单组目标值临床试验多指标时的样本量估计

Sample size calculation for single-arm OPC trials with multi endpoints

目的:针对多个以率为共同终点评价指标的单组临床试验,探讨控制全局把握度的样本量估计方法。方法:根据单组目标值法单指标时的样本量估计方法,在理论上阐明单组临床试验多个共同指标时样本量估计的原理,提出控制全局把握度的样本量估计方法。事先设定各单指标目标值、靶值、Ⅰ类错误率及总的Ⅱ类错误率水平,在给定的样本量下可算得各单指标对应的把握度,相乘后即为全局把握度,利用逐步寻值即可获得达到全局把握度时所需的样本量。结果:结合一个单组临床试验案例,基于4个率指标,利用样本量估计的正态近似法和确切概率法,分析对比了传统直接校正把握度的方法和本文逐步寻值法所估计的样本量结果,后者能更准确地控制全局把握度,所需的样本量更少。结论:本文给出的单组目标值临床试验多指标情形下逐步寻值的样本量估计方法,能较好地控制全局把握度、节约样本量,具有很强的实用价值。

AIM：To explore the sample size calculation by control global power for multiple rates as multi endpoints in single-arm Objective Performance Criteria（OPC） trials.METHODS：Based on the sample size calculations for single-arm OPC with single endpoint,we illustrated the statistical principle of sample size calculations with multi endpoints in theory and proposed a method by control global power.According to the pre-determined level of OPC,target value,type Ⅰ error and total type Ⅱ error,the power of each endpoint could be evaluated under the given sample sizes and the multiplication of all powers was the actual global power.The sample size reached objective global power was evaluated by the stepwise search method.RESULTS：We compared the sample sizes using the conventional direct power correction and stepwise search method through an example of one-sample clinical trial with four proportions based on normal approximation method and exact method.The results showed that the latter could more accurately control the global power and reduced the sample sizes.CONCLUSION：We proposed a solution to correct sample sizes for single-arm OPC trials with multi endpoints,which could control global power and reduce sample sizes significantly.