The authors conducted this study to determine which of the three third-generation IOL formulas is best depending upon the axial length (AL) when using optimized IOL formula constants and the IOLMaster biometer (Carl Zeiss Meditec AG) for AL and K measurements. Using optimized Hoffer Q, Holladay 1 and SRK/T formulas, they retrospectively calculated hypothetical prediction errors using data from electronic medical records for patients implanted with one of two different IOLs. They found that the Hoffer Q performed best for ALs from 20.00 to 20.99 mm, the Hoffer Q and Holladay 1 for ALs from 21.00 to 21.49 mm and the SRK/T for ALs 27.00 mm or longer.
To conduct the study, the authors used prospectively collected data from 8,108 eyes that underwent phacoemulsification and implantation with the Sofport AO or Akreos Fit IOL. They believe this is the largest reported series of retrospectively calculated refractive outcomes after cataract surgery using the three formulas with optimized IOL formula constants and the IOLMaster device for biometry.
Their calculations determined that in short eyes, the Hoffer Q had the lowest mean absolute error (MAE) for ALs from 20.00 to 20.99 mm. The Hoffer Q and Holladay 1 had a lower MAE than the SRK/T for ALs from 21.00 to 21.49 mm. They found no statistically significant differences in MAE for ALs from 21.50 to 21.99mm. In medium eyes, they found no statistically significant differences in MAE for any IOL formula for ALs from 22.00 to 23.49 mm but a trend toward lower MAEs for the Holladay 1 for ALs from 23.50 to 25.99 mm. They determined that in long eyes, the SRK/T had the lowest MAE, with statistically significant differences for ALs of 27.00 mm or longer.
They say that the differences in refractive outcomes measured using the MAE and the percentages of eyes within target refraction margins were, with the exception of one subgroup, due to differences in the mean error (P < 0.00001) and not differences in the standard deviation between formulas in specific AL groups. They say this suggests that if the IOL constant varied with respect to the AL, the formulas may have improved refractive outcomes in these groups.
The authors conclude that use of optimized constants can produce refractive outcomes in the range of 40 percent within 0.25 D, 75 percent within 0.50 D and 95 percent within 1.00 D. They say that the study's findings support previous recommendations for the use of IOL formulas and further define the groups of eyes for which the effect of IOL formula choice is particularly significant.
While this paper was a joy to see, we wish they had looked at the Haigis and Holladay 2 formulas, which are both known to be highly reliable across a wide range of ALs and ocular geometry.
We'd also like to see future software revisions for partial coherence interferometry biometry include "smart" output options customized to the measured axial lengths. The IOLMaster currently spits out a full page of lens choices based on a single formula, regardless of AL, but it might be more helpful if it included on one page several formulas known to provide the highest accuracy given the biometry. This would reduce errors from physicians choosing lenses using suboptimal formulas. Naturally, savvy surgeons would want to customize the parameters, and the software should be flexible enough to permit this.