RANDOMIZATION

Rajalakshmirajalakshmi.shanmugasundaram@takesolutions.com

IASCT - ConSPIC201208-Sep-2012

Agenda

Overview

Types of randomization

SAS Procedures

Overview

What is randomization?Assigning research participants by chance - To

treatment group or control group.

Why it is Important?- Balances the Treatment groups.- Eliminates the selection bias.

Types of Randomization

Simple Randomization

Block Randomization

Stratified Randomization

Covariate Adaptive Randomization

Simple Randomization

- Single sequence of random assignments of subjects to a particular treatment.

E.g. Using Random Number tables

AdvantageIt is easy and simple to implement for larger clinical trials,

say [n>200].

DisadvantageProblematic in relatively small sample size, which results

in imbalance in the assignments of treatment groups.

Simple Randomization

Randomization in SAS

Using PROC PLAN

SEED – An integer ; Blind the other team member; regenerate the same randomization.

FACTORS – Specifies the Factors for the design of the study –FACTORS; BLOCK SIZE.

TREATMENTS – Specify the treatment. Treatment number >= last factor values in the factor statement.

OUTPUT – It creates the dataset containing the end point.

- Randomize subjects into groups with equal sample size.- Balance in sample size across groups over time.- The number of participants are similar at all times, when the blocks are small and balanced among treatment groups.

Block Randomization

Output:

Ex1: Generate a randomization schedule for 24 subjects for 2 treatment groups.

Block Randomization

Code:

1= Treatment Group2= Control Group

Advantage:- The block is used to reduce the selection bias, when the

assignments can be determined with certainty.

Disadvantage: - Although balance in sample size may be achieved with this

method, groups may be generated that are rarely comparable in terms of certain covariates.

Block Randomization

- Control and balance the influence of covariates.- The specific covariates must be identified by the researcher who understands the potential influence each covariate has on the dependent variable.- It can be achieved by generating a separate block for each combination of covariates and participants are assigned to the appropriate blocks of covariates.- Once the participants have been assigned into blocks, the subject will be assigned randomly to either the treatment group or the control group.

Stratified Randomization

Stratified Randomization

Ex2: Generate a randomization for 48 subjects with 2 treatment groups, controlling for the covariates Gender 2-levels (Male; Female) and Body mass index 3 levels (Underweight; Normal; Overweight)

Random assignment of “Male” and

“Normal”

Stratified Randomization

Code

Output

Stratified Randomization

Advantage: - Simple and useful technique, especially for smaller clinical trials.

Disadvantage:- Becomes complicated, for many covariates to be controlled.- Two many block combination lead to imbalance treatment allocation.- When Baseline character is not available before assignment, stratified randomization is difficult.

Stratified Randomization

-New participant will be sequentially assigned to a particular treatment; based on the specific covariate and pervious assignments of participants.-Uses the method of minimization by assessing the imbalance of sample size among several covariates.- Frane method controls for quantitative covariates in addition to categorical ones.

Covariate Adaptive Randomization

Step 1: Temporarily assign new patient to the treatment group a.

Step2: Calculate the Pearson’s χ2 Goodness-of-Fit statistics for the

covariate groups for which the new patients would belong.

Step 3: Identify the maximum χ2 test statistics among all the covariate

group.

Step 4: Remove the patient from group A and repeat step 1-3 for all

other treatment groups.

Step 5: Identify the minimum χ2 test statistics over all the identified test

statistics (one from each repetition of step 1-3).

Step 6: Assign the new patient to the group for which the minimum χ2

test statistics was achieved.

Covariate Adaptive Randomization

Ex3: Consider the efficacy of two blood pressure reducing drug is being compared in patients with high blood pressure. We have 2 covariates Baseline blood pressure (hypertensive; Pre-hypertensive) and age (>=65 and <65).

A new patient arrives at the trial who is hypertensive and <65. Temporarily assign this new patient to Drug A and calculate the χ2

Goodness-of-Fit test statistics for both the covariate.

Covariate Adaptive Randomization

- The largest χ2 test statistics when subject in Drug A is 3.0. - Repeat the process by replacing the subject to Drug B.- The largest χ2 test statistics when subject in Drug B is 1.33.- Assign New patient to Drug B; Smaller imbalance for the covariate when patient in B.

Covariate Adaptive Randomization

Input dataset for the SAS macro:

Covariate Adaptive Randomization

Covariate Adaptive Randomization

Output

Advantage:- It balance the covariates across the treatment groups.

Disadvantage:-It determines the treatment group of the next patients, rather than increasing the probability of being randomly assigned to that treatment group, selection bias is present.-Further investigation should consider randomizing patients with increased probability rather than directly determining the group assignment of the next patient.

Covariate Adaptive Randomization

- Simple randomization works well for large trial (n>200).

- Block randomization is used to achieve balance in Sample size.

- To achieve balance in Baseline characteristics, stratified randomization is widely used.

- Covariate adaptive randomization, can achieve better balance than other randomization methods.

Conclusion

Reference

Issues in Outcomes Research: An Overview of Randomization Techniques for Clinical Trials; J Athl Train. 2008 Mar-Apr; 43(2): 215–221. Minsoo Kang, PhD,1 Brian G Ragan, PhD, ATC,2 and Jae-Hyeon Park, PhD3

An overview of randomization techniques: An unbiased assessment of outcome in clinical research; J Hum Reprod Sci. 2011 Jan-Apr; 4(1): 8–11. KP Suresh

Thank You!