Refractive error and axial length prediction using fundus parameters of normal eyes in the Kumejima population study.

Purpose: To evaluate the accuracy of predictive refractive error (RE) and axial length (AL) using regression analysis of fundus parameters in the Kumejima study and to identify RE- or AL-related fundus changes.

Study design: Prospective cross-sectional observational population study.

Methods: Non-mydriatic color fundus photographs (CFPs) from 1,646 right eyes of healthy Kumejima participants were analyzed. Mean red (R), green (G), and blue (B) intensities at eight locations around the optic disc and fovea were quantified, and the tessellation fundus index was calculated as R/(R + G + B). Optic disc ovality ratio, papillomacular angle, and retinal vessel angle were measured. Least absolute shrinkage and selection operator regression with leave-one-out cross-validation predicted RE and AL, validated using Pearson's correlation coefficient.

Results: The mean ± standard deviation actual RE and AL of participants (834 men and 812 women) were -0.14±1.62 diopter and 23.50±0.88 mm. The mean ± standard deviation predicted RE and AL based on fundus parameters was -0.14±1.05 diopter and 23.50±0.48 mm, with a mean absolute error of 0.91 diopter and 0.59 mm, and the correlation coefficients between actual and predicted RE and AL were 0.63 and 0.51 (p<0.001). Eyes with a longer AL had narrower temporal vessel angles, weaker green intensities, stronger blue intensities, and increased tessellation of the fundus color (p<0.001).

Conclusion: RE and AL could be predicted using CFP parameters; the RE- or AL-related changes in the fundus, such as vessel angles and peripapillary color intensity, may enhance our understanding of myopia mechanisms.