Document Lifecycle

Protocol Amendment Impact Assessment: Which Documents to Check

By Kamila Kiriek, MD, PhD, Chief Medical Officer, Skaldi

7 min read

After a protocol amendment, check every document that repeats or depends on the content that changed: the informed consent form, the Investigator's Brochure, the Statistical Analysis Plan, the case report forms and data management documents, pharmacy and laboratory manuals, monitoring and safety plans, site training materials and any public registry entry. An impact assessment is the written record of that check. For each document it states whether the document is affected, why, and who decided.

This article describes how to run that assessment so that nothing is missed and so that a reviewer can later see how each conclusion was reached. It does not tell you whether a particular amendment is substantial or whether participants must be re-consented. Those are decisions for the sponsor, the investigator and the ethics committee, and the sections below say who makes them.

When an impact assessment is needed

Any change to an approved protocol starts one, including changes that look administrative. A corrected visit window, a new exclusion criterion and a changed sample-handling instruction all travel into other documents, and the size of the edit in the protocol says little about how far it reaches.

The assessment belongs before the amended protocol is finalised, not after. Three reasons:

  • Submission packages are built from it. The amended protocol usually goes to the regulatory authority and the ethics committee together with the documents it changes, most often a revised consent form.
  • Implementation depends on it. With narrow exceptions for removing an immediate hazard to participants, a change is not put into practice until the required approvals are in place. In the United States this follows from 21 CFR 312.30 for studies under an IND and from 21 CFR 56.108(a)(4) for IRB review. In the European Union, a substantial modification is authorised under Chapter III of Regulation (EU) No 536/2014. ICH E6(R3) sets the same expectation for investigators and sponsors.
  • A late discovery costs a second cycle. A consent form found to be inconsistent after approval needs its own submission, its own approval and its own site training.

What the assessment contains

A usable assessment has one row per document and the same few columns throughout:

ColumnWhat goes in it
ChangeThe protocol section, the previous wording and the new wording
Document and sectionThe exact place in another document that repeats or relies on that content
Affected?Yes, no, or needs assessment
BasisThe sentence in the source that supports the conclusion
OwnerThe function that decides and makes the change
Action and statusWhat will be done, and whether it is done

Two rules keep the table honest. A "no" needs a basis just as a "yes" does: "not affected" should mean that someone looked, and the record should show where. And "needs assessment" is a legitimate entry. It marks a question the source does not answer, which is different from a question nobody asked.

Steps

1. Describe the change itself

Compare the previous and the amended protocol section by section and list each change as a pair: what the text said, and what it says now. Work from the two approved versions, not from a summary of changes written from memory. A summary describes intent; the documents downstream were written against the actual wording.

Split compound changes. "Visit schedule updated" hides separate facts: a visit was added, a procedure was added at that visit, and the total number of blood draws changed. Each of them may land in a different document.

2. List the documents that could be affected

Start from the study's own document inventory, not from a generic list. The usual candidates are:

  • informed consent forms and other participant-facing materials, including translations
  • the Investigator's Brochure. The Investigator's Brochure is more often the source of a protocol change than a document changed by it, but it may describe the study design, visit schedule or assessments in its clinical-study section, and those descriptions need to match the amended protocol.
  • the Statistical Analysis Plan
  • case report forms, completion guidelines and the data management plan
  • randomisation and supply documents, and the pharmacy manual
  • laboratory and imaging manuals
  • the monitoring plan and the safety management plan
  • site training materials and delegation or training logs
  • clinical trial registry entries
  • contracts and budgets, where visits or procedures change

Record the version of each document you checked. An assessment against a superseded consent form is worse than none, because it looks complete.

3. Find the affected content in each document

For every change from step 1, look for the places where the same fact is stated or assumed. Search for the concept and not only for the words: participant-facing text rarely uses the protocol's wording. A protocol says "blood samples on Days 1, 8 and 29"; a consent form says "you will come to the clinic three times".

