Adaptive trials
Adaptive and master protocol type trials (such as basket trials, umbrella trials, platform trials, multi-arm, multi-stage trials) have designs that allow the treatment arms and allocation ratios to change during the trial. This means that users may require changes to be made to live randomisation systems. The ability to make different changes to systems depends on the randomisation method used and the Red Pill version.
Outside of a manual change request, the ability to make changes to live randomisation systems is summarised below by Red Pill version.
Red Pill version 28 onwards
Only limited changes to the list based method is allowed as shown below. Most changes require a change request to be submitted.
| Change | Minimisation | Dynamic blocks | List |
|---|---|---|---|
| Add treatment group | ❌ | ❌ | ✅ † |
| Remove treatment group | ❌ | ❌ | ✅ † |
| Change allocation ratio | ❌ | ❌ | ✅ † |
| Add stratification factor | ❌ | ❌ | ❌ |
| Remove stratification factor | ❌ | ❌ | ❌ |
| Add or remove categories from existing stratification factor | ❌ | ❌ | ✅ |
| Randomisation method | ❌ | ❌ | ❌ |
† By using the Upload new list feature.
Red Pill version 31 onwards
Red Pill 31 introduced much more flexibility into the ability to make changes to minimisation and dynamic block randomisation using the CRF builder and re-releasing the live system. Note that changes to treatment groups or allocation ratios result in the previous minimisation scores or dynamic blocks being disregarded - see stratified minimisation for more details.
| Change | Minimisation | Dynamic blocks | List |
|---|---|---|---|
| Add treatment group | ✅ | ✅ | ✅ † |
| Remove treatment group | ✅ | ✅ | ✅ † |
| Change allocation ratio | ✅ | ✅ | ✅ † |
| Add stratification factor | ✅ | ✅ | ❌ |
| Remove stratification factor | ✅ | ✅ | ❌ |
| Add or remove categories from existing stratification factor | ✅ | ✅ | ✅ |
| Randomisation method | ❌ | ❌ | ❌ |
† By using the Upload new list feature.
Red Pill version 35 onwards
Version 35 introduced the R-code method. This allows complete flexibility to change the randomisation method at any time by activating a new R-code function. This is done within the application itself and does not require using the CRF builder or re-releasing the live system.