一种临床试验中的适应性样本量调整方法  

A Method of Sample Size Re-estimation During Adaptive Clinical Trials

作　　者：谢之辉[1] 胡芳[2] 任建玲[1] 贺江南[1] 贺佳[1] 

机构地区：[1]第二军医大学卫勤系卫生统计学教研室,200433 [2]上海市普陀区疾病预防控制中心,200333

出　　处：《中国卫生统计》2010年第1期2-6,共5页Chinese Journal of Health Statistics

基　　金：国家十一五"重大新药创制科技重大专项"(2008ZX09312-007和2008ZX09312-025)

摘　　要：目的介绍一种临床试验中的适应性样本量调整方法,并探讨样本量调整后统计分析方法的第Ⅰ类错误率及检验效能。方法通过montecarlo模拟的方法研究n1大小对最终样本量Nf的影响,并估计最终方差偏移大小;同时模拟研究样本量调整后统计分析方法的第Ⅰ类错误率及检验效能大小。结果(1)模拟结果显示运用该样本量调整方法所得到的最终样本量Nf非常接近其真实值N0,尤其在π=0.4时进行样本量调整。(2)同时模拟结果显示所介绍的样本量调整后的校正t检验方法不仅能有效控制第Ⅰ类错误率α并且能充分满足试验检验效能(1-β)。结论该样本量调整方法研究结果是在一般两样本单侧t检验条件下得到也可应用于优效或非劣效设计的临床试验中。Objective Introducing a method of sample size re-estimation in adaptive clinical trials,and studying the type I error rate and the power of statistical mehod after the sample size adjustment.Methods We studied the impact of magnitude of n1 to final sample size,assessing the bias of finally calculated variance to true variance;meanwhile studying the type I error rate and the power of the adjustment statistical mehod by monte carlo simulations.Results（1）The simulation results showed that the final sample size was very near to the true size,especially when the adjustment was made at π=0.4.（2）Meanwhile our results indicated that the bias in the type I error rate was negligible and the desired power was maintained with the introduced corrected t-test after the sample size adjustment.Conclusion All simulation results are got under the two-sample t-test with one-side and it can be applied in non-inferiority or superiority clinical trials.

关 键 词：样本量重估 内部预试验 适应性临床试验 

分 类 号：R195[医药卫生—卫生统计学]