多终点变量对药物疗效评价的影响

Effect of multiple endpoints on the efficacy assessment in clinical trials

目的:探讨在新药临床试验中多终点变量对药物疗效评价的影响。方法:通过Monte Carlo模拟探索在不同疗效分布类型中多终点变量间的相关性及其对疗效客观评价的影响,并比较不同校正方法控制试验总的Ⅰ类错误的优劣。结果:在不同疗效分布类型中多终点变量间存在不同程度的相关性,在控制整个试验的alpha消耗中若考虑到多终点变量间相关性的存在,较之于一般的校正法能更好地控制试验总的Ⅰ类错误。结论:在新药临床试验中应严格控制主要指标个数,可根据疗效分布类型选择合适的评价指标,当存在多个主要指标时应考虑到指标间的相关性,并采用合适的校正方法控制试验的假阳性率。

Objective： To determine the effect of multiple endpoints on efficacy assessment in clinical trials.Methods： By Monte Carlo simulation,we studied the correlations of multiple endpoints in different efficacy distributions and corresponding effect on the efficacy assessment in clinical trials,and compared the performances of several adjusted methods on controlling the Family-Wise type I error rate（FWER）.Results： Simulation results showed that different degrees of correlations existed between different multiple endpoints.Conclusion： The number of endpoints should be controlled strictly.Correlations between endpoints should be considered where multiple endpoints are needed.Properly adjusted method should be chosen according to the correlation degree.