Refractive prediction error after intrascleral intraocular lens fixation in eyes with long axial length

To evaluate the refractive prediction accuracy of intraocular lens (IOL) power calculation formulas after flanged intrascleral IOL fixation (SF-IOL; Yamane technique) in long axial length (AL) eyes.

University of Osaka Hospital, Osaka, Japan.

Retrospective observational cohort study.

Eyes undergoing SF-IOL between January 2023 and April 2024 were classified into normal (N, <26.0 mm) and long (L, ≥26.0 mm) AL groups. Barrett Universal II, Hoffer Q, Holladay 1, and SRK/T were evaluated. Prediction error (PE) was defined as 3-month postoperative spherical equivalent minus predicted value. Prediction accuracy was assessed using root mean square absolute error (RMSAE) and the Formula Performance Index (FPI).

53 eyes from 50 patients were analyzed (N = 35; L = 18). Mean AL was 24.6 ± 0.8 mm (N) and 27.1 ± 1.0 mm (L). Mean PE (N and L) was -0.55 ± 0.72 diopters (D) and -0.64 ± 0.62 D with Barrett Universal II ( P = .52), -0.68 ± 0.77 D and -0.45 ± 0.62 D with Hoffer Q ( P = .32), -0.78 ± 0.74 D and -0.31 ± 0.72 D with Holladay 1 ( P = .043), and -0.74 ± 0.72 D and -0.62 ± 0.72 D with SRK/T ( P = .60). In the L group, Holladay 1 showed PE nearest zero; however, demonstrated steep AL-PE regression slope and low FPI. Barrett Universal II yielded the lowest RMSAE and the highest FPI.

Barrett Universal II demonstrated the highest overall predictive accuracy in long AL eyes undergoing SF-IOL. Although Holladay 1 produced a smaller PE in long AL, this reflected systematic offset rather than superior accuracy, as indicated by its lower FPI.