# Innovative and effective subject randomization methods

Published on March 29, 2020   46 min
• Review

#### Other Talks in the Category: Methods

0:00
Hi, everyone. My name is Wenle Zhao. I'm working at the Medical University of South Carolina in the United States. I'm a professor for Biostats, mainly working on clinical trials including clinical trial design and implementation.
0:16
This presentation covers the following contents. First, I will review the advantages and disadvantages of the commonly used permuted block design. Then, I will present three new randomization designs that overcome the disadvantages of the permuted block design. At the same time, we will keep the good properties of that design. So, the block urn design offers a much higher allocation randomness than permuted block design under the same block size. The mass-weighted urn design can actually target any unequal allocation ratios. Not only those with small integers like 1-2 or 2-3, but also those with irrational numbers like one to the square root of two to the square root of three. The minimum sufficient balance method prevents serious imbalance on a larger number of baseline covariates, meanwhile maintaining a high level of allocation randomness.
1:22
First, let's review the advantages and disadvantages of the most commonly used permuted block design.
1:31
Permuted block design uses a pretty generated randomization sequence with a random permutation of blocks. The first step is to fill each block with treatment assignment based on the allocation ratio and assign two random numbers to each sequence code. Here is an example of a drug study with a 1-1 allocation ratio and a block size of six. Step 2: sort by the first random number and then assign the drug kit ID accordingly so that drug kit ID carries no information on treatment assignment. Step 3: sort by the second random number with each block and assign randomization sequence number. Step 4: pack study drug kit by the treatment group in the groups. Step 5: assign drug kit to a subject based on a randomized sequence number.
Hide