Statistics in scientific articles published in the European Annals of Otorhinolaryngology Head & Neck Diseases.

ABSTRACT

OBJECTIVES: To evaluate biostatistics in scientific articles published in the European Annals of Otorhinolaryngology Head & Neck Diseases. MATERIALS AND METHODS: Retrospective analysis of 76 scientific articles published in 2018 and 2019. The main goal was to describe the percentage of articles using biostatistics and to explore for potential associations with the article's main characteristics. Secondary goals were, within the group of articles using statistical tests, to describe the type of statistics used, the assessment of normality in case of comparison of mean values, the p-value threshold for significance, the use of confidence intervals, and power analysis. RESULTS: Statistics were used in 73.7% of articles, without any significant association with main characteristics. Within the group of articles using statistics, the tests used were parametric, non-parametric and not specified in 77.7%, 51.4% and 1.8% of cases, respectively. Normality was checked in 14.2% of article using parametric tests to evaluate mean values. The p-value significance threshold was set at .05, .01, .005 levels and not defined in 60.7%, 1.8%, 1.8% and 35.7% of articles, respectively, while confidence intervals and power analyses were documented in 10.7% and 5.3%, respectively. CONCLUSION: This article underlines the need for better use of statistics in the European Annals of Otorhinolaryngology Head & Neck Diseases, to improve the quality of scientific articles published in the organ of written expression of the French and International French-speaking Societies of Otorhinolaryngology, and to support the ongoing move toward better medical science.