Amendment 3 versus amendment 2
A protocol amendment is issued as a complete new document. The summary of changes is written by the sponsor and is usually accurate about the big items: a changed dosing schedule, a new inclusion criterion. It is less reliable about the small ones: a changed visit window, a re-worded endpoint, a corrected table.
Those small items are the ones that cause protocol deviations. A site that keeps scheduling the week 12 visit within plus or minus seven days, when the amendment quietly tightened it to plus or minus three, is out of compliance without knowing it.
A worked example
Take a hypothetical 110-page protocol. Amendment 3 changes the dosing schedule in section 6, adds one inclusion criterion, and according to its summary of changes, nothing else. Comparing it against amendment 2 produces a change list of eleven entries. Two are the dosing changes. One is the new criterion. Four are in the schedule of assessments table, where a visit window narrowed and a safety lab was added at week 8. Two are a re-worded secondary endpoint. Two are a renumbered cross-reference and a corrected typo.
The table changes and the endpoint wording were not in the summary. None of this is unusual, and none of it implies anything wrong with the sponsor's process. Summaries are written by people, and people summarise what they think matters. The change list simply records everything, with the page number in both versions, so the reviewer decides what matters.
Note the document size. A protocol of that length is above the 30-page anonymous limit, so it needs a free account (up to 100 pages) or an Enterprise plan for longer documents up to 300 pages. See pricing for the tiers.
Who needs the list
- Sites, to update their procedures and re-consent participants where needed.
- Monitors, to check that the changes were implemented at each site.
- Ethics committees and IRBs, to review what they are being asked to approve.
- Sponsors and CROs, to confirm that the submitted version matches the approved version.
Each group reads the same change list differently. A site coordinator scans the schedule of assessments. A medical monitor reads the eligibility and safety sections. An ethics committee reviewer often wants the redlined PDF itself, because it shows each change in its context on the page rather than as an extract.
Informed consent forms
Consent forms are revised often and translated into several languages. Comparing each revised form against its previous version, per language, shows which paragraphs changed and therefore which participants need to be re-consented. The comparison aligns content by paragraph, so an inserted paragraph does not make every later page look changed.
The redlined PDF is often the easiest thing to hand to a site. It shows the new form with insertions and deletions marked in place, so the coordinator can see at a glance whether the change affects the sections participants actually discuss with staff, such as risks, visit burden or compensation, or only administrative wording.
One pattern worth checking for: the English master changes, and one translation is updated while another is not. Comparing each translation against its own previous version, and then reading the change lists side by side, shows quickly whether every language picked up the same change.
Edge cases
Amendments often renumber sections. Because content is aligned by paragraph rather than by section number or page, a section that moved from 8.4 to 8.5 is matched to its earlier self, and the renumbering appears as a small change rather than as a whole deleted section and a whole new one. Text moved to a different part of the protocol is reported as a move.
Signed copies that come back from a site as scans need an OCR text layer. A pure image scan is refused because there is no text to compare. With OCR, formatting differences are ranked low so they do not hide real wording changes. Running headers with the protocol number and version are recognised and excluded, so the version bump in the header is reported once instead of on all 110 pages.
Keeping the record
The simplest workflow is to file the redlined PDF alongside the amendment in the trial master file, with the change list exported from the side-by-side view. If anyone later asks what the team knew changed in amendment 3, the answer is a document, not a recollection.
Sponsors handling many studies can use the comparison API to run this automatically whenever a new protocol version is finalised, and attach the JSON change list to the document record.
Reproducibility
The comparison is deterministic. The same two PDFs always produce the same change list, which is what a regulated record requires. Enterprise deployments run inside the sponsor's or CRO's own infrastructure, with pinned versions, so documents never leave their network. The clinical research industry page describes the options.
Try it on your own documents. Upload two versions on the homepage. Documents up to 30 pages need no account, and nothing is stored.
Industry page: PDF comparison for clinical research