Comparison of the accuracy of 11 intraocular lens power calculation formulas

Purpose To compare the accuracy of 11 intraocular lens (IOL) power calculation formulas (SRK-T, Hoffer Q, Holladay I, Haigis, Holladay II, Olsen, Barrett Universal II, Hill-RBF, Ladas Super formula, EVO and Kane). Setting Private university hospital (QuironSalud, Madrid, Spain). Design Retrospective case series Methods Data were compiled from 481 eyes of 481 patients who had undergone uneventful cataract surgery with IOL insertion. Preoperative biometric measurements were made using an IOL Master® 700. Respective ULIB IOL constants (http://ocusoft.de/ulib/c1.htm) for each of 4 IOL models implanted were used to calculate the predictive refractive outcome for each formula. This was compared with the actual refractive outcome determined 3 months postoperatively. The primary outcome was mean absolute prediction error (MAE). The study sample was divided according to axial length (AL) into three groups of eyes: short (⩽22.00 mm), normal (22.00–25.00 mm) and long (⩾25.00 mm). Results The Barrett Universal II and Haigis formulas yielded the lowest MAEs over the entire AL range (p < .01, except EVO) as well as in the long (p < .01, all formulas) and normal (p < .01, except Haigis, Holladay II, Olsen and LSF) eyes. In the short eyes, the lower MAEs were provided by Haigis and EVO (p < .01 except Hoffer Q, SRK/T and Holladay I). Conclusions Barrett Universal II was the most accurate for IOL power calculation in the normal and long eyes. For short eyes, the formulas Haigis and EVO seem best at predicting refractive outcomes.