IRT Commandment #6
The Randomization Schedule Is Sacred & Shall Not Be Modified
There’s a question that surfaces sooner or later in many studies:
“This patient was randomized in error; can you just remove them from the list?”
It sounds reasonable. It never is.
Why This Matters
Randomization is not a visit. Randomization It is a moment in time, a discrete system action that assigns treatment based on the information available at that exact moment.
Once it happens, it becomes part of the study’s permanent record.
Any attempt to remove, adjust, or “clean up” that action undermines the integrity of the randomization process itself.
The Risk You Inherit
When a randomized subject is removed, the questions don’t stop at why the error occurred. They multiply:
Was the subject removed because the allocated treatment was undesirable?
Was the decision influenced by emerging data?
Was someone trying to improve the apparent performance of the study?
Even if none of those things are true, the optics alone can be damaging.
Intent-to-treat principles exist precisely to protect against these risks.
Undoing randomization, even with good intentions, cuts directly across that protection.
System Record vs. Clinical Record
The data used at the moment of randomization, whether later found to be correct or not, is the data that actually determined the treatment assignment. That record must not be changed.
If more accurate or corrected information becomes available later, it should be captured as new/revised clinical data and documented appropriately. It should not be used to rewrite the randomization history.
The clinical record can evolve, but the randomization record cannot.
Trying to correct randomization after the fact does not improve data quality. It undermines the credibility of the process itself.
A Line That Shouldn’t Be Crossed
Once a subject is randomized, that record must stand, even if the subject is later withdrawn, deemed ineligible, or never dosed.
That’s not rigidity, that’s integrity and scientific rigor.