Reply: Biostatistics and Ophthalmology: The Case of Two Eyes, What Is Correct and What Is Customary
Year: 2024
Type: letter
Abstract: We read with interest the article by Waring et al, titled "Safety and Effectiveness of Laser in Situ Keratomileusis Using the Teneo 317 Model 2 for Correcting Myopia and Myopic Astigmatism: Results of the U.S. FDA Clinical Trial." 1 For the statistical analysis, the authors treated the eyes as independent sampling units, ignoring the correlation between the two eyes of each patient despite the well-known issue of interocular symmetry, the correlation of features between the two eyes of the same individual.2 In standard statistical methods, sample independence is a key assumption, and violating this principle can be comparable to doubling the sample size, akin to data fabrication.The consequences of disregarding interocular symmetry in statistical analyses may include artificially narrowed confidence intervals and misleading P value significance.3,4 Given the significance of this study and its publication in the esteemed Journal of Refractive Surgery, it was expected that the distinguished team of authors would adhere strictly to scientific methods, avoiding customary yet inaccurate approaches found in other publications.To account for interocular symmetry in statistical analyses, the researchers might choose to include only one eye per participant randomly or apply more advanced statistical techniques to address the sample dependence.5,
Source: Journal of Refractive Surgery
Author: George O. Waring
Language: English
Subfield: Statistics and Probability
Field: Mathematics
Domain: Physical Sciences
Open Access status: bronze
OpenAlex ID: W4404299383