Editors can read your registry record alongside your manuscript. If the dates or the outcomes do not line up, no amount of careful writing will fix it at submission.
Why editors check registration first
Clinical trial registration is one of the few requirements an editor can check directly, without judging the science at all. The record is public, it is dated, and it states what you planned to measure.
The ICMJE Recommendations ask medical journal editors to require registration in a public trials registry at or before the time of first patient consent for enrollment, as a condition of considering the paper. The reasons are practical: to prevent selective publication and selective reporting of outcomes, to avoid duplicated research, to let patients find trials they might join, and to show ethics committees what similar work is already under way.
The ICMJE is blunt about late registration: registering at the time of manuscript submission meets none of those purposes. That is why the problem cannot be solved at the end: by the time you write the paper, the date on the record is already set.
Expect your registry entry to be read next to your manuscript. The ICMJE also puts responsibility for explaining any discrepancy between results in the registry and results in the paper on the authors, not the editors.
Does your study count as a clinical trial?
The ICMJE defines a clinical trial as any research project that prospectively assigns people or a group of people to an intervention, with or without concurrent comparison or control groups, to study the relationship between a health-related intervention and a health outcome.
Three questions settle most cases:
- Did the research team assign the intervention? If participants, or clusters such as wards or clinics, were allocated by the study rather than by routine care, the answer is yes. Randomisation is not required.
- Is the intervention health-related? The ICMJE's examples are broad: drugs, surgical procedures, devices, behavioural treatments, educational programs, dietary interventions, quality improvement interventions and process-of-care changes.
- Is at least one outcome a health outcome? This includes any biomedical or health-related measure in participants, including pharmacokinetic measures and adverse events.
Three yes answers mean the study needs prospective registration.
| Study | ICMJE registration | Note |
|---|---|---|
| Randomised trial of a drug, device or procedure | Required | The standard case |
| Single-arm intervention study | Required | The definition applies with or without a control group |
| Cluster trial of a quality improvement or process-of-care change | Required | Named in the definition |
| Pilot or feasibility trial | Required when it meets the definition | The definition does not exempt small or preliminary studies |
| Trial that assigns clinicians, measuring their patients' outcomes | Required | ICMJE FAQ |
| Trial that assigns clinicians, measuring only their knowledge or attitudes | Not required | ICMJE FAQ |
| Cohort, case-control or cross-sectional study | Encouraged, not required | Investigators do not assign the exposure |
| Secondary analysis of a registered trial | Do not register separately | Cite the parent trial's number |
If you are still unsure, the ICMJE FAQ gives one rule: err on the side of registration, or ask the editorial office of the journal you plan to submit to. The ICMJE secretariat does not review individual studies.
What "prospective" means in practice
The deadline is the first participant's consent for enrollment. Not ethics approval, not the first randomisation, and not the first data point.
Four details are easy to get wrong.
The registration date is the submission date. The ICMJE treats the date registration materials were first submitted to a registry as the date of registration. If there was a substantial delay between that submission and the record being posted, editors may ask why. ClinicalTrials.gov records the date a study was first submitted separately from the date its record first met quality control review criteria and the date it was first posted, so a long gap is visible. Keep your own confirmation of the first submission with the trial master file.
Ethics approval is not registration. The ICMJE states that approval from an ethics committee or institutional review board does not fulfil the requirement for prospective registration.
Legal deadlines are not journal deadlines. Under US regulations, the registration information for an applicable clinical trial is due no later than 21 calendar days after the first human subject is enrolled. A trial registered on day 10 can meet that rule and still fail the ICMJE test. Other jurisdictions and funders set their own rules, so check each one that applies, but plan to the strictest date.
A hidden record does not count. The ICMJE lists registrations that are not publicly accessible as inadequate. Its examples are phase I trials submitted to the EU's Clinical Trials Information System (CTIS) and device trials whose information sits in a "lock box". For device trials on ClinicalTrials.gov, the ICMJE FAQ says to opt out by selecting "Yes" for the "Post Prior to U.S. FDA Approval or Clearance" data element.
A simple start-up rule prevents the timing problems: no consent form is signed until the registry submission has gone in, the record is set to post publicly, and the confirmation is filed with the protocol.
Which registries count
The ICMJE accepts publicly accessible registration in ClinicalTrials.gov or in any primary registry of the WHO International Clinical Trials Registry Platform (ICTRP). The ICMJE endorses them because they are free for the public to access, open to all prospective registrants, managed by not-for-profit organisations, have a mechanism to ensure the registration data are valid, and are electronically searchable.
The WHO list of primary registries includes, among others, ANZCTR, ChiCTR, CRiS, CTRI, CTIS, DRKS, ISRCTN, jRCT, PACTR and ReBec. The list changes, so check it on the day you register rather than relying on a colleague's memory.
An accepted registry is necessary but not sufficient. The ICMJE considers a registration inadequate if any of the 24 items in the WHO Trial Registration Data Set is missing, or if a field contains uninformative content. Outcome fields deserve the closest look, because they are what a reviewer compares with your results. An entry that says only "efficacy" is the kind to fix: the WHO data set asks for the outcome's name, the metric or method of measurement, and the timepoint.
Two further rules from the ICMJE are worth knowing:
- If a trial is registered in a registry that is not WHO-approved, register it in a WHO-approved registry as well.
- The ICMJE FAQ asks for the minimum data items to be registered in English as well as in the registry's own language.
Where the registration number goes
Put the number in every place an editor or reviewer will look, and make it identical in each.
- End of the abstract. The ICMJE recommends that journals publish the registration number there. The CONSORT extension for abstracts asks for the registration number and the name of the register.
- Methods, or the open science section. CONSORT 2025 item 2 asks for the name of the trial registry, the identifying number with its URL, and the date of registration.
- First use of a trial acronym. The ICMJE recommends giving the registration number the first time you use an acronym, for your own trial and for any other registered trial you mention.
- The submission form. If the journal's system has a registration field, enter the same identifier.
A line you can adapt for the abstract:
Trial registration: [Registry name], [identifier]. Registered [date first submitted].
Two situations need extra care. A secondary analysis of a trial should use the primary trial's registration number, cite the primary publication, and say plainly that it reports secondary analyses. For a large trial that planned several papers from the same participants, the ICMJE says the original registration number can be used if all the outcomes were defined in the original registration. If substudies were registered separately, give the identifier for the substudy you are reporting.
A registration audit before you submit
Run this audit when the manuscript is close to final, and add it to your general manuscript submission checklist.
- Confirm the study is a clinical trial. Apply the three questions above. If the answer is yes, the rest of the audit applies.
- Compare the registration date with first consent. Take the date the registration was first submitted from the registry record, and the first consent date from your enrollment log. The submission must come before that first consent, or at the latest at the same time.
- Check the record is complete and public. Every item in the WHO data set should hold real information, and the record should be visible to anyone without logging in. For trials that began enrolling on or after 1 January 2019, the ICMJE expects a data sharing plan in the registration.
- Reconcile outcomes with the manuscript. Match primary and secondary outcomes, measures and timepoints. Where the registry history shows a change, explain in the manuscript what changed, when, and whether it happened before or after anyone saw outcome data.
- Update the registry record. Correct recruitment status and completion dates. If the data sharing plan has changed, the ICMJE asks for the change to appear in both the manuscript statement and the registry. The ICMJE also encourages a note in the registration that results are not yet published in a peer-reviewed journal, and an update with the full citation once they are.
- Place the number where editors look. Abstract, methods, first use of the acronym, and the submission form.
- Disclose any timing problem up front. If registration came after enrollment began, say so in the manuscript and the cover letter.
Under the ICMJE Recommendations, posting results in an accepted registry does not count as prior publication, as long as the posting is limited to a brief structured abstract or tables. You still need to meet any results-reporting duties set by your funder or regulator.
If registration came late
Late registration is not automatically fatal, but it is serious. The ICMJE lets editors consider whether the circumstances were likely intended to bias reporting, or resulted in biased reporting. If an editor makes an exception, the trial must be registered, the authors should state in the paper when registration was completed and why it was delayed, and the journal should publish a statement explaining why the exception was allowed. The ICMJE says such exceptions should be rare, and that authors who fail to register prospectively risk the paper being inadmissible.
If this is your situation:
- Do not hide the dates. The registry shows them to anyone who looks. State both dates in the methods and give the reason for the delay.
- Give the editor evidence to judge bias. Point to anything that fixed your outcomes before enrollment, such as a dated protocol version approved by the ethics committee, and state whether any registered outcome changed after enrollment started. This does not replace registration, but it helps the editor judge the risk of biased reporting.
- Ask before you submit. Policies on late registration differ between journals. Read the author instructions, and if they are unclear, ask the editorial office with the dates in hand.
- Say it in the cover letter. One or two factual sentences are enough. Our guide on how to write a journal cover letter covers where disclosures fit.
The better fix is upstream. The WHO data set makes the primary sponsor responsible for ensuring the trial is properly registered, and inside a research group someone still has to own the task. Registration belongs with ethics submission and consent forms in the start-up checklist that a lab lead signs off, alongside the other research integrity responsibilities of a lab lead.
The one check worth doing today
If you have a trial in progress, open its registry record now and compare the date the registration was first submitted with the first consent date in your enrollment log. That single comparison tells you whether you have a routine submission or a disclosure to prepare, and it is far easier to deal with now than after an editor finds it.
Directive Publications gathers its author guidelines and publication ethics policies on its for authors page, so you can read them before preparing a trial report. When your registration details and manuscript agree, you can submit your manuscript.
Frequently asked questions
Can a retrospectively registered trial still be published?
Sometimes, but you cannot count on it. The ICMJE says exceptions should be rare: editors consider whether the late registration was likely intended to bias reporting or resulted in it, and if they make an exception the authors must state when registration was completed and why it was delayed. Journal policies differ, so read the author instructions or ask the editorial office before you submit, and disclose the dates plainly.
Does an observational study need to be registered as a clinical trial?
Not under the ICMJE policy. The ICMJE encourages registration of observational and other non-trial designs, because the same transparency purposes apply, but does not require it, since the investigators do not assign the exposure. If the research team assigns participants to an intervention, the study is a trial whatever it is called, and it needs prospective registration.
Which trial registries do medical journals accept?
Journals that follow the ICMJE Recommendations accept ClinicalTrials.gov and any primary registry in the WHO International Clinical Trials Registry Platform network, such as ISRCTN, ANZCTR, CTRI, ChiCTR or DRKS. The record must contain the 24-item WHO trial registration data set and be publicly accessible. If your trial is registered only somewhere outside that list, register it in an accepted registry as well.
Where should the trial registration number appear in the manuscript?
Put it at the end of the abstract with the registry name, and in the main text with the registry name, identifier, URL and date of registration, as CONSORT 2025 item 2 asks. Also give the number the first time you use the trial's acronym. If the submission system has a registration field, enter the identifier exactly as it appears in the abstract.
Do pilot and feasibility studies need to be registered?
If a pilot or feasibility study prospectively assigns participants to a health-related intervention and measures health outcomes such as adverse events, it meets the ICMJE definition of a clinical trial, and that definition has