The optimal cut-off value of non-cycloplegic autorefraction for diagnosing myopia in school-aged children.

Abstract

To determine the optimal cut-off value of non-cycloplegic autorefraction for diagnosing myopia and estimating myopia prevalence in school-aged children. Retrospective case-control study. Based on age and sex, case-control data were extracted from the medical records of children who underwent autorefraction before and after instillation of 1% cyclopentolate eye drops. We regarded a spherical equivalent (SEQ) of -0.50 D or less found by cycloplegic autorefraction as myopia and determined the optimal cut-off value of non-cycloplegic autorefraction for diagnosing myopia using a receiver operating characteristic (ROC) curve and diagnostic performance (DP) plots. Of a total of 232 children (mean age 8.2 [range: 6-12] years, 126 boys [54.3%]), 116 (50.0%) had myopia. In this cohort, the optimal cut-off SEQ for diagnosing myopia was -0.75 D with a sensitivity of 90.5% and specificity of 95.7%. However, the DP-plots indicated that the cut-off value significantly varied with prevalence of myopia: -1.31 D, -0.81 D, and -0.65 D for the prevalence of 30%, 50%, and 80%, respectively.For non-cycloplegic autorefraction, we found greater accommodation in children aged 6 years and hyperopic eyes (p < 0.001). When diagnosing myopia using non-cycloplegic autorefraction alone with a theoretical cut-off SEQ of -0.50 D, the prevalence of myopia will be overestimated, and we need to set the cut-off value lower (more myopic) especially in younger children with low prevalence.