Dual Minimization in Clinical Trials

• Activity Number: 618
• Type: Contributed
• Date/Time: Wednesday, August 3, 2016 : 2:00 PM to 3:50 PM
• Sponsor: Mental Health Statistics Section
• Author(s): Jay Taves* and Donald Taves
• Companies: and University of Washington
• Keywords: Clinical Trials ; Minimization ; Randomization ; Study Design

Abstract

Minimization as currently employed is not an ideal solution for the type of selection bias that has been considered in the literature. It prevents selection bias with an informed cheater by unbalancing non-cheaters patients of the same type in the opposite direction. We can make it more difficult for a person to cheat by taking the balances by investigator into account. Three ways to make minimization pay attention to the imbalances by investigator were evaluated by simulations using the same sequence of investigators and their patients. The endpoints were the balances achieved overall and by investigator. As expected there was a reciprocal trade-off between overall and investigator balance for all three methods but the least severe trade-off appeared in dual minimization, which separately takes into account the magnitude of the imbalances by investigator and the usual intrinsic variates and makes the assignment that causes the greatest gain of balance. It gives less severe trade-offs, and the more complex programing makes cheating more difficult than the usual minimization.