Randomization

1. Timing of when to randomize, consider when this will occur in your projects operational workflow…

   • Will this occur after determining eligibility or perhaps as close to treatment time as possible (to avoid death or withdrawal before treatment starts)?

2. Having Standards of Practice (SOPs) to ensure the integrity of the randomized trial. You should have defined processes for…

   • Randomization procedure

   • Unblinding procedure for blinded studies

     • Assigning Data Access Groups if you plan to utilized these

3. Your Allocation table needs to be at least 20% larger than your sample size. Major issues can occur mid-study if you have more subjects than spots in your allocation table.

   • A good rule is that for any strata, you should have as many options as the number of participants. For example, if you have 4 strata options, and you plan to enroll 100 participants, you should have 400 slots in your allocation table. This is well over the sample size but ensures you have enough slots.

     • Accommodates drop-out rate

     • Accommodates if study gets larger

   • You will need to create two allocation tables, one for development mode (testing) and one for production:

     • A new allocation table must be used for Production, you may NOT use the same table used in development mode.

     • When you have multiple sites, definitely talk to your statistician to ensure you include site in randomization allocation to prevent all controls at one site and treatment at the other site.

   • NOTE: This article is not meant to supplant any advice you get from your statistician. The size of your allocation table is dependent upon block size, design features, sample size, type of sites, etc.

4. Determine which type of randomized study will be conducted:

   • Unblinded

   • Single blind

   • Double blind

5. Define which strata you will be randomizing from (e.g., age, gender). It is a clinically best practice to pipe these fields into your randomization instrument to help you remember the criteria without going back to another instrument. For example, below is a descriptive field which has piped information into the Randomization instrument from a different form so users do not have to navigate back and forth to confirm inclusion/exclusion criteria:

   Open

   

6. Create a field in REDcap where the data you are randomizing will be piped (as shown above). 

   1. Do this before configuring your Randomization settings/parameters

During the randomization review, be prepared to address the following:

1. Project workflow/operations: Be prepared to present an overview of your project purpose, configuration, # participants, recruitment processes, etc. A screen-share will be hosted to provide BMIC team with a walkthrough through of your project workflow, configuration and user rights set up, etc..

2. Consent process and configuration

3. Inclusion/Exclusion instrument & criteria: Be prepared to showcase your inclusion criteria and how the data is captured

4. Randomization type: Blinded or Unblinded

5. Randomization: Who will perform the randomization?

6. Randomization field & Safeguards: Be prepared to review the instrument where your randomization field is located and show the safeguards you have in place to ensure the Randomization button only appears when a participant meets all inclusion criteria. This can be accomplished using branching logic.

   1. For example, if your inclusion criteria is that the participant is over 18 and female, then you would create branching logic for your randomize field which hosts logic that enforces the field only be shown when the participant has selected they are over 18 AND female.

   2. More info on branching logic: Branching Logic 101

   3. Be prepare to discuss your allocation table numbers which account for potential attrition

7. User Rights: Discussion of user rights

   1. Users who are Blinded vs Unblinded should have the appropriate user permissions assigned for randomization

   2. Consider hiding your randomization instrument from users who should not be granted the ability to randomize

The randomization model is configured via "Project Setup" > “Option modules & customizations” as shown below.

1. Define your instruments and set up your project to be a Randomization Project

   1. Turn on the Randomization module by clicking "Enable" on the Project Set-Up page under the section Enable Optional Modules & Customizations 

      Open

      

       

2. Activate User Rights from the left side to ensure you and your staff have access to appropriate rights to assist with the randomized study.

   1. Click to activate access to specific Randomization user rights

   2.  
      
      

   3. User privileges for randomization

      • User privileges can be set to allow users

      1. Set up randomization

      2. Perform randomization

      3. View allocation dashboard

      • Someone with randomize privileges will be able to view and modify existing data used for stratification even if they do not have Forms privileges
      
      

3. Return to Project Setup page and click on Set Up Randomization 

   Open

   

   1. READ the information provided after clicking on More Details 
      
      

      Open

      

   2. Define your Randomization model

      1. Complete (Step 1)

         • Select your randomization type (stratified is commonly used)

         • Select relevant criteria fields

         • Choose Randomization field

         • Save Randomization model
           
           

           Open

           

            

      2. Download Demo Allocation Tables (these tables are NOT to be used for your project, examples only)

         Open

         

      3. Upload Your Project's Allocation table for the Development instance

         • You will need to create two allocation tables, one for development mode (testing) and one for production:

           • A new allocation table must be used for Production, you may NOT use the same table used in development mode.

4. TEST your project

   • Use the Record Dashboard to add test records and test your randomization process & setup
     
     

5. Move to Production

   • After you have fully tested your project, request that your project be moved to Production. Note that you will need to have your Production randomization allocation table uploaded.

   • The BMIC team will coordinate a meeting to review as noted above to ensure. The Prod excel template

1. User Privilege

   • If someone is given 'Randomize' privileges, they will be able to view and modify any existing data already collected for the randomization strata fields (if stratification is used) when they are performing the randomization, even if they do not specifically have form-level rights to view the form on which a strata field exists.

   • Caveat: Randomize rights trump form-level rights in this way, but only for the randomization strata fields.

2. Extending Allocation Tables

   • Allocation tables can be extended but it is not a great idea - create enough size upfront to handle the trial

   • Numbers assigned to the treatment groups should match the numbers in the allocation table. 

3. Group assignments are assigned based on order of randomization, not enrollment order

   • i.e. If you enroll record 1 > randomize record 1, then enroll record 2 > randomize record 2, the resulting randomization assignment will be…

     • record 1 gets the 1st assignment from the allocation table

     • record 2 gets the 2nd assignment from the allocation table