The role and significance of sensitivity analyses in enhancing the statistical validity of clinical studies

The Journal of Thoracic and Cardiovascular Surgery

article

Michael Baiocchi¹  Y. Joseph Woo²  Peter Chiu²  Andrew B. Goldstone² 

[1] Department of Epidemiology and Population Health, Stanford University;Department of Cardiothoracic Surgery, Stanford University

关键词: bias;    causality;    confounding;    matching;    propensity scores;   

DOI  :  10.1016/j.jtcvs.2020.09.134

学科分类：心脏病和心血管学

来源: Mosby, Inc.

【 摘 要 】

As surgeons, we operate with the intent of causing a patient's health to improve. In statistical jargon, we are interested in the causal effect of one treatment in contrast to another, for example, for a mitral valve in need of replacement, will a patient survive longer with a mechanical or bioprosthetic valve?1 The surest way to generate useful data to answer a causal question is to run a randomized controlled trial (RCT). There are many nuances, but it is largely true that RCTs produce strong causal conclusions because the necessary assumptions are true by design; that is, the statistical assumptions needed to obtain reliable causal estimates are made true by researchers randomizing people into treatment A or treatment B.

【 授权许可】

Unknown