We read with interest the paper by Jeon and associates dealing with the refractive stability of toric intraocular lenses (IOLs). We were surprised to see that IOLs with higher cylinder (ie, the T4 and T5 Acrysof) led to a more myopic outcome after cataract surgery, and could not come up with any explanation for such a result. Before claiming that a refined formula is required to select IOL power in cases of high astigmatism, we feel that the data of this study should be carefully reanalyzed for several reasons.
First, rather than using the SRK/T formula for all eyes, the Hoffer Q and the Holladay 1 formulas should be adopted. The axial length, in fact, was between 22 and 25 mm, and the latter formulas have been proven to be superior to the SRK/T in short and medium eyes. On the contrary, the SRK/T formula is the best for long eyes (>26 mm), which were not enrolled in this study. Constant optimization should also be carried out in order to achieve a zero mean arithmetic error between the predicted and the measured postoperative refraction. Optimized constants should be reported in the manuscript. Once constant optimization is carried out, the authors should report the median absolute error (MedAE) in refraction prediction. Comparing these values by analysis of variance (ANOVA) would then allow us to really understand if IOLs with higher toricity lead to a myopic outcome.
By following this approach, we did not find any difference in a sample of consecutive eyes operated by 2 of us (G.S. and P.B.) with the same technique. Using the Hoffer Q and Holladay 1 formulas, we found that the MedAE was, respectively, 0.27 and 0.33 diopter (D) in a sample of 21 consecutive eyes (mean axial length: 23.88 ± 0.86 mm) with the aspheric non-toric AcrySof SN60WF. In a sample of 19 eyes (mean axial length: 23.26 ± 1.37 mm) with the AcrySof T3 toric IOL, the MedAE was, respectively, 0.28 and 0.17 D. In a sample of 23 eyes (mean axial length: 23.33 ± 1.01 mm) that received the AcrySof T4, T5, T6, T7, T8, and T9, the MedAE was 0.32 D for the Hoffer Q and 0.29 D for the Holladay 1 formula. One-way ANOVA did not disclose any statistically significant difference among the 3 groups of IOLs with both Hoffer Q and Holladay 1 formulas.
Second, the authors should report the mean axial length for each subgroup of eyes. Was there any difference? Moreover, was there any difference in the mean corneal power? These differences may be related to the refractive outcome.
Third, using the Alcon online calculator may lead to some inaccuracy since it does not take into consideration the influence of anterior chamber depth on the conversion of the toricity from the IOL to the corneal plane.
Finally, stating that there was no difference between manual keratometry and other techniques in astigmatism assessment without performing any kind of vectorial analysis seems quite unscientific.