Comparative Performance Evaluation of Ten Intraocular Lens Power Calculation Formulas Based on Two Different Optimization Methods

Purpose Comparison of the accuracy of 10 different intraocular lens power calculation formulas with two different optimization methods. Methods Data from 213 patients with normal axial length who underwent cataract surgery were retrospectively reviewed. SRK/T, Barrett Universal II (BUII), Hill-RBF 3.0, Hoffer Q, Haigis, Kane, EVO 2.0, PEARL-DGS, T2, and K6 formulas were used to calculate the mean prediction error (PE), mean absolute error (MAE), and median absolute error (MedAE) based on A-constants calculated according to the User Group for Laser Interference Biometry (Group 1) and three variable optimization (TVO) (Group 2). Additionally, the percentage of PE within ± 0.25 D, ± 0.50 D, ± 0.75 D, and ± 1.00 D was calculated. Results For all formulas, the mean PE in Group 2 was significantly lower than in Group 1. The lowest MAE and MedAE were calculated using the BUII formula in both groups. The MAE was found to be lower in both groups in BUII formula compared to Haigis formula (p = 0.001, p = 0.03). In group 1 the MedAE, in BUII formula was lower than the Haigis, Kane, and K6 formulas (p < 0.05 for all). In group 2 the MedAE was higher in Haigis formula compared to the BUII and Hill-RBF 3.0 formulas (p < 0.05 for both). In Group 2, the percentage of eyes with PE within ± 0.50 D using EVO 2.0 formula and the percentage of eyes with PE within ± 1.00 D using the Haigis formula were significantly higher compared to Group 1. Conclusions The TVO method yields successful results, and its ease of use is its most important advantage.