Bias That Should Be Avoided to Obtain a Reliable Study of IOL Power Calculation After Myopic Refractive Surgery

Abstract

We read with great interest the article by Lanza et al 1 concerning intraocular lens (IOL) power calculation after myopic refractive surgery.We congratulate the authors for their study; however, we would like to comment on some points that in our opinion need to be clarified.First, the authors only analyzed the methods that did not require special devices to evaluate corneal parameters except for optical biometer, because other formulas could be not accessible to all clinicians.We agree with them, but we wonder why they did not consider the Advanced Lens Measurement Approach method. 2 This method requires only the knowledge of mean keratometry and axial length.The authors should have known this method because it is the evolution of a formula that was also described Rosa et al. 3 Second, we would also comment on the use of the American Society of Cataract and Refractive Surgery (ASCRS) Online Post-Refractive IOL Calculator to perform calculation with some formulas.We wonder how the authors calculated the prediction error (PE) for the Barrett True-K, Haigis-L, and Shammas formulas.With the ASCRS calculator, only the differences between ASCRS suggested IOL and implanted IOL powers, corresponding to IOL-PEs, can be obtained.A limitation of using IOL-PEs is the need to convert them to PEs by using an estimate. 4Moreover, according to most updated protocols regarding IOL power calculation accuracy studies, 5 converting IOL-PEs to PEs using a constant factor over the entire axial length range is an error. 5 Third, Lanza et al divided the population in subgroups according to different ranges of axial length and mean keratometry, but they did not explain how they chose cut-off values for these parameters.There are no references that could explain this choice.Fourth, the authors declared that they followed the criteria of Wang et al 4 to perform statistical analysis, but Wang et al suggest using the Cochrane-Q test when comparing the percentage of eyes within 0.50 and 1.00 diopters of PE of more than two formulas; instead the authors applied the McNemar test.In addition, Wang et al 4 recommend defining sample size by calculation and checking whether mean errors were significantly different from zero: Lanza et al did not perform such evaluations.Consequently, they did not seem to follow Wang et al. 4 At the end, the main problem of this study was why the authors did not perform lens constant optimization through zeroing-out mean errors of PEs.They used the

Topics & Concepts