Click on the magnifying glass to the right. Starting typing a topic you wish to learn about

Skip to end of metadata
Go to start of metadata

You are viewing an old version of this page. View the current version.

Compare with Current View Page History

« Previous Version 8 Next »


It is strongly recommended that you have your statistician involved in both the project design as well as development of the allocation tables. All projects using Randomization MUST be reviewed by BMIC Faculty prior to project start.


Initial Considerations

Randomization is a process that assigns participants by chance (rather than by choice) into specific groups, typically for clinical research and clinical trials. REDCap DOES NOT create the randomization table for you.
Planning the randomization is crucial and must consider factors such as:

  1. Timing of when to randomize
    • After determining eligibility
    • As close to treatment time as possible (to avoid death or withdrawal before treatment starts).
  2. Having Standards of Practice in place to ensure the integrity of the randomized trial is dependent on these SOPs. You should have defined processes for
    • Randomization procedure
    • Unblinding procedure for blinded studies
    • Assigning Data Access Groups if you plan to utilized these
    • Other
  3. Your Allocation table needs to be at least as large as 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
    • Consider a 15% larger allocation table
    • 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. A key step in planning your randomized study is to determine which type of randomized study you wish to conduct:
    • 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 we have piped in strata of Age & Gender captured in the Demographics Instrument into the Randomization instrument:
  6. Create a field in REDcap where the data you are randomizing will be placed. Do this in advance of setting up the REDCap randomization parameters

7.  You will need to create two allocation tables: One for development (testing) and one for production. They can NOT be identical.

8.  Blocks support equal allocation to both control and treatment groups. Consult with your project Statistician

9.  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.


NEVER use the REDCap Template Allocation examples as the allocation tables for your project. They are solely meant to provide examples and NOT be utilized in a project.  Your statistician should create your allocation tables in the format of these template tables.


How the REDCap Randomization Framework Works


REDCap DOES NOT create the randomization table for you.

  • This table must be generated outside of REDCap using other software (e.g. SAS, Stata, R), most likely by the statistician/data analyst involved in your project.
    • Then formatted as the REDCap template allocation provides an example
  • REDCap then uses the table as a lookup for determining how to randomize the subjects
  • Creating your allocation table outside of REDCap lets you choose exactly how you wish to structure your allocations and assignments. Examples: block sizes, permutations, and stratification balancing




The Process – REDCap Randomization Setup

The randomization model is defined via the "Setup" tab

  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 

  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 
    1. READ the information provided after clicking on More Details 


    2. Set Up your randomization
      1. Select desired choices at the bottom of the page on the Setup Tab 


      2. Save by clicking Save Randomization model in lower left corner 


      3. Review Demo Allocation Tables (THESE TABLES ARE EXAMPLES ONLY; NOT TO BE USED for your project)
      4. Upload Your Project's Allocation tables
        1. Need one allocation table for test.
        2. Need second table for production. They cannot be identical copies of each other 
  4. TEST your project by adding records
    1. You can use the Record Dashboard to see how participants have been randomized to slots in the allocation table




Additional Considerations

  1. Randomization must be set during development
    • You should not add randomization to a project already in production with existing data
  2. 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.
  3. 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. 
  4. Testing:
    • Ensure there is a reconciliation between the group assignments in the Randomization field, and the allocation tables 
  5. Randomization Field must be on a data collection form
    • This is field the where the allocated randomization group is saved and stored and where the randomization button will appear on your data collection form
    • Randomization must be done in data collection mode. Can not randomize on a survey
  6. Can't modify a randomized field nor upload information into a randomized field
  7. Group assignments are assigned based on order of randomization, not enrollment order
    • i.e. If you enroll record 1, then randomize record 1, then enroll record 2, then randomize record 2, the resulting randomization assignment will be record 1 gets the 1st assignment from the allocation table and record 2 gets the 2nd assignment
    • If you enroll record 1, then enroll record 2, then randomize record 2, then randomize record 1, the resulting randomization assignment will be record 2 gets the 1st assignment from the allocation table and record 1 gets the 2nd assignment










      


         


Final notes about Randomization

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

2. The integrity of any RCT is dependent on everything having SOPs, especially randomization!




  • No labels