Ioannidis John P A
Department of Hygiene and Epidemiology, University of Ioannina School of Medicine, Ioannina, Greece 45110.
Bull NYU Hosp Jt Dis. 2008;66(2):135-9.
Randomized trials are an excellent research design with major advantages. However, randomized trials are not immune to biases, and inferences from them may be sometimes flawed or irrelevant. The present review addresses, in brief, some of the major threats to the credibility and relevance of the results of clinical trials: power problems, biases affecting internal validity (poor design, conduct, and analysis), biases affecting the total randomized evidence on a specific topic (publication bias and selective outcome and analysis reporting bias), lack of relevance, poor generalizability, and biases in the interpretation of the results.
随机试验是一种具有诸多主要优势的出色研究设计。然而，随机试验并非不会受到偏倚影响，基于它们得出的推论有时可能存在缺陷或不相关。本综述简要阐述了对临床试验结果的可信度和相关性构成的一些主要威胁：效能问题、影响内部效度的偏倚（设计、实施和分析不佳）、影响特定主题全部随机证据的偏倚（发表偏倚以及选择性结果和分析报告偏倚）、缺乏相关性、可推广性差以及结果解释中的偏倚。
