PURPOSE. The most frequent cause of amblyopia in early childhood is significant refractive error. The purpose of this study is to evaluate the performance of autorefractors as screening tools using the area under the receiver operator characteristic (ROC) curve to assess the overall performance of each instrument. METHOD. Forty-two preschoolers were screened using non-cyclopleged retinoscopy and three autorefractors: Nikon Retinomax K-Plus and Welch Allyn (WA) instruments: Prototype and Sure Sight (SS) 1.22. Cyclopleged retinoscopy was performed as the gold standard. Area under the ROC curve was calculated for each refracting method for the categories of hyperopia, anisometropia, and astigmatism. Referral criteria for each refracting method were picked to simultaneously maximize sensitivity and specificity for each category.
RESULTS. For hyperopia, noncyclopleged retinoscopy (.96) shows the best overall performance as a screening test followed by the Retinomax (.89), WA Prototype (.89), and WA SS 1.22 (.75). For astigmatism, all the refracting methods perform very well (>.96). All refracting methods performed poorly for anisometropia: retinoscopy (.54), Retinomax (.64), WA Prototype (.59), and WA SS 1.22 (.70). The referral criteria to maximize sensitivity and specificity for hyperopia per instrument were as follows: retinoscopy +1.50D, Retinomax +1.00D, WA Prototype +2.50D, WA SS 1.22 +2.50D. The referral criteria for all of the refracting methods for astigmatism is -1.50DC.
CONCLUSIONS. For hyperopia, sensitivity and specificity for each instrument are maximized with different referral criteria. When screening for astigmatism, the overall performance of all refracting methods is similar at -1.50DC. ROC curves show that all refractive methods for detecting anisometropia are weak.