Accuracy of Noncycloplegic Refraction for Detecting Refractive Errors in School-aged African Children

Cycloplegia is often restricted in screening settings. Previous studies have evaluated noncycloplegic refraction for screening, but their data are not fully applicable to school-aged African children. This article adds to the literature by investigating this in school-aged African children.

The purpose of this study was to evaluate the ability of noncycloplegic autorefraction (NCA) and noncycloplegic retinoscopy (NCR) to detect refractive errors in school-aged African children and quantify differences between noncycloplegic and cycloplegic refraction measures.

Autorefraction and retinoscopy were performed on 150 children aged 6 to 15 years before and after cyclopentolate cycloplegia. Subjective refraction was performed by the same examiner after cycloplegia. Noncycloplegic and cycloplegic measurements were compared, and the data were assessed with receiver operating characteristic curves and Bland-Altman plots.

With the exception of cylindrical vector J0 (P = .17) and J45 (P = .08) obtained with NCR, all mean measures of NCA and NCR significantly differed from mean cycloplegic subjective refraction measurements by -0.71 ± 0.67 (P ≤ .001; 95% confidence interval, -2.02 to +0.60) and -0.75 ± 0.68 (P ≤ .001; 95% confidence interval, -2.08 to +0.58), respectively. Mean spherical power between the two tests did not differ (NCA -NCR, mean bias of -0.037 ± 0.675 and 95% limits of agreement of -1.36 to 1.286; P = .51). The sensitivities for detecting any refractive error for NCR and NCA were 86.8 and 82.9%, respectively, and the specificities were 82.0% for NCR and 82.4% for NCA. Performance of both tests was excellent for myopia (area under the curve, >0.90; sensitivity, ≥90%) and similar (P = .13), but the specificities for detecting hyperopia were 78.0% for NCR and 80.0% for NCA.

Noncycloplegic refraction may accurately detect astigmatism and myopia in this population but may elicit higher myopic values and lower hyperopic values than using cycloplegic refraction.