months) was 1.5 times those in JK (mean age=55.6 months; 95% CI 1.1 to 2.1), with sensitivity improved to 89% (95% CI 80 to 96) and specificity improved to 57% (95% CI 50 to 64) among SK children. CONCLUSIONS: A school-based screening programme correctly identified 84% of those kindergarten children who were found to have a visual problem by a cyclopleged optometry exam. Additional analyses revealed how accuracy varies with different combinations of screening tools and referral criteria.