Abstract
Purpose :
There are several studies comparing the intraocular lens (IOL) power calculations for Caucasian eyes. Whereas, there are morphological differences in Japanese eye such as axial length (AL), corneal power, and anterior chamber depth, so that outcomes of IOL power calculation would be influenced. This retrospective study was aimed to evaluate the accuracy of 4 IOL power calculations for Japanese eyes of short and long ALs.
Methods :
There were 37 and 80 eyes of 117 cataractous patients with AL shorter than 22.0mm and longer than 26.0mm, respectively. Biometry data was measured with an optical biometer (OA-2000, TOMEY). Predicted postoperative refractions for implanted IOL were obtained in the use of the SRK/T, Haigis, Barrett Universal II, and Hill-Radial Basis Function (Hill-RBF). Refractive errors from manifest refraction 1 month postoperatively were calculated and mean absolute prediction error (MAE) and rate of ±0.5D of errors for 4 calculations were compared in short and long eyes each.
Results :
The mean ALs were 21.73mm (range 21.14 to 21.99mm) in short eyes and 27.54mm (range 26.01 to 32.74mm) in long eyes. In short eyes, the MAE were 0.39D, 0.56D, 0.50D, and 0.44D with the uses of SRK/T, Haigis, Barrett Universal II, and Hill-RBF, respectively, and SRK/T was smaller than Haigis (P=0.022). There were 78.4%, 54.1%, 51.4%, and 59.5% of eyes within ±0.5D of errors: SRK/T was significantly higher than Haigis and Barrett Universal II (P=0.049, P=0.028). In long eyes, the MAE were 0.49D, 0.48D, 0.45D, and 0.40D, and the Hill-RBF was smaller than SRK/T (P=0.041). There were 56.3%, 56.3%, 61.3%, and 62.5% of eyes within ±0.5D of errors, and no significant difference was not found.
Conclusions :
For Japanese eyes, the use of SRK/T and Hill-RBF resulted in the lowest predictive error in short and long eyes, respectively. The comparison results did not coincide with previous studies completely. Adaptation of IOL power calculations for morphological differences in patients would be important before use.
This abstract was presented at the 2019 ARVO Annual Meeting, held in Vancouver, Canada, April 28 - May 2, 2019.