The IVR Subject RANDOMIZATION System:  A brief description

 

The Department of Bioinformatics has available a computerized randomization system that is accessible by a touch-tone phone.  Features of the system include:

• 24 hour accessibility
• ease of use
• immediate randomization, even with complex allocation schemes
• accurate data collection

 

 

You have already used the technology that we call IVR (for Interactive Voice Response) when you’ve called your bank or a utilities company and heard “Please enter your account number…”  When you call, you are contacting a computer and hearing recorded voices. 

 

 

A variety of schemes for treatment assignment are available.  All can be adapted to control for factors known to influence the outcome.  Following is a list of the methods with a short description.

·        Stratified block allocation (Zelen, 1974).  Arm counts are balanced within each combination of factor levels (stratum) and across strata.  The method is appropriate only if there are few strata relative to the sample size.  Three factors maximum.

• Minimization (Simon & Pocock, 1975).  Balances treatment assignment within each factor.  Appropriate to use when there are many prognostic factors relative to the sample size.  Two versions are available, one that will result in equal arm counts if the target sample size is known.  Five factors maximum.
• Distance function (Raghavarao, 1980).  This is the only method that can have prognostic factors that are continuous rather than categorical (e.g., the actual age versus age groups).  Two versions are available, one being more appropriate for smaller sample sizes relative to the number of factors.  Five factors maximum.

 

Summary tables will be sent periodically (e.g., monthly) to the project director.  These provide interim arm counts at each factor level as well as totals across all levels.  Other types of reports can be developed based on the researchers’ needs.

 

 

A DEMONSTRATION RANDOMIZATION PROGRAM

 

You can access our demonstration program by following the following instructions.  Accessing the system is a two-part process:  both a study number and a site id need to be entered.  Three stratification variables are used for the randomization in our demonstration.  The computer voice script is written in italicized bold font.  The entries you should enter are in a bolded font.

 

Phone number:  (802) 656-4652 if calling from outside UVM.

 

Welcome to the Interactive Voice Response System.  Please enter the study number.

-         103

 

You have reached the subject allocation program.  Please enter your site id.

-         301

 

Stratification variable.  Please enter the subject’s sex:  1 if male, 2 if female.

            - 1

 

Please enter the subject’s age. 

            - 33  (Any number between 18 and 50 will be accepted.  Any other value would get an “answer out of range” error message.)

 

Please enter the lab result.

            - 22  (Valid range is 1 to 200.)

 

Please listen as your entries are read back to you.  Variable 1, Sex, male.  Variable 2, Age, 33).  Variable 3, Lab result, 22).  Press 1 to confirm, 0 to re-enter.

            - 1

 

Please record the following information for your records.

The subject id number is:  (number).

The subject is assigned to:   (the control group/ the first treatment group/ the second treatment group).

Press 1 to listen to the above information again.  Press 2 to enter another subject.  Press 0 to exit.

            - 0

 

If you have any questions about the demonstration program, please contact:

 

Dr. Shelly Naud, phone 1-802-656-3822, email click here.

 

If you would like to use the Bioinformatics Department randomization system for a research project, please contact Dr. Taka Ashikaga, Bioinformatics Department Director, for a cost estimate:

 

            Dr. Taka Ashikaga, phone 1-802-656-2526, email click here