Biometry-driven selection of intraocular lens power calculation formulas using the ESCRS IOL calculator.

Abstract

Purpose: To evaluate the accuracy of a biometric subgroup-based intraocular lens (IOL) power calculation formula selection strategy using the ESCRS IOL Calculator.

Setting: Kim's Eye Hospital and Korea University Ansan Hospital, South Korea.

Design: Retrospective comparative study.

Methods: 626 eyes that underwent uneventful cataract surgery with in-the-bag IOL implantation were included. Eyes were divided into a derivation dataset (351 eyes) and an independent external-validation dataset (275 eyes). Biometric subgroups were defined based on combinations of keratometry (K), anterior chamber depth (ACD), and axial length (AL), resulting in 640 subgroups. For each subgroup, the formula with the lowest mean absolute error (MAE) was selected using the derivation dataset. For subgroups not represented in the derivation dataset, formulas were assigned using a median-based approach. Prediction accuracy in the validation dataset was compared between the subgroup-based method and 7 individual formulas.

Results: The subgroup-based method demonstrated the lowest MAE (0.24 ± 0.19 diopters [D]) and median absolute error (0.19 D) among all methods ( P < .001). The proportion of eyes within ±0.50 D was higher with the subgroup-based method (89.1%) compared with individual formulas (approximately 80%; P < .001). The subgroup-based method also showed the highest IOL Formula Performance Index (0.609) compared with individual formulas (0.377 to 0.560).

Conclusions: Selecting IOL power calculation formulas according to biometric subgroups defined by K, ACD, and AL improves refractive prediction accuracy compared with individual formulas. A biometry-driven formula selection strategy using the ESCRS IOL Calculator may provide a clinically useful approach for optimizing refractive outcomes after cataract surgery.