Biostatistics pitfalls: Lessons learned from analysis of medical data

Abstract

In several recent issues of The Lancet, we identified a few common pitfalls in the analysis of clinical trial and medical data in the published articles (Mok et al., 2019; Herrlinger et al., 2019; Reindl-Schwaighofer et al., 2019; He et al., 2019). Without careful validation of model assumptions, even the primary endpoint of the trial might be analyzed using improper statistical methods. We carried out an in-depth analysis of the statistical issues in four real clinical trials, which highlights the importance of statistics in the medical field. With every effort, biostatisticians need to work with clinicians closely to take the most appropriate statistical approaches to data analysis; otherwise the conclusions drawn from the data might be problematic or misleading.