LAUSR

Purpose To investigate the accuracy of 6 intraocular lens (IOL) power calculation formulas in predicting refractive outcomes in extremely long eyes. Setting Department of Ophthalmology, Far Eastern Memorial Hospital, Taiwan. Design Retrospective comparative study. Methods In this retrospective single-center study, we reviewed 70 eyes of 70 patients with axial length (AL) ≥ 28 mm who had received an uneventful 2.2 mm corneal wound phacoemulsification and in-the-bag IOL placement. The actual postoperative refractive results were compared to the predicted refraction calculated with 6 formulas (Haigis, Hoffer Q, Holladay 1, SRK/T, T2, Barrett Universal II formulas) using IOLMaster 500 as optical biometry in the User Group for Laser Interference Biometry (ULIB) constants. Results Overall, the Haigis and Barrett formulas achieved the lowest level of mean prediction error (PE) and median absolute error (MedAE). Hoffer Q, Holladay 1, SRK/T, and T2 had hyperopic prediction errors (p < 0.05). The Hoffer Q and Holladay 1 had significantly more MedAE between the 6 formulas. After the mean PE was zeroed out, the MedAE had no significant difference between each group. The absolute error tends to be larger in patients with longer AL. The absolute errors were 30.0–37.1% and 60.0–64.3% within 1.0 D of all patients compared to predicted refraction calculated using various formulas. Conclusion The Haigis and Barrett Universal II formulas had a better success rate in predicting IOL power in high myopic eyes with AL longer than 28 mm using the ULIB constant in this study. The postoperative refractive results were inferior to the benchmark standards, which indicated that the precision of IOL power calculation in patients with high myopia still required improvement.