Comparison of 10 biometric formulas in combined phacovitrectomy for different underlying retinal pathology

The refractive accuracy of intraocular lens (IOL) formulas varies in eyes that undergo combined phacovitrectomy for different underlying vitreoretinal pathology. A total of 401 eyes that underwent uncomplicated phacovitrectomy (23-25 g) with implantation of a plate haptic IOL (CT Asphina 409 M) in the capsular bag between April 2020 and December 2022 were included in the study. Inclusion criteria were postoperative best corrected visual acuity of 0.4 LogMAR or better at least 8 weeks after surgery. The Barrett Universal II (BUII), Haigis, Hill-Radial Basis Function (Hill-RBFv3.0), Hoffer Q, Holladay I, Holladay II, Kane, K6, Pearl-DGS and SRK/T formulas were compared for their accuracy in mean absolute error (MAE). Furthermore, all formulas were additionally tested by using the prediction for an IOL power 0,5 or 1D above the IOL used (IOLdown). Wilcox-Holladay-Wang-Koch (WHWK) statistical tests with Holm correction were applied. The Barrett IOLdown formula showed the lowest refractive prediction error (0.03D; together with Hill IOLdown), lowest MAE (0.40D), lowest median absolute error (MedAE; 0.31D), lowest standard deviation of MAE (0.55D) and lowest root mean squared absolute error (RMSAE; 0,54D). Barrett IOLdown had the highest percentage of eyes with predicted error within ± 0.25D (41.6%), Kane IOLdown within ± 0.50D (70.5%) and ± 0.75D (89%) and Hill IOLdown within ± 1D (95.2%). Except for Haigis, Hoffer Q and Pearl-DGS, all other formulas showed a statistically significantly lower MAE after IOLdown modification. The newest formulas with IOLdown modification performed better than old generation formulas, with Barrett IOLdown exhibiting the best results.