临床试验中评价处理效应的协变量调整问题

Statistical adjustment of treatment effect for covariates in clinical trials

临床试验设计和分析中常常包括影响疾病结果的非处理因素 ,这些因素被统称为协变量。对处理效应的评价 ,若不合理考虑协变量的影响 ,常常会导致结论的偏倚乃至错误。因此 ,在临床试验中要审慎地对待这一问题 ,在评价处理效应时应该考虑用协变量对其进行适当的调整 ,以达到处理效应的正确评价。试验设计阶段的协变量调整称为“事先调整” ,常用的方法有分层法、动态分配最小化法等。试验分析阶段的协变量调整称为“事后调整”,常用的方法有分层检验法、回归模型法等。通过协变量调整 ,一方面阐明病人特征在处理组间的不均衡并没有影响到处理效应的正确评价 ,提高临床试验结果的可信度 ;另外 ,协变量调整方法的合理运用还可提高统计效能。本文对处理效应评价协变量调整涉及的概念、方法、步骤等进行介绍和探讨 ,尤其强调了应用中的一些统计学事项。

Some non treatment variables that affect the outcome of a disease are often called covariates. These covariables should be considered in the design and analysis of clinical trials to obtain unbias conclusion. To ensure that any observed treatment effect is not influenced by an imbalances in baseline characteristics, both preadjustment and postadjustment are provided in the design stage and analysis stage of the trials respectively. They can improve the credibility of the trial results and increase the statistical efficiency. Based on a few papers published about adjustment for covariates and some documentations of the International Conference on Harmonization (ICH), we review the concepts, methods and procedures for adjustment of treatment effects for the influence of covariates. The statistical issues on the application of adjustment are especially discussed in great depth.