RESUMO
In recent decades, evidence-based medicine acquired special importance in medicine. Therefore, proper presentation of data obtained in scientific research is extremely important. The statistical data processing, being an integral part of this process, often causes difficulties for researchers and its incorrect application results in distortion of results obtained. The purpose of the study is to comparatively analyze programs and methods of statistical data processing applied in dissertations on obstetrics and gynecology in 2011-2021, to examine trends in choosing them depending on specificity of research issue and to identify shortcomings erred by authors in choosing or describing data processing methods. The sampling for analysis included 258 abstracts of candidate's dissertations in the specialty "obstetrics and gynecology", defended in 2011-2021. The analysis covered the programs and methods of mathematical data processing. Over the past decade, significant complication of statistical processing of results of clinical trials in obstetrics and gynecology occurred in part of methods applied. The application of binary logistic regression and discriminant analysis increased most significantly over the past decade. Such sophisticated methods of statistical data processing as factor analysis, decision trees, ordinal logistic regression and neural networks began to be used too. The trend of gradual replacement of parametric methods (Student's t-test, one-way analysis of variance) by such corresponding non-parametric methods as Mann-Whitney test, Kruskal-Wallis test. The Microsoft Excel and Statistica were used most often for data processing. In recent years, the software SPSS Statistics is actively applied. However, problems in describing statistical methods used in dissertations continue to be present. In significant part of dissertations information about statistical program applied, methods of assessing of quantitative data distribution and criteria of significance of obtained results is absent. The proper application of statistical programs, methods of information processing, adequate interpretation of results as well as provision of complete information about methodological support are the key points to carry out modern research resulting in trusted attitude to scientific work and its results.
