Protocol Deviations: Important, or Just Recorded

A protocol deviation is any departure from the approved protocol. Every trial has them. A trial reporting none has a recording problem rather than a perfect record.

The question that matters is not whether you had deviations. It is whether you classified them correctly, because classification drives what you report and what happens to the data.

Important versus not

ICH E6 distinguishes deviations generally from those that are important: deviations that may significantly affect the completeness, accuracy or reliability of the trial data, or that may significantly affect a subject's rights, safety or wellbeing.

Some organisations use major and minor; the E6 term is important. Whatever the vocabulary, the line is the same, and it is a judgement rather than a category you can automate entirely.

Typically important:

• Enrolling a subject who did not meet eligibility criteria.
• Missing or late safety assessments where the assessment protects the subject.
• Administering the wrong treatment, wrong dose, or dosing outside the permitted window in a way that affects exposure.
• Informed consent obtained late, on the wrong version, or not at all.
• Failure to report a serious adverse event within the required timeframe.
• Unblinding outside the procedure.

Typically not:

• A visit a few days outside the window with no assessment consequence.
• A non-safety assessment missed where the endpoint is unaffected.
• Administrative and documentation errors corrected contemporaneously.

Why classification decides your CSR

Important deviations must be described in the clinical study report, and ICH E3 requires them in the study patients section along with their effect on the analysis.

That has a downstream consequence people meet late: a deviation affecting eligibility can move a subject between analysis populations. If your per-protocol population excludes subjects with important deviations, the classification decision is a statistical decision, and it should be made before unblinding rather than after.

Deciding classification after seeing the results is the thing an inspector is looking for.

The reporting obligations

To the IRB or ethics committee. Deviations affecting subject safety or the conduct of the study are reportable, and many require prompt notification. A deviation made to eliminate an immediate hazard to a subject is permitted without prior approval but must be reported promptly afterwards.

To the sponsor. From the site, per the protocol and the monitoring plan.

In the trial master file. The deviation log and its supporting documentation, filed contemporaneously.

In safety reporting, where the deviation itself relates to an adverse event.

Where it goes wrong

Under-recording. Sites that record only what monitoring finds. The log then describes the monitoring, not the trial.

Reclassifying late. Downgrading a deviation after seeing its effect on the analysis.

No trend analysis. Twenty minor deviations of the same type at one site is a systemic finding regardless of individual severity, and ICH E6(R3) expects that pattern to be noticed.