Statistics usage in the American Journal of Obstetrics and Gynecology : Things have changed

Abstract

Appropriate use of statistical analysis and a clear description of the statistical methods applied are essential elements of a scientific paper. In 1996, Welch and Gabbe reviewed the statistical usage of all papers included in the Clinical Articles section and all articles published as transactions of societies in the Journal from January through June 1994.1Welch GE Gabbe SG. Review of statistics usage in the American Journal of Obstetrics and Gynecology.Am J Obstet Gynecol. 1996; 175: 1138-1141Abstract Full Text Full Text PDF PubMed Scopus (50) Google Scholar That study revealed inappropriate usage of statistics in nearly one third of the published papers, and in more than half of these the statistical methods were judged to be seriously flawed. Furthermore, fewer than half the papers completely cataloged the statistical methods used to analyze the data. In an effort to improve statistical usage in the Journal, the Editors wrote new Instructions to Authors requiring a complete listing of the statistical procedures used, developed a checklist to help reviewers and Editors assess the appropriateness of statistical analysis, sent out a statistical education tool to help peer reviewers identify common errors, and planned to expand the evaluation of manuscripts by statistical consultants. This issue of the Journal includes a follow-up study by Welch and Gabbe that asks the question, “Has anything changed?”2Welch GE Gabbe SG. Statistics usage in the American Journal of Obstetrics and Gynecology: has anything changed?.Am J Obstet Gynecol. 2002; 186: 584-586Abstract Full Text Full Text PDF PubMed Scopus (21) Google Scholar Examination of the data provides a clear answer. Yes, things have changed and for the better! This review of papers published in the general obstetrics and gynecology section and the transactions of societies section during the first 6 months of 1999 reveals that nearly 75% of manuscripts have completely cataloged the statistical methods used, that there has been a significant improvement in the appropriateness of statistical usage, and that inappropriate use of statistics in published papers has fallen below 10%. What is responsible for this significant improvement? In addition to the efforts of our reviewers and Editors, we agree with Welch and Gabbe that it is our contributors who deserve most of the credit for providing a more complete listing of the statistical methods used and for using the most appropriate tests for their analyses.