Amendments and Version Control

Part of Clinical Research Associate (CRA) Training: The Protocol as the CRA's Rulebook.

About 5 minutes. Free, in your language, no account needed.

Why this matters

Amendments are where well-run sites fail. A change is approved centrally, communicated by email, and implemented at eleven sites in eleven slightly different ways on eleven different dates.

Version control is unglamorous and it is one of the highest-yield things a monitor checks.

Why protocols are amended

• New safety information about the product.
• Changes to eligibility criteria, often because recruitment is slower than expected.
• Adjustments to the visit schedule, assessments or endpoints.
• Operational refinements and clarifications.
• Regulatory or ethics committee requirements.

Some changes are substantial and require regulatory and ethics approval before implementation. Others are administrative. The sponsor classifies them; you verify the site implemented them correctly.

The sequence, and it never changes

1. The sponsor issues the amendment.
2. Regulatory authorization is obtained where required.
3. The ethics committee approves the amended protocol and any amended participant-facing documents.
4. The site is trained on the change, and that training is documented against the new version.
5. Superseded versions are withdrawn from use the same day.
6. Participants are re-consented where the change affects them, before the changed activity applies to them.
7. The change takes effect from the implementation date at that site.

Steps 3 and 6 are where sites get into trouble. Implementing before approval is a serious deviation. Performing a newly added procedure before re-consenting the participant is a consent violation.

What a monitor verifies

Approvals. The ethics committee approval for the specific amended protocol version and each amended consent form version, and where required the regulatory authorization, both dated before the site's implementation date.

Training. Every delegated staff member trained on the new version, with the version number and date recorded. New staff joining later trained on the current version, not on whatever is nearest.

Withdrawal of old versions. Physically and electronically. Walk the clinic. Superseded consent forms in a drawer will eventually be used.

Re-consent. A tracker showing, for every enrolled participant, which consent version they signed and when. Compare re-consent dates against the date the changed activity first applied to that person.

Practice matches version. Visit procedures match the current schedule of assessments. Eligibility checks use the current criteria. Data entries reflect the requirements in force at the time.

Records reflect the transition. For participants who spanned the change, the earlier data should follow the earlier version and the later data the later version. This is normal and should not be "corrected".

The version log

Ask every site to maintain a simple version log, and check it at every visit:

Four dates, one row, and most amendment findings disappear. Sites that do not keep this cannot answer an inspector's simplest question: which version were you using in March?

The multi-site problem

You will monitor several sites running the same amendment. Expect them to implement on different dates, because ethics approval timing differs. That is legitimate.

What is not legitimate is a site that cannot say which date it implemented, or that applied part of an amendment and not the rest. Watch particularly for the site that trained staff but never withdrew the old consent forms, and the site that started the new procedure before the ethics committee approved it because "it is obviously an improvement".

In practice

Vantis Pharma amends VT-301 to add a liver function test at each visit. Consent form version 4.0 is approved at Leah's four sites on four different dates.

At three sites the sequence is clean. At the fourth, Leah finds a participant whose liver test was drawn on 6 August but who was re-consented on 13 August.

She checks the facts before writing anything. The site's ethics approval was 1 August. Staff were trained on 4 August. The test was added to the visit worksheet on 5 August. The participant attended on 6 August, had the new test, and was re-consented at the next visit a week later because nobody had flagged that she was outstanding.

Leah classifies it as a major deviation: a study procedure performed before the participant consented to it. She confirms the site reported it to the ethics committee, and works with the coordinator on a re-consent tracker so that no participant can have the new test drawn before their row is ticked.

She also checks her other three sites for the same gap. One has two participants at risk of the same error the following week.

Finding it at one site is monitoring. Checking your other sites for it is the job.

Key takeaways

• The amendment sequence never changes: sponsor issues, regulatory authorization where required, ethics approval, site training, withdrawal of old versions, re-consent, then implementation.
• Implementing before approval is a serious deviation, and performing a newly added procedure before re-consent is a consent violation.
• A four-column version log — document, version, approved, implemented, withdrawn — prevents most amendment findings and answers an inspector's simplest question.
• Sites legitimately implement the same amendment on different dates, but every site must be able to state its own implementation date and must implement the amendment in full.
• A finding at one site should immediately be checked at your other sites, because the same gap is usually present somewhere else.