How would you distinguish a protocol deviation from a data correction or a planned amendment?

Short answer

A deviation is something done differently from the approved protocol. A data correction fixes a record that did not reflect what actually happened. An amendment is a deliberate change to the protocol that is approved before it is used. The test is whether the conduct changed, the record changed, or the plan changed.

Why

These three get confused because they can show up as the same edit in the database. Suppose a visit date in the case report form says day 21, outside a window of day 14 plus or minus 2 days. If the source shows the visit really happened on day 15, the entry was a transcription error and the fix is a data correction: attributed, justified, supported by source, with the audit trail intact.

If the visit really happened on day 21, the conduct departed from the protocol, and that is a deviation to document and assess. Correcting the date to make the deviation disappear would be falsification. An amendment is different again.

If the team decides the window is too tight for all sites, it changes the protocol and gets approval before applying the new window. E6(R3) allows a change before approval only to remove an immediate hazard to participants, which is then reported. Deviations that repeat for the same reason often signal that an amendment is needed.

A strong answer covers

Conduct versus record versus plan
Source records decide which one it is
Corrections attributed, justified and audit trailed
Amendments approved before use, except to remove an immediate hazard
Recurring deviations as evidence for an amendment

Common mistakes

Editing data to bring an out of window visit inside the window.
Applying a planned protocol change at sites before approval because it seems minor.
Logging a typing error as a protocol deviation and inflating deviation counts.

The interviewer might ask next

How would you decide which deviations are important for a given trial?