Вопросы статистического анализа в российских офтальмологических журналах

Abstract

Purpose. To rate the quality of the statistical analysis in the leading Russian journals in ophthalmology. Material and methods. The study was performed of the quality of the statistical analysis and the formation of research groups in 90 original articles published in 2015 in leading Russian ophthalmology journals. Twenty-three articles (case reports, surgical techniques, etc.) did not contain statistical data and were excluded. As a result, 67 articles were studied. Results. Appropriate statistical analysis without serious mistakes was found in 14 publications (21%) but only five of them (7%) did not contain any shortcomings. The most common errors were: inclusion in the analysis of both eyes of patients without correction for the correlation of the parameters of the fellow eyes (30%), comparing of multiple groups by incorrect methods (28%; 82% of the articles requiring this comparison), comparing proportions without the use of proper methods (24%), absence (25%) or a clear deficiency (18%) of the statistical analysis section, lack of information on the format of the data used M±SD or M±m (16%), lack of guidance on the use or incorrect use of unpaired or paired Student’s t test (15%). Concerning the formation of research groups the most common errors were: sample size calculation was not present (all articles), sampling order was not defined (82%), exclusion criteria were absent (64%). Conclusion. The analysis of original articles in leading Russian ophthalmology journals in 2015 indicates the need for a significant increase in quality requirements for statistical data processing and the formation of research groups. Based on this analysis, it was proposed to amend the rules for authors in ophthalmology journals.