Study Of Statistical Method Based On Weighted Effect Size For Simultaneous Global Drug Development Program

Background:With the increasing globalization of new drug clinical trials,multi-regional clinical trials(MRCT)play an important role in global drug development.However,the design and implementation of MRCT tend to focus more on the overall effect.In order to ensure the power of regional effect test,some scholars have proposed Simultaneous Global Drug Development Program(SGDDP),which is to supplement a local clinical trial simultaneously or sequentially on the basis of MRCT.At present,it is recommended to use weighted Z method to test the regional effects of SGDDP.However,the interpretation of the weight parameter remains unclear in this method,particularly about its clinical and regulatory meaning which is critical for the evaluation.And the strong assumption that there is no interaction between treatment efficacy and regional population should be hold,which limits its popularization and application.Objectives:In this study,we propose a regional effect test and sample size calculation method using the idea of weighted effect size for continuous and binary data in independent and paired non-inferiority SGDDP design,and compare it with weighted Z method by simulation,in order to provide reference for the SGDDP participated by China.Methods:?For independent SGDDP designs,firstly,the statistical performance of weighted Z method and weighted effect size method for continuous data is compared by simulation.Then,weighted effect size method for binary data is proposed with OR as efficacy indexes and compared with weighted Z method.?For paired SGDDP design,we construct the statistics and sample size calculation formulas of weighted Z method and weighted effect size method,with mean difference and RD as efficacy indexes for continuous and binary data,respectively.We also compared the two methods by simulation study.All statistical simulations are implemented by Monte Carlo method through the IML process of SAS 9.4.The evaluation indicators of simulation comparison include type I error and power.Results:Considering continuous and binary data,weighted effect size method is applied to independent and paired SGDDP design and compared with weighted Z method.The simulation results show that:?When the new drug is inferior to the positive control drug in non-target regions,both two methods can commendably control type I error rates,but the power of the weighted effect size method is higher.?When the new drug is non-inferior to the positive control drug in non-target regions.weighted effect size method controls type I error rates well.But the type I error rates of weighted Z method is greatly affected by the data from non-target regions.With the increase effect size of non-target ethnic(NTE),type I error rates inflate seriously.Although weighted Z method can gain higher power with smaller sample size and weight than weighted effect size method,but this is at the cost of increasing type I error rates,which will lead researchers and regulators to mistakenly accept drugs that do not meet the requirements in the target region,harming patients' health and interests.Conclusion:The weighted effect size method proposed in this study uses the ideal sample size proportions of target and non-target regions as weights,which is easy to calculate and interpret.The study results show that this method has good performance under various parameters combination.Especially when the efficacy of new drugs is significant in non-target regions,the method proposed in this study controls type I error rates better and make up for the shortcomings of the original method.To sum up,compared with weighted Z method recommended at present,weighted effect size method has better statistical performance and application value in the regional effect test of non-inferior SGDDP design.