违背随机分组方案时的临床试验统计分析方法研究

Exploring Statistical Methods in Clinical Trials When Some of the Patients are Treated against the Randomization

目的在临床试验出现随机分组被部分破坏的情况下,探索适合的统计分析方法。方法利用随机模拟的方法考察三种分析方法的第一类错误发生概率是否小于α;比较第一类错误发生概率不大于α的分析方法的功效。结果对违背随机分组方案的情况,按客观分组的t检验是的第一类错误发生概率远大于α;ITT方法和协方差分析方法的第一类错误发生概率的95%置信区间包含α;协方差分析方法的第二类错误发生概率显著小于ITT方法的第二类错误发生概率。结论对违背随机分组方案的情况,按客观分组的t检验是不合适的;ITT方法和协方差分析方法能较好地控制第一类错误的概率在α附近;协方差分析方法的功效高于ITT方法的功效。

Objective To explore adequate methods for statistical analysis when some of the patients are treated against the randomization.Methods Simulation studies were conducted to see if the probability of the TYPE I error for each method is not greater than α and to compare the power of the methods with probability of TYPE I error not greater than α.Results When some of the patients are treated against the randomization,the probability of the TYPE I error for the t test based on the treatment eventually administered is far greater than α,while the probability of the TYPE I error for ITT t test and the ANCOVA based on the treatment eventually administered has a 95% confidence interval containing α.Conclusion When some of the patients are treated against the randomization,the t test based on the treatment eventually administered is not an appropriate method,while the ITT t test and the ANCOVA based on the treatment eventually administered are able to control the probability of TYPE I error around α;the ANCOVA based on the treatment eventually administered has a higher power than the ITT t test.