Comparison of the Accuracy of the Web-Based ESCRS IOL Power Calculator Formulas

Purpose To compare the accuracy of modern and conventional intraocular lens (IOL) power calculation formulas available on the ESCRS calculator and the IOLMaster 700 using 6-month postoperative refraction, and to assess the correlation between prediction errors and preoperative biometric variables.Methods This retrospective cross-sectional study analyzed a total of 223 eyes from 223 patients undergoing cataract surgery with TECNIS Eyhance ICB00 IOL. Preoperative biometry was obtained using the IOLMaster 700 and CASIA2. Seven formulas from the ESCRS calculator and 4 conventional formulas from IOLMaster 700 were evaluated. Lens constants were optimized to achieve a mean prediction error (PE) of zero. Primary outcome measures included median absolute error (MedAE), root mean square absolute error (RMSAE), and the percentage of eyes within ±0.50 diopters (D) of prediction error. Formula Performance Index (FPI) was used to assess overall accuracy. Correlation between PE and biometric variables was examined using univariate and multivariate regression.Results All formulas demonstrated comparable refractive accuracy. MedAE ranged from 0.29 to 0.33 D (p = .921) and RMSAE from 0.416 to 0.465 D (p = .437), and 72.2% to 78.5% of eyes were within ±0.50 D (p = .776), with no statistically significant differences among formulas. FPI ranged from 0.617 (SRK/T) to 0.666 (Cooke K6). In multivariate regression, the proportion of PE variance explained by preoperative variables was lowest for Cooke K6 (R2 = 0.057) and highest for SRK/T (R2 = 0.242).Conclusions Both modern and conventional IOL formulas provided comparable refractive outcomes when lens constants were optimized and standardized biometry was used. The strength of the association between prediction error and preoperative biometric parameters varied across formulas, which merits further evaluation in studies with broader biometric representation.