Accuracy of intraocular lens power calculation in pediatric aphakia with the Artisan iris-claw lens

To compare the refractive accuracy of six intraocular lens (IOL) power calculation formulas in pediatric aphakia managed with anterior iris-claw (Artisan) IOL implantation.

Children aged 18 years or younger with ectopia lentis or microspherophakia who underwent lensectomy and anterior Artisan Aphakia IOL implantation between 2013 and 2024 at the Hospital for Sick Children, Toronto, were reviewed retrospectively. Optical biometry was obtained using the IOLMaster (version 5.4). Predicted refraction was calculated using SRK/T, Holladay 1, Hoffer Q, T2, Barrett Universal II, and EVO 2.0 formulas, with lens constants iteratively optimised to zero mean prediction error. The primary analysis used first-operated eyes; cumulative data for all eyes are presented separately.

Twenty-six first-operated eyes of 26 children were analysed. With manufacturer-supplied constants, SRK/T showed the least systematic bias (mean prediction error - 0.18 D), a mean absolute error (MAE) of 0.75 D, and the highest observed proportion of eyes within ± 0.50 D (61.5%). After lens-constant optimisation, SRK/T had the lowest observed MAE (0.76 D) and median absolute error (0.37 D), and 61.5% of eyes were within ± 0.50 D. T2 and EVO 2.0 showed comparable overall performance. Absolute prediction error did not differ significantly across the six formulas (Friedman test, p = 0.71). Although the proportion within ± 0.50 D differed across formulas overall (Cochran's Q, p = 0.03), no pairwise comparison remained statistically significant after Bonferroni adjustment. Differences at the ± 1.00 D and ± 2.00 D thresholds were not significant. The cumulative 44-eye analysis showed a similar overall pattern.

In this cohort, SRK/T showed the least systematic bias with the manufacturer-supplied constant and the lowest observed absolute prediction errors after optimisation, with T2 and EVO 2.0 performing comparably. No individual formula demonstrated statistically significant superiority after correction for multiple comparisons. These findings require prospective validation in independent cohorts before informing routine formula selection.