Abstract: Real-world studies have increasely become the focus in the field of clinical research because their findings are closer to the real world. However, real-world study is not a specific type of research design parallel with observational studies and experimental studies, and the control of confounding factors and covariates in the data analysis of real-world studies should be paid more attention. Correct identification of confounding factors/covariates and reasonable application of statistical methods can effectively improve the accuracy and authenticity of research conclusions. This article introduced the common methods of confounding factors/covariates control in real-world studies, including hierarchical analysis, analysis of covariance, multi-factor regression, propensity score, disease risk score and instrumental variable analysis, and cases analysis combined with clinical research application of cerebrovascular disease, so as to provide reference for understanding and applying the relevant methods.

Key words: Real-world study; Control of confounding factor; Cerebrovascular disease; Covariate

摘要： 真实世界研究由于其研究结果更贴近真实世界，在临床研究领域逐渐受到关注和重视。真实世界研究不是并列于观察性研究和试验性研究的某种特定的研究设计类型，在真实世界研究的数据分析中更需要关注混杂因素和协变量的控制。正确识别混杂因素和协变量以及合理应用统计方法对其进行控制，可有效提高研究结论的准确性和真实性。本文介绍了真实世界研究中混杂因素和协变量控制的常见方法，包括分层分析、协方差分析、多因素回归、倾向性评分、疾病风险评分和工具变量分析等，并结合脑血管病的临床研究应用实例展开分析，以期为研究者理解与应用相关方法提供参考。

关键词: 真实世界研究; 混杂因素控制; 脑血管病; 协变量