Independent from the study team
The individuals who execute the code break are firewalled from those producing TLF outputs for the blinded study team, preserving the integrity of any analyses that remain to be run.
What's included
What goes into a controlled unblind
- An unblinding plan naming who may request and who may authorize a code break, agreed before the study starts collecting data
- Unblinding triggered only at the pre-specified point — database lock, a serious safety event, or another documented trigger — never on an ad hoc basis
- Access to unblinded data restricted to the named individuals, with the rest of the study team kept blinded
- A full audit log of every request, approval, and code-break event, timestamped and retained
- Confirmation that treatment assignment is disclosed only to the extent, and only to the people, specified in the plan
Traceable end to end
The same audit discipline applied here carries through to statistical deliverables archival, where the unblinding log is retained alongside the SAP and outputs.