New method to assess the accuracy of intraocular lens power calculation formulas according to ocular biometric parameters.

To develop a new method that makes it easy to detect accuracy deficiencies of any intraocular lens (IOL) power calculation formulas and to test it on 9 different formulas. IOA, Madrid, Spain. Retrospective observational case series. This study's first stage included 3519 eyes from 3519 candidates for cataract surgery for which frequency distributions for the following biometric eye parameters were computed: axial length (AL), anterior chamber depth (ACD), lens thickness (LT), white-to-white (WTW), and mean corneal radius (Rm). The resulting data for each parameter were fifth, 25th, 75th and 95th percentiles, which allowed definition of the corresponding normality range. In a second stage, the new graphic-representation method was tested for 9 different formulas in a sample of 70 eyes undergoing cataract surgery with multifocal IOL (mIOL) implantation. Normality ranges (defined as the 25th to 75th percentile interval) were 22.84 to 24.42 mm for AL, 2.86 to 3.39 mm for ACD, 4.36 to 4.88 mm for LT, 11.64 to 12.19 mm for WTW, and 7.52 to 7.87 mm for Rm, with lower sizes in women. No significant differences were found among the 9 formulas for percentage of eyes in ±0.50 diopters (D) (P = .82) or ±1.00 D (P = .97). The graphical method showed less accuracy in ±0.50 D for ACDs from 2.46 to 2.85 mm (5th to 25th percentile) for several formulas (P < .05). Nine formulas showed nonsignificant differences in the general predictability for a sample of eyes that were candidates to mIOL implantation. Predictability in this sample decreased for short ACDs.