薬理と治療

Abstract

In clinical and epidemiological studies, poor conducts of statistical analyses produce biases in the results and sometimes cause fails of reproducibility. The current standards of statistical analyses in clinical studies have been established to prevent these biases and to maximize scientific validity and precision. In this article, we provide a gentle introduction to these biostatistical methodology for understanding research papers published in medical journals. In particular, we focus what is “confounding,” a typical bias in clinical studies and how it influences to results and interpretations of clinical researches. In addition, we explain how we can prevent or adjust the bias in designing and analyzing of clinical studies. Also, we review the development of the STROBE statement that is a guideline to improve reports of observational studies to gaining scientific validity and reproducibility.