Agreement and diagnostic accuracy of noncycloplegic retinoscopy, subjective and autorefraction in a young adult Ghanaian population

Objective: To determine the agreement and diagnostic accuracy of retinoscopy, subjective and autorefraction among young adults in a sampled Ghanaian population. Design: A descriptive cross-sectional design. Setting: Optometry Department of Bishop Ackom Memorial Christian Eye Centre, Cape Coast, Ghana. Participants: Refractive error data were obtained from 100 adults aged 18-35 years (200 eyes)using autorefraction, retinoscopy and subjective refraction. Main outcome measures: Spherical equivalent, J0 and J45 values, and sensitivity and specificity for myopia and hyperopia detection. Results: The mean age was 23.38 ± 4.39 years, 61% females. This study used Bland-Altman analysis and intraclass correlation to compare refractive techniques. For spherical equivalents, retinoscopy and subjective refraction yielded high agreement (ICC=0.9), and autorefraction and subjective refraction exhibited significant agreement (ICC=0.8, p < 0.001). Retinoscopy and subjective refraction showed a significant correlation (ICC = 0.81, p < 0.001). Similarly, autorefraction and subjective refraction for J0 correlated (ICC=0.67, p < 0.001), and retinoscopy and subjective refraction showed low agreement on the J45 axis (ICC=0.01, p=0.48). Autorefraction and subjective refraction were weakly but significantly correlated (ICC=0.22, p < 0.001). The specificity and sensitivity of retinoscopy in detecting myopia are similar to those of autorefraction. The accuracy of retinoscopy was greater than that of autorefraction in detecting myopia. However, retinoscopy and autorefraction demonstrated equal accuracy in detecting hyperopia. Conclusion: In young adults, retinoscopy shows strong agreement with subjective refraction, demonstrating its consistency in evaluating spherical equivalents and the J0 axis. Although efficient, autorefraction showed reduced precision and less consistency for J45 axis determination. Funding: None declared.