ROLE OF PROGNOSTIC STRATIFICATION IN PREVENTING BIAS PERMITTED BY RANDOM ALLOCATION OF TREATMENT
ROLE OF PROGNOSTIC STRATIFICATION IN PREVENTING BIAS PERMITTED BY RANDOM ALLOCATION OF TREATMENT
复制标题
DOI：
10.1016/0021-9681(76)90080-1
复制
发表时间：
1976-01-01
期刊：
影响因子：
--
通讯作者：
LANDIS, JR
中科院分区：
文献类型：
--
作者：
FEINSTEIN, AR;LANDIS, JR
在治疗试验中使用随机化是为了防止比较治疗分配中的偏差，并允许在结果的统计评估中使用概率的数学理论。随机化时间表已准备好，治疗之间有或没有数值平衡，希望机会能够提供预后特征的均等分布。进行模拟以确定这种希望在非分层随机化程序中徒劳的频率。注意到机会特别有可能产生不平衡结果的情况，并确定了通过分层内的随机化可以在多大程度上防止不平衡。分层随机化在降低治疗试验中出现虚假显着性结果的风险方面的价值已得到实证证明。
Randomization is used in a therapeutic trial to prevent bias in the assignment of the compared t reatments and to allow mathematical theories of probability to be used in statistical appraisals of the results. The randomization schedule is prepared, with or without numerical numerical balancing among treatments, and it is hoped that chance will provide an equal distribution of prognostic features. Simulations were performed to determine the frequency with which this hope would be futile in an unstratified randomization procedure. Circumstances in which chance was particularly likely to produce unbalanced results were noted, and how much of the imbalance could be prevented by randomization within strata was determined. The value of stratified randomization in reducing the hazard of spuriously significant results in a therapeutic trial was empirically demonstrated.