Skip to main content

Table 3 Summary of Flexibility/Benefits and Challenges/Obstacles of Various Less Well-Understood Adaptive Designs

From: Benefits, challenges and obstacles of adaptive clinical trial designs




Adaptive Randomization Design

■ Unequal probability of treatment assignment

■ Assign subjects to more promising treatment arm

■ Randomization schedule not available prior to the conduct of the trial

■ Not feasible for large trials or trials with long treatment duration

■ Statistical inference is often different, if not impossible, to obtain

Adaptive Dose Finding Design*

■ Drop inferior dose group early

■ Modify/add additional dose groups

■ Increase the probability of correctly identifying the MTD with limited number of subjects

■ Selection of initial dose

■ Selection of dose range under study

■ Selection criteria and decision rule

■ Risk of dropping promising dose groups

Two-stage Seamless Adaptive Design

(either phase I/II or phase II/III)

■ Combine two studies into a single study

■ Fully utilize data collected from both stages

■ Reduce lead time between studies

■ Shorten the development time

■ Additional adaptations such as drop-the-loser, adaptive randomization, and adaptive hypotheses may be applied at the end of the 1st stage

■ The control of the overall type I error rate

■ Sample size calculation/allocation

■ How to perform analysis based on combined data collected from both stages?

■ Is the O'Brien-Fleming type of boundaries feasible?

  1. *For example, adaptive dose escalation designs for cancer trials.