Optimization Research Of Randomized Methods In Clinical Trials

Background and objectiveRandomization is the cornerstone of ensuring the quality of clinical trials,however,it has not attracted commensurate attention with its importance in practice.Although some new randomization methods have emerged in the past decade and can significantly increase the statistical performance of randomization,the real applications in clinical trials are rare.The reason may be due to the fact that the research on randomization methodologies is still controversial and does not form a good academic consensus;on the other hand,it may be due to the lack of practical operation and implementation tools corresponding to it.This study will endeavor to optimize the choice of methods based on the comparison of different randomization methods,form recommendations,and try to optimize the application of randomization through the development of tools.MethodsThis study reviewed various common randomization methods through literature review,selected several common restrictive randomization methods,and performed comprehensive performance evaluations from two aspects of randomness and imbalance by means of stochastic simulation methods.The randomness assessment index uses the probability of deterministic assignment(DA)and the probability of correct guess(CG);the imbalance evaluation index uses the maximum absolute imbalance in the randomization sequence(MI),the mean of the difference in treatment group sizes(Dn),and the probability of achieving exact balance in the randomization sequence(EB).First,the traditional block randomization methods,including the permuted block design(PBD)and the varied permuted block design(VBPD),are compared;then the several new randomization methods proposed in recent years are allowed to be compared.The MTI randomization methods selected in this study mainly include the big stick design(BSD),Chen’s biased coin design with imbalance tolerance(BCDWIT),and the block urn design(BUD).).We have separately compiled randomized SAS macro programs,developed a random allocation book technique that can achieve random distribution concealment,developed a randomized App tool based on the mobile phone client,and performed a report on the application of the randomization method.ResultsThe study found that when the block length of the variable block randomization method is greater than or equal to the permuted block randomized,its comprehensive statistical performance is higher,that is,under special circumstances,the variable block randomization method is also worth recommending in use.Comparing the random performance of multiple randomization methods based on the MTI procedures,the results show that the BSD,BUD,and BCD WIT methods is superior to PBD for the two-arm equal allocation trials.However,in addition to the MTI,the BCDWIT method require the adjustment probability to be set in advance.In contrast,the BSD and BUD methods are simpler and more efficient.The random allocation book technique can achieve random mask and concealment.The investigation results of its practical application show that the random allocation concealment technique is simple in principle,easy to be operated.At the same time,this study also developed a APP "Jinling Rat" based the mobile phone client,which is feature-rich,flexible setting,convenient operation,free from time,location and work environment and other factors,and can meet the need for randomization in most clinical trials.The SAS code of the multi-randomization methods in the end can control the imblance between groups,meanwhile,the SAS code of the selection of an optimal random assignment sequence also provides convenience for clinical researchers and tries to improve the items of the randomization report.ConclusionsIn this study,two groups of balanced open trials were carried out in the framework of restrictive randomization from the perspectives of randomized method selection,performance comparison,tool development,and randomized method specification reporting.In the selection of random methods for clinical trials,new methods such as BSD,BUD,and BCDWIT should be actively recommended,such as the use of block randomization methods,and the design of the block length needs to be concerned.The random allocation book technology in this research and the App based on the mobile phone client provide practical tools for the implementation of these randomization methods,as well as attempted to propose the items of the randomization reporting,which provided strong reference for clinical trial randomization applications.