Comparison Of 6 Methods For The Intraocular Lens Power Calculation In High Myopic Eyes

Objective: To compare the accuracy of Barrett Universal II,Hill-RBF SRK/T,Haigis,Holladay 2 and Holladay 1 for the calculation of intraocular lens power in high myopic eyes with cataract.Methods: Retrospective clinical study.Consecutive patients with axial length equal to or more than 26.0 mm that had cataract surgery and IOL implantation at Department of Ophthalmology,Shanghai Ninth People's Hospital,Shanghai Jiao Tong University School of Medicine,Shanghai,China,from June 2018 to October 2018 were retrospectively reviewed.77 eyes of 56 patients were analyzed.All of the patients underwent routine ocular examination before cataract surgery and preoperative biometry data measurement using IOLMaster,uneventful phacoemulsification cataract surgery with in-the-bag implantation of an posterior chamber intraocular lens(Human Optics,MC X11 ASP IOL).The predicted refraction outcomes were calculated by Barrett Universal II,Hill-RBF SRK/T,Haigis,Holladay 2 and Holladay 1.The difference between the actual refraction one month postoperatively and the refraction predicted by the six methods was evaluated as the numerical prediction error.The absolute value of the prediction error was analyzed as the absolute prediction error.Meanwhile,The percentages of eyes that had MAE within ±0.25D?±0.5D?±0.75 D and ±1.0D were calculated for all methods.For the further exploration of relationship between axial length and prediction errors,the eyes were divided into subgroups according to the axial length.Results: The mean axial length and K value of the eyes was 29.23 ± 1.92 mm and 44.03 ± 1.54 D.The mean numerical prediction error(MNE)± standard deviation of the six intraocular lens power calculation methods was Barrett Universal II(0.41 ± 0.56 D),Hill-RBF(0.44 ± 0.57 D),SRK/T(0.51 ± 0.63 D),Haigis(0.59 ± 0.60 D),Holladay 2(0.96 ± 0.68 D),Holladay 1(1.08 ± 0.67 D).The results of median absolute errors(Med AE)for all methods were Barrett(0.49 D),Hill-RBF(0.54 D),SRK/T(0.54 D)and Haigis(0.64D).There were significant difference among the MNEs of six methods,as well as Med AEs(P < 0.001).Barrett Universal II,Hill-RBF SRK/T and Haigis had smaller MNEs and Med AEs than the Holladay 2 and Holladay 1 statistically significantly(all P < 0.001).The percentages of eyes within ±0.50 D from target refraction of Barrett Universal II,Hill-RBF SRK/T and Haigis were 0.65%,46.75%,42.86% and 37.66% respectively,larger than the Holladay 2(19.48%)and Holladay 1(19.48%)statistically significantly(all P < 0.05).The Barrett Universal II(84.42%)?Hill-RBF(81.82%)?SRK/T(80.52%)and Haigis(74.03%)methods had larger percentage of eyes within ±1.00 D from target refraction than the Holladay 2(51.95%)and Holladay 1(41.56%)(all P < 0.01).There was no significant difference among the mean numerical errors of Barrett Universal II,Hill-RBF,SRK/T and Haigis.The mean numerical errors predicted by Barrett Universal II,Hill-RBF,SRK/T and Haigis were significantly lower than Holladay 2 and Holladay 1(P < 0.05).The results of the median absolute errors were similar as the mean numerical errors(P < 0.05).There was no significant difference within different AL subgroups of Barrett Universal II,Hill-RBF and Haigis methods.However,with the increase of AL,the SRK/T,Holladay 1 and Holladay 2 formulas produced lower predictability(P < 0.05).Furthermore,the Holladay 1 and Holladay 2 formulas produced higher prediction errors than the Barrett Universal II,Hill-RBF,SRK/T and Haigis in all AL subgroups.Conclusions: The Barrett Universal II,Hill-RBF,SRK/T and Haigis were more acurate than Holladay 2 and Holladay 1 in the calculation of IOL power for the high myopic eyes.The Barrett Universal II formula performed best in the prediction error,percentage of eyes within target refraction and subgroup of axial length for the high myopia eyes.The new algorithm,Hill-RBF method,was comparable to the theoretical formulas,such as SRK/T and Haigis.