Comparison Of Prediction Errors Between Iolcon And Escrs Constants Using Eyetemis
Published 2026 - 44th Congress of the ESCRS
Reference: POCAT624 | Type: Free paper | DOI: 10.82333/5xav-n449
Authors: Myung-Sun Song* 1 , Yea Eun Lee 1 , Nahyun Park 1 , Chung Min Lee 1 , Jeewon Han 1 , Alison Hong 2 , Kyu Sang Eah 1 , Ho Seok Chung 1 , Jae Yong Kim 1 , Dante Buonsanti 3 , Hun Lee 4
1Ophthalmology,Asan Medical Center,Seoul,Korea, Republic Of, 2Ophthalmology,George Washington University,Washington DC,United States, 3Centro Buonsanti,Buenos Aires,Argentina, 4Ophthalmology,Asan Medical Center,Seoul,Korea, Republic Of;Center for Cell Therapy,Asan Medical Center,Seoul,Korea, Republic Of
Purpose
To compare the refractive prediction performance of IOLCon and ESCRS constants across multiple modern intraocular lens (IOL) power calculation formulas.
Setting
Single center, retrospective study
Methods
This retrospective analysis included 164 eyes. Biometric data were analyzed using the Eyetemis platform, which was used to calculate the prediction errors (PE) based on IOLCon and ESCRS constants. Six contemporary formulas were evaluated: Barret Universal II (BUII), Cooke, EVO, Kane, Pearl-DGS, and Hoffer QST (HQST). Trueness was assessed using the trimmed mean spherical equivalent prediction error (SEQ-PE), precision by the dispersion of SEQ-PE, and accuracy by absolute SEQ-PE. The proportion of eyes within ±0.25D, ±0.5D, ±0.75D, and ±1.00D were compared. Cochran’s Q test with post hoc McNemar analysis was performed for categorical comparisons.
Results
Most constants demonstrated a significant negative deviation of trimmed mean PE from zero (P <0.05), except for BUII(IOLCon), Cooke(IOLCon), Pearl(IOLCon), and Cooke(ESCRS). Pairwise comparisons of absolute SEQ-PE were non-significant between most constants, except for between HQST IOLcon and HQST ESRCS (P = 0.022), with the ESCRS constant showing a larger absolute SEQ-PE than the IOLCon constant. The proportion of eyes within ±0.25D, ±0.5D, ±0.75D, and ±1.00D ranged from 37.2-42.7%, 66.5-75.6%, 84.1-89.0%, and 91.5-96.3%, respectively. Although Cochran’s Q test showed a significant overall difference at the ±1.00D threshold (P = 0.03), post hoc pairwise McNemar tests revealed no significant differences between individual constants.
Conclusions
Both IOLCon and ESCRS constants demonstrated largely comparable refractive prediction performance across modern IOL calculation formulas. Although a mild systemic myopic bias was observed in most constant-formula combinations, clinically meaningful differences in absolute prediction accuracy were limited. Notably, for the HQST formula, the ESCRS constant yielded a higher absolute error compared with IOLCon.