Accuracy of noncycloplegic refraction in primary school children in southern Thailand.

OBJECTIVE: To evaluate the accuracy of noncycloplegic refraction compared with cycloplegic refraction in diagnosing refractive error in primary school children in southern Thailand. MATERIAL AND METHOD: This is a cross-sectional study. One hundred twenty children aged from 6-13 years who had visual acuity of 20/40 or worse in at least one eye were included. All the children underwent autorefraction using a Nikon model NRK-8000, retinoscopy and subjective refraction without cycloplegia followed by cycloplegic refraction. The spherical power (SP), cylindrical power (CP), cylindrical axis (CA), and spherical equivalence (SE) from each noncycloplegic technique were compared to cycloplegic refraction using the mean difFerence. The authors also showed the percentage agreement between the data obtainedfrom the three noncycloplegic techniques with those from cycloplegic refraction. RESULTS: The SE mean diFference of noncycloplegic autorefraction, retinoscopy and subjective refraction with cycloplegic refraction were -0.85, -0.19, and -0.26, respectively (p < 0.0001). The data for the SP was similar The cylindrical power mean diFferences were -0.18, -0.13, and -0.02, respectively. The percentage agreements of SE between noncycloplegic autorefraction, retinoscopy, and subjective refraction with cycloplegic refraction within +/- 0.5 diopter (D) were 31.25%, 80.84%, and 81.66%, respectively. For the cylindrical power, the percentages of patients who were within +/- 0.5 D were 87.50%, 94.58%, and 97.50%. The percentages of patients who were within 10 degrees of the cylindrical axis were 73.46%, 96.91%, and 97.53%. CONCLUSION: Noncycloplegic retinoscopy and subjective refraction are clinically accurate and can be applied for refractive error screening in primary school children. Noncycloplegic autorefraction has a tendency towards minus over-correction.