Abstract
Purpose :
To assess the agreement of refreactive error (RE) measures between the PowerRef 3 (PR3) and the Grand Seiko WAM-5500 (GS).
Methods :
Monocular, simultaneous measures (10 seconds) of RE were obtained from the right eye by the GS (6Hz, output-spherical equivalent (SE)) and the PR3 (50Hz, 90th meridian) with a hot mirror placed 45 degrees in front (~1cm) of the eye. Subjects (n=27,14-42 yrs) viewed a high-contrast Maltese cross at 3M (0D) and five near distances (2, 2.5, 3, 4, and 5D demands). Additional measures were obtained at each demand by the GS in static mode using the cylinder and axis values to calculate the power of the eye in the 90th meridian during the dynamic measures. Agreement between the two instruments was assessed using interclass correlation coefficients (ICC) and Bland Altman plots. Analysis was performed using an aggregate of all measures and separately for measures obtained at demands ≤3D.
Results :
The mean differences between the instruments at each demand were: 0D: -0.13±0.52; 2D: 0.03±0.42; 2.5D: -0.16±0.49; 3D -0.14±0.36, 4D: -0.12±0.61; 5D: -0.16±0.98. The absolute mean differences between the instruments were: 0D: 0.41±0.34; 2D: 0.32±0.26; 2.5D: 0.35±0.37; 3D: 0.27±0.26; 4D: 0.47±0.39; 5D 0.78±0.60. Bland Altman plots showed the total mean difference (bias) of -0.09±0.63 with a 95% LOA of ±1.23D. There was no significant relationship between the differences in measures and mean RE (B-coeff: -0.006, P=0.87). Bland Altman plot using the absolute mean difference showed a significant decrease in the differences in measures (B-coeff: -0.06, P=0.01) as RE decreased. The ICC of all measures was 0.91 (95%CI = 0.88, 0.93). For demands ≤3D, the mean difference between the instruments was -0.10±0.45 and 95% LOA of ±0.88D with ICCs of 0.88 (95%CI = 0.83, 0.92).
Conclusions :
Agreement was high between the two instruments for all demands tested and when limiting analysis to demands ≤3D. The mean differences suggest the PR3 may have a small bias towards measuring less myopia, but the bias is not likely to be clinically significant. The differences between the two instruments were greatest at the 4 and 5D demands. Limiting the analysis to measures obtained at demands ≤3D improved the agreement between the two instruments to similarly reported limits of agreement between RE measures obtained by subjective refraction, retinoscopy, and autorefraction.
This is an abstract that was submitted for the 2018 ARVO Annual Meeting, held in Honolulu, Hawaii, April 29 - May 3, 2018.