Friday, March 9, 2018

Understanding randomization in clinical trials

I am writing this to clear basics about randomization. It is a very important concept for understanding the clinical trial design and can come handy while critically analyzing any trial or designing your own study. 
This is not very important for any med school exam. I believe this is really important because of more extensive use of evidence-based medicine (EBM) in clinical practice and many clinicians lack the ability to skillfully evaluate a scientific manuscript.

I am planning to write more blogs related to evidence-based medicine, which might help our readers across the world to become expert in EBM.

  • RANDOMIZATION - randomly allocating participants into different treatment arms, purely on the basis of chance.  

  • Randomization is the cornerstone of clinical trial design. It's a very tricky concept and gets trickier when you start evaluating scientific literature critically or start designing a robust clinical trial.

  • It is pivotal in distributing confounders (eg. sex, age, history) equally in every treatment arm. Except for chance variation among the randomized group at baseline

Two most important  features of successful randomization:

1. Procedure truly allocates treatments randomly (based on chance)
2. Assignments are tamper proof

Randomization techniques:

1. Simple randomization:
By coin flipping (one side for treatment 1 and another side for treatment 2), shuffled deck of cards (even numbers for treatment 1 and odd numbers for treatment 2), throwing dice (numbers <3 for treatment 1 and numbers >3 for treatment 2). More better methods are random table method in stats books and computer software like excel.

Uses: in large sample size (>100 it should be preferred over block randomization)
Drawback: problematic in small sample size because it can create  unequal numbers in groups.

2. Block randomization: Ensure that participants are equally distributed among each group. Randomization is done in blocks, eg block size of six.
For example, a scientist enrolls only 6 patients per visit for a trial of total 60 patients. On each visit, he divides 3 patients each to treatment group A and B. At the end he will have 30 patient in both groups. See the figure 1 below.

 Figure 1. Block randomization of 60 patients in 6 patient blocks.

Drawbacks: Not suitable for randomization in non blinded trials, because randomization in small blocks makes a prediction of sequence easy.

3. Stratified Block randomization: It ensure that important predictor of outcome is more evenly distributed among study groups.
For example, if the age is a major determining factor in effectiveness or toxicity of the treatment then its imperative to have a similar distribution of ages in both treatment groups. Hence patients will be the first stratified into age groups and then they will be equally randomized in each arm. Like we did for Block randomization.

Drawback: only small number of baseline variables (2-3) can be managed by this technique.

4. Adaptive randomization: used for balancing more than 2-3 baseline variables.
5. Minimization: more complex adaptive randomization

I will continue more in next blog on randomization or other important concepts. Kindly post comments or question, which might help me, you, or other readers.

Dr. Gee


Hulley SB, Cummings SR, Browner WS, Grady DG, Newman TB. Designing clinical research. Lippincott Williams & Wilkins; 2013 May 8.

Suresh, K. (2011). An overview of randomization techniques: An unbiased assessment of outcome in clinical research. Journal of Human Reproductive Sciences, 4(1), 8–11.

No comments:

Post a Comment

This is express yourself space. Where you type create something beautiful! <3
Wondering what do I write? Well...
Tell us something you know better. You are a brilliant mind. Yes, you are! ^__^
Ask about something you don't understand @_@?
Compliment... Say something nice! =D
Be a good critic and correct us if something went wrong :|
Go ahead. Comment all you like here! (:

PS: We have moderated comments to reduce spam. ALL comments that are not spam will be published on the website.