Sampling > Stratified Randomization in Clinical Trials
You may want to read this article first: Permuted Block Randomization
What is Stratified Randomization?In stratified randomization (sometimes called Stratified Permuted Block Randomization), trial participants are subdivided into strata, then permuted block randomization is used for each stratum. The goal is to create a balance of clinical/prognostic factors, because the trial may not have valid results if factors are not well balanced.
This form of randomization is only important for small clinical trials (fewer than 400 patients), where known clinical factors (like gender, age, disease stage, or obesity) are thought to effect treatment outcomes. Larger trials don’t use the technique because it’s unlikely to find imbalances in clinical factors for a randomized large group.
Don’t try to balance every clinical factor — choose the most important ones. Too many strata can result in too few patients in each. As an extreme, you could end up with only one patient — or even zero patients — in each strata. Therefore, you should keep strata to a minimum. Between 1 and 5 factors (i.e. randomization variables) is commonly recommended; Each factor usually has between 2 and 4 levels. The recommended number of stratification factors is usually one or two.
The number of patients in each strata do not have to be even. The important thing is to make sure an equal number of patients are assigned to block A or B, which is taken care of if you use permuted block randomization within the strata.
- Polit DF Beck CT (2012). Nursing Research: Generating and Assessing Evidence for Nursing Practice, 9th ed. Philadelphia, USA: Wolters Klower Health, Lippincott Williams & Wilkins.
- Therneau, T. How many Stratification Factors is “Too Many” to Use in a Randomization Plan? Retrieved July 21, 2016 from http://www.mayo.edu/research/documents/biostat-57pdf/doc-10027486.
If you prefer an online interactive environment to learn R and statistics, this free R Tutorial by Datacamp is a great way to get started. If you're are somewhat comfortable with R and are interested in going deeper into Statistics, try this Statistics with R track.Comments? Need to post a correction? Please post on our Facebook page.