Purpose: To compare Netra smartphone-based and automated refraction with subjective refraction for screening of refractive errors.
Methods: Cross-sectional study at the University of Malaya Medical Centre, Kuala Lumpur. Subjects underwent subjective refraction, then automated refraction, and finally Netra smartphone-based refraction. All results were converted to power vectors (M, J0 and J45) and were analysed using repeated-measures ANOVA and Bland-Altman plots. Sensitivity and specificity were determined. The best cut-off points were determined from ROC curve analysis. P < .05 was considered statistically significant.
Results: Data from the right eyes of 204 subjects were analysed. Mean age was 36.6 ± 15.7 years (range 16-78 years). Spherical equivalent [mean (95% CI)] from Netra and automated refraction were similar, and both more myopic than subjective refraction; -2.87 (-3.23 to -2.51), -2.85 (-3.21 to -2.49) and -2.46 (-2.83 to -2.10) respectively (p < .001). Differences in J0 and J45 between Netra and subjective refraction were not statistically significant (0.10 vs 0.11 and 0.01 vs -0.02 respectively, both p > .05), but those between automated and subjective refraction were (0.06 vs 0.11 and 0.07 vs -0.02, p = .004 and p < .001 respectively). Bland Altman plots showed the 95% limits of agreement with Netra refraction were wider than with automated refraction (-2.21D to 1.42D vs. -1.90D to 1.16D respectively).
Conclusion: Netra smartphone-based refraction gives similar readings to automated refraction, and both show myopic overestimation when compared to subjective refraction. However, due to non-insignificant practical usage issues, its use as a screening tool for refractive errors is limited.
Keywords: Netra; automated refraction; screening; smartphone-based; subjective refraction.