We read with great interest the article by Garcia-Gonzalez and associates and compliment the authors for their study, which tries to provide additional data regarding laser in situ keratomileusis-induced monovision. The authors provided details of visual acuity (VA) outcomes. As well known, in laser in situ keratomileusis and other refractive surgical procedures, VA is the main postoperative measure. We believe any author should be careful in reporting VA outcomes, because this is the parameter that most readers will value the highest. Therefore, we cannot avoid raising a point of criticism regarding the way the visual outcomes were reported in this article.
First, it was mentioned that the investigators used the logarithmic Early Treatment Diabetic Retinopathy chart 2000 for their VA tests. This chart does not contain a 0.9-decimal VA line, and we wonder how it was possible for the authors to form a 0.9 decimal VA group for the results presented in Figure 1.
Second, the authors used the logarithm of the minimal angle of resolution (logMAR) values of all visual acuities for the statistical analysis and then converted them into the decimal quotation throughout the text, reporting mean and standard deviation (SD) of VA results. Most researchers have encountered problems with converting logMAR values into Snellen values. Although it is easy to convert a mean logMAR results into a decimal value, the same is not possible for SD. Data points that show a linear correlation in a system with logarithmic scale, as used for logMAR VA and the calculation of mean and SD, show a logarithmic distribution in a system with linear scale, not allowing the use of SD in decimal scale. Therefore, it is most likely that the presented SDs in the text are incorrect. We would challenge the authors to reconsider their calculations and conversions of VA values.
Third, there are well-accepted ways to report VA values (Snellen/decimal, logMAR). Generally, authors should be encouraged to avoid presenting a mix of these scales in the same report, as seen in Figures 1 and 2 of this article. For statistical reasons, and for correct reporting of mean and SD, we suggest that every author use logMAR values. This also would bring the reports in line with published standards.