提高医疗数据库研究可重复性:研究设计可视化流程图

Ways to Improve Reproducibility of Healthcare Database Studies: Graphical Depiction of Longitudinal Study Designs

近些年来,相较于评估药物有效性和安全性的金标准——随机对照试验(RCT)的结果,基于大型医疗数据库(如行政索赔数据库和电子健康记录)开展研究获得的证据受到越来越多医疗决策者的关注。由于数据库研究的多样性和复杂性,研究者常常对相同概念采用不同的术语表达。对于观察性研究而言,对研究报告中术语和关键参数等基本内容进行统一规范,将会大大提升研究的透明度和可重复性,及对结果的有效性评估。基于此,哈佛医学院的Sebastian Schneeweiss博士带领的研究团队对于医疗数据库研究提出了研究设计可视化流程图展示框架。本文首先对医疗数据库研究的几个重要概念进行了简单归纳,着重介绍了这个框架的基本构成和含义,并给出几个经典研究设计的案例以供读者加深理解。

In recent years, evidence generated from large health care databases, such as administrative claims and electronic health records, have received increasing attention from medical decision makers, compared with randomized controlled trials(RCTS), which is termed as the gold standard to assess the efficacy and safety of drugs. Due to the diversity and complexity of database studies, researchers often use different terminologies for the same concept. For observational studies, the standardization of the most basic contents such as terminology and key parameters in the study report will greatly improve the transparency and reproducibility of study and the evaluation of effectiveness of study results. Based on that, a research team led by Dr Sebastian Schneeweiss from Harvard Medical School proposed a framework of graphical representation that visualizes study design implementations of health care database. This paper began with a brief introduction to some key concepts of health care database study, focusing on the basic structure and meaning of the framework, and gave some examples of classic study designs for readers for further understanding.