正如小胖在前文中提及的，由于分层随机化对分层因素个数的限制，因此分层随机化只能应付有限的协变量，然而有些临床试验特别是一些肿瘤试验中，预后因素较多而且也都比较重要，那怎么解决呢？这时便出现了动态随机化，它可以解决分层因素较多导致层数过多的问题，可以同时使多个协变量达到组间均衡。关于动态随机化的介绍，可参见小胖说统计之三十八。动态随机化的应用还存在着一定的争论，如果使用此方法，其使用的理由要有充分的临床和统计学依据。

Paul Dorian. Placebo-Controlled, Randomized Clinical Trial of Azimilide for Prevention of Ventricular Tachyarrhythmias in Patients With an Implantable Cardioverter Defibrillator. Circulation. 2004;110:3646-3654.

“The randomization was conducted in a ratio of 1:1:1 (75 mg of azimilide, 125 mg of azimilide, and placebo); patients were stratified within a region (every country was a separate region except for the United States, which had 4 regions) by ß-blocker usage, left ventricular ejection fraction ( 40% or >40%), and ICD type (existing ICD or new ICD). A dynamic randomization scheme using Schouten’s methods16 was used through an interactive voice response system to randomize patients to treatment groups. Patients, investigators, and the sponsoring agency were blinded to the treatment assignment, and the codes were only available to the Data and Safety Monitoring Board. Blinding was maintained throughout the entire study. Patients were maintained on the originally assigned blinded therapy for 365 days (unless withdrawn for any reason), regardless of the number of intervening arrhythmia events”

上边是一个简单的使用动态随机化的试验的例子，仅供大家参考。