By MedClinRes.org Editorial Team
A clinical trial is not one document. It is at least three: the registry entry, the protocol (and its amendments), and the published paper. Each can be revised or finalized on a different date, by a different party, for a different audience. Before leaning on a published result as a direct answer to the question a trial originally asked, it's worth checking whether that question stayed the same from registration to publication — and if it didn't, when it changed and how.
This isn't about catching anyone doing something wrong. Protocol amendments are a routine, expected part of running a trial. The practical problem is simpler: if you only read the final paper, you may be reading the answer to a question that was revised partway through, without any way of knowing that from the paper alone.
The Three Documents and What Each One Is For
Each document serves a different purpose, which is part of why they don't always move in sync with each other.
- The registry entry — the public record on ClinicalTrials.gov: a structured listing of study status, outcome measures, enrollment figures, and key dates that sponsors submit and update directly.
- The protocol (and its amendments) — the detailed document that governs how the trial is actually run: eligibility criteria, procedures, and the statistical analysis plan. When investigators change the trial's design, they amend the protocol, and that change may or may not show up in the registry right away.
- The published paper — the final write-up submitted to a journal, often appearing months or years after the trial concludes, written for a peer-review audience rather than a public registry.
Because these three documents are created and maintained on different timelines, by different processes, a reader working from just one of them has an incomplete picture by default — not because anyone is hiding anything, but because that's how the system is built.
Why a Registry Entry Can Change Without a New Paper Appearing
ClinicalTrials.gov entries are not static once posted. They're living records: sponsors can and do update fields like status, enrollment, or outcome measures as a trial proceeds, and the registry reflects that history by carrying both an original registration date and a separate date for when the record itself was last changed. Those two dates are often not close together, and treating a registry page as a snapshot from “whenever the trial started” can be misleading if the most recent edit happened much later.
For background on how a trial's status field moves through recruiting, active, and completed stages before any results appear, see our piece on how clinical trial status is reported. For how outcome measures are defined in the first place — the fields most likely to shift between registration and publication — see our explainer on primary versus secondary outcomes.
Building a Three-Document Change Log
The most reliable way to catch a mismatch between what a trial originally planned and what it eventually reported is to lay out all three documents side by side, each with its own dates, before reading the paper's conclusions. A simple spreadsheet or notes file works. The file paths below are hypothetical examples to illustrate the structure — not live documents.
- Registry entry (example path: /registry/NCT-XXXXXXXX_v-current.pdf) — record the original registration date, the date first posted publicly, and the record's most recent update date. Note any outcome measure, enrollment target, or status field that differs from the earliest version you can find.
- Protocol version (example path: /protocol/trial-name_amendment-02.pdf) — record the original protocol date, each amendment's effective date, and a one-line summary of what changed (eligibility, dosing, endpoints, analysis plan).
- Published results (example path: /results/trial-name_published-paper.pdf) — record the submission date (if stated), the publication date, and which protocol version the paper's methods section says it followed.
- Unresolved discrepancies — a running list of anything that doesn't line up: an outcome named in the paper but missing from the current registry entry, an amendment dated after enrollment was supposedly closed, or a published analysis plan that doesn't match the registry's listed outcomes.
With that log in hand, you can ask a narrower and more useful question than “is this a good study?” — namely, “does this paper's conclusion match the plan the trial was registered and amended under, or does it answer a somewhat different question than the one it started with?”
Reading Discrepancies Without Overreaching
A gap between versions is a prompt to look closer, not a verdict. Amendments happen for ordinary, often protective reasons: slow enrollment, a safety signal that narrows eligibility, a statistical review that refines the analysis plan, or a sponsor correcting an administrative error in the registry. A single amendment — or a single discrepancy in your change log — is not enough information on its own to infer bias or wrongdoing.
What a documented discrepancy does tell you is where to focus your reading. If a published primary outcome doesn't match the registry's current primary outcome field, the paper's methods and discussion sections are where that change should be explained — and a full explanation depends on the complete protocol and analysis plan, not on registry fields alone. If no explanation is there, treat that as a limit on what you can conclude from the paper, not as proof of a problem with the trial.
Why This Level of Scrutiny Exists at All
Registry reporting isn't optional for a large share of U.S. clinical research. Under the NIH Policy on Dissemination of NIH-Funded Clinical Trial Information — which applies to competing applications and contract proposals submitted on or after January 18, 2017, implementing Section 801 of the FDA Amendments Act (FDAAA 801) through the 42 CFR Part 11 final rule — NIH-funded clinical trials are expected to register and submit results information to ClinicalTrials.gov. That reporting requirement is part of why the registry functions as a running, dated record rather than a one-time announcement: it exists to be checked against, not just read once.
That also means the registry is a floor, not a ceiling. It documents that a study happened, roughly what it currently looks like, and when that listing was last touched — but it won't tell you on its own whether the design it currently shows is the one the published results actually used. That's what the change log above is for.
The Practical Takeaway
Before leaning on a published trial result to make a decision, pull the registry entry's dates, the protocol's amendment history if you can obtain it, and the paper's stated methods, and put them in one dated log. Most of the time they'll line up cleanly. When they don't, the discrepancy tells you where more reading is needed — not what conclusion to draw. See our guide to trial status fields and our explainer on primary and secondary outcomes for the two registry fields most relevant to this process.
Sources