Windows, Deviations and What Counts as Late

Part of Research Nurse (RN) Training: Reading the Protocol Like a Nurse.

About 2 minutes. Free, in your language, no account needed.

Why this matters

A deviation is any departure from the protocol. That is all it is. It is not an accusation.

Sites that fear deviations under-report them. Under-reported deviations are far more damaging than the deviations themselves, because they hide the pattern that would have been fixable.

What counts

A missed assessment. A visit outside its window. A dose given at the wrong time. An eligibility criterion applied incorrectly. A sample processed late.

Some are trivial. Some threaten a participant's safety or the reliability of their data. The protocol or the sponsor usually defines which are major.

Your job is not to grade them. Your job is to notice and report them.

Report it even when it turned out fine

A participant was given a dose twenty minutes outside the specified window and had no ill effect. It is still a deviation.

"No harm resulted" is an outcome, not a reason to omit the record. Reporting it lets the sponsor see whether twenty-minute drifts are happening across many sites, which is a question no single site can answer.

Prevention beats reporting

Most deviations are predictable. A participant who works nights will struggle with morning visits. A fasting requirement will be missed if nobody tells the participant the day before.

The nurse who calls a participant two days before a complex visit prevents more deviations than any form ever fixed.

In practice

You realize mid-visit that a required ECG was not done at the previous visit.

You do not backdate it. You report it as a missed assessment, do the ECG now if the protocol allows, and note what was done. The investigator decides whether anything further is needed.

The difference between a small finding and a serious one is usually whether the site found it or the monitor did.

What this looks like when it goes wrong

A site records only the deviations it considers important. The monitor finds six unreported ones in a single file.

The sponsor now has a harder question than the deviations themselves. If these were not reported, what else was not reported? Every other record at that site is now less trusted.

One judgment call about what mattered cost the site its credibility.

Key takeaways