An illustration. Suppose an amendment to the Schedule of Assessments adds a clinic visit and a blood sample on Day 8, where the previous version had visits on Days 1 and 29 only. The consent form's section on what happens during the study still says that the next visit after Day 1 is on Day 29. That sentence is affected, and the basis is the amended schedule. The same amendment says nothing about how much blood is drawn at the new visit, so whether the consent form's statement on blood volume or participant burden needs to change cannot be decided from the amendment alone. That line goes into the table as "needs assessment", with the question written out.

4. Decide and record, document by document

Fill the table. Where the source settles the matter, write the conclusion and quote the basis. Where it does not, record the open question and who will answer it. Resist the two shortcuts that damage assessments most: marking a whole document "not affected" because its main sections are unchanged, and marking a document "affected" without saying where.

5. Assign owners and the order of work

Some updates can run in parallel; others wait for a decision. The consent form cannot be finalised while a question about blood volume is open. The statistical plan cannot be revised until the change to an endpoint or analysis population is fixed. Put the dependencies in the table so that the schedule reflects them.

6. Review the assessment as a document in its own right

Have the assessment reviewed by the functions that own the affected documents before work on those documents starts. It is quicker to correct one table than several revised documents, and the reviewed assessment becomes the reference against which the finished package is checked.

7. Close it out and file it

When every action is complete, confirm that each revised document says what the assessment said it would, record the final versions, and file the assessment with the amendment documentation. ICH E6(R3) lists the essential records a trial is expected to keep; the reasoning behind an amendment's implementation is part of what makes those records reconstructable.

Common mistakes

  • Working from the summary of changes. Summaries omit small edits, and small edits are the ones that surface later as inconsistencies.
  • Checking only the obvious documents. The consent form and the statistical plan get attention; the pharmacy manual, the registry entry and translated materials are found by inspection.
  • Recording conclusions without a basis. "ICF: not affected" cannot be reviewed. "ICF section on study visits: not affected; the amendment changes laboratory handling only (protocol section on sample processing)" can.
  • Treating open questions as answered. A question that the source cannot settle is closed by a named person with the missing information, not by the person filling the table.
  • Assessing once. If the amendment text changes during review, the assessment is repeated for the changed sections.
  • Forgetting documents already in the field. Sites hold printed and translated versions. The assessment should say which versions are superseded and how they are withdrawn.

What remains a specialist's decision

An impact assessment organises facts. It does not make the following decisions, and it should not appear to:

  • Whether the amendment is substantial. In the European Union the term is defined in Article 2 of Regulation (EU) No 536/2014, and the sponsor makes the classification, which the authorities may question. In the United States, 21 CFR 312.30(b) describes which protocol changes require a protocol amendment to the IND. Regulatory affairs owns this judgement.
  • Whether participants need new information or renewed consent. This is decided by the sponsor and the investigator and reviewed by the IRB or ethics committee. The assessment supplies the list of affected consent wording; it does not decide what participants must be told or when.
  • Whether the change alters the risk to participants. This is a medical judgement, made by the responsible physicians and the safety function.
  • Whether the statistical analysis changes. The study statistician decides, and the decision is recorded in the Statistical Analysis Plan and, later, in the clinical study report, where ICH E3 section 9.8 asks for changes in the conduct of the study or planned analyses to be described.

The assessment earns its place by making those decisions easier to take: the people who own them see exactly what changed, where it appears, and what is still unknown.

Sources

  • ICH E6(R3), Guideline for Good Clinical Practice.
  • ICH E3, Structure and Content of Clinical Study Reports, section 9.8.
  • 21 CFR 312.30, Protocol amendments; 21 CFR 56.108(a)(4), IRB functions and operations.
  • Regulation (EU) No 536/2014 on clinical trials on medicinal products for human use, Article 2 and Chapter III.

🔄 Skaldi connects a protocol amendment to proposed, source-linked revisions in the linked ICF, with expert review.

See the amendment example

Protocol changed. Documents in step.

When a source changes, every affected document follows. Pilots start with protocol amendments.

See how it works