How to Conduct School Myopia Screening: Comparison Among Myopia Screening Tests and Determination of Associated Cutoffs

Purpose: To compare the accuracy for various screening tests and their combined uses for myopia screening among children and adolescents and explore age-specific cutoffs. Design: Cross-sectional study. Methods: A total of 6017 children and adolescents aged 4 to 15 years participated in the study. Uncorrected visual acuity (UCVA, recorded in decimal notation), cycloplegic and noncycloplegic refraction (NCR), axial length (AL), and corneal curvature radius (CR) examinations were performed. Cycloplegic spherical equivalent ≤−0.50 D was considered as the gold standard for myopia. Receiver operating characteristic (ROC) curves were drawn to determine optimal cutoffs for all age groups, and sensitivity, specificity, as well as screening prevalence of myopia were calculated. Results: The overall estimate of myopia prevalence was 31.8% using the gold standard. The sensitivity and specificity of the UCVA alone for the commonly used cutoff (1.0) were 97.7% and 33.1%, respectively. The areas under the ROC curve were optimally estimated to be 0.985 (95% CI, 0.982–0.988) for the combined use of UCVA and NCR tests, and 0.987 (95% CI, 0.983–0.989) for the combined use of AL/CR and NCR tests, with no significant difference (P = 0.208). The best cutoffs for UCVA-NCR combinations were UCVA <1.0 and NCR <−0.25 D in 4 to 6 years; UCVA <1.0 and NCR <−0.50 D in 7 to 12 years; UCVA <0.8 and NCR <−0.75 D in 13 to 15 years. If those screening positive were all referred to clinics and corrected with cycloplegic autorefraction data, the relative difference between screening prevalence and the actual prevalence by the gold standard would reduce from 13.2% to 4.7%. Conclusions: UCVA test alone for detecting myopia demonstrated a poorer accuracy among these tests. The combined use of UCVA and NCR tests and the combined use of AL/CR and NCR tests achieved optimal accuracy for myopia screening. Setting age-specific cutoffs would increase the accuracy, and the prevalence obtained from primary screening should be corrected according to the data of cycloplegic refraction after referral, especially in younger ages.