[Reliability of biomedical research]

Abstract

In the last century, use of statistics became widespread in biomedical research, making medicine more scientific. However, the misuse of statistical methods may have a negative impact on the reproducibility of experiments, which instead should be at the core of scientific method. The consequence is the so called "replicability crisis", which can affect the reliability of scientific results: if one cannot replicate a published finding, she would hardly trust it. In this paper, firstly, we will provide a short review of the replicability crisis. Indeed, this issue has recently drawn the attention of the entire scientific community. Secondly, we will try to figure out some potential explanation for this phenomenon. Finally, we will show how this problem deeply affects also the reliability of randomized clinical trial, which is fundamental for the medical and regulatory decision-making.