Abstract
Purpose :
With the rising prevalance of myopia, there is increasing need for validated screening tools to identify and monitor refractive error in pediatric populations. We conducted a prospective clinical study to compare the spherical equivalent (SE) refractive error measurements of a newly developed mobile app to manifest refraction (MR) and subjective cycloplegic refraction (SCR) performed in clinic.
Methods :
Data was collected from 54 myopic subjects (61.1% female, mean age 11.7 ± 3.22 years). Eyes with astigmatism more than -1.75D were excluded. Refractive error was measured with the app before cycloplegia. Using a selfie camera to measure the viewing distance in real time, the app displayed three distance-compensated Tumbling E letters in 20/25 angular size. The subject’s refractive error was estimated based on the far point distance at which they correctly identified the orientation of 2/3 letters. Pearson correlation (R) and Bland-Altman 95% limits of agreement were used to compare app measurements to clinical refraction.
Results :
For the 95 eyes enrolled, the SE refractive error range was -0.75D to -6.38D as measured by SCR. 6 eyes with a difference in app measurement and MR more than 1.5x IQR (0.795) were outliers and excluded. The app measurement was highly correlated with MR (R= 0.93, p<0.001), as well as SCR (R= 0.93, p<0.001). A paired t-test showed that the app systematically underestimated SE refractive error by -0.306D (p<0.001) as compared to MR, and by -0.242D (p<0.001) as compared to SCR. The Bland-Altman 95% limits of agreement were [-1.46, 0.846] for the app and MR and [-1.37, 0.887] for the app and SCR. Clinically relevant limits of agreement are ±0.50D and 51.6% of eyes were within 0.50D of difference.
Conclusions :
The moderate Bland-Altman agreement suggested that one-time measurement with the app may not be sufficiently accurate for some myopic patients. However, the high correlation and small systematic bias of the app with both MR and SCR may support its potential for screening for the presence of myopia and its relative progression in pediatric populations, if the agreement can be improved, for example, by repeated measurement. The similarities in comparing the app to MR and SCR demonstrated that the app could be used without cycloplegia and therefore operated by non-professionals.
This abstract was presented at the 2022 ARVO Annual Meeting, held in Denver, CO, May 1-4, 2022, and virtually.