Performance of formulas included in the ESCRS intraocular lens power calculator

Purpose: To compare the refractive prediction errors (PEs) of formulas included in the ESCRS intraocular lens (IOL) power calculator to aid in informed decisions on IOL power selection based on the output of this tool. Setting: Cullen Eye Institute, Baylor College of Medicine, Houston, Texas. Design: Retrospective case-series. Methods: 748 eyes of 748 patients after implantation of 1 of 3 lenses were included, single-piece: the SN60WF, PCB00/ZCB00 and 3-piece: MA60MA. IOL constants recommended by the calculator were used for the study. We performed analysis for the whole dataset, short (<22 mm) and long eyes (>25 mm) as well as in subgroups based on the type of the implanted IOL. SD and root mean square absolute error (RMSAE) were selected as the primary endpoints. Results: Cooke K6 had the lowest SD of PEs in the whole dataset ( P < .05) when compared with Barrett, EVO, and Hoffer QST. In the subgroup of long eyes, the Kane formula had the lowest RMSAE ( P < .05) when compared with Barrett and EVO. No significant differences in primary endpoints for implantation of the 3 types of IOLs were found. However, the median absolute error after implantation of the MA60MA was significantly higher for Hoffer QST than for all other formulas except for Pearl-DGS. Conclusions: Significant differences in the performance of formulas included in the calculator were found. In the whole dataset, Cooke K6 had the lowest SD of PEs among the analyzed formulas.