INTRAOCULAR LENS POWER CALCULATION IN SHORT EYES

To comparatively evaluate the accuracy of newer intraocular lens (IOL) calculation formulas (Barrett Universal II, Kane and Hill-RBF 3.0) and common third-generation formulas with and without using a novel axial length (AL) adjustment in predicting refractive outcomes in eyes with short AL. 

Centro Hospitalar e Universitário de Coimbra, Portugal

Retrospective study including eyes with AL less than 22.0 mm submitted to uneventful cataract surgery and implantation of an AcrySof SN60AT IOL. All patients underwent optical biometry (Carl Zeiss IOLMaster 700) and the post-operative spherical equivalent for the same implanted IOL was estimated using SRK/T, Holladay 1, Hoffer Q, Haigis, Barrett Universal II, Kane and Hill-RBF 3.0 formulas. The Cooke-modified axial length (CMAL) method was used in the SRK/T, Holladay 1 and Hoffer Q formulas. Analysis was performed before and after lens constants optimization.  Outcomes included the mean (ME) and median (MedE) prediction error, the mean absolute (MAE) and median absolute prediction error (MedAE) and the proportion of eyes within 0.50, 0.75 and 1.00 diopters (D) of the pre-operative prediction.

Sixty-four eyes with a mean axial-length of 21.54 ± 0.57 mm were included. Without adjustment the Hoffer Q was the only formula with a slightly myopic refractive prediction error –0.157D ± 0.60 and Hill-RBF 3.0 had the lowest standard deviation in the prediction error 0.031D ± 0.58. After optimization the mean absolute error in ascending order was Kane 0.43D, Hill-RBF 3.0 0.43D, Barrett 0.44D, Hoffer Q 0.45D, Haigis 0.45D, Holladay 1 0.48 and SRK/T 0.53D. The Kane formula, with the lowest MAE, yielded a prediction error within 0.50D, 0.75D and 1D in 71.9%, 84.4% and 90.6% of cases, respectively. Using CMAL did not improve predictions. The use of optional variables in the Kane (LT and CCT) and Barrett Universal II (LT and WTW) formulas changed the prediction error >0.1D in less than 30% of cases and most without further improvement.

Recent formulas like the Barrett Universal II, Kane, Hill-RBF v3.0 perform well, particularly after constant optimization. Without optimization the Hoffer Q is the only one with myopic prediction error which might explain its popularity in this subset of patients. The CMAL adjustment, originally developed for another optical biometer (OLCR device) did not improve outcomes. Also, the use of optional variables in the Kane and Barrett Universal II formulas did not further enhance predictions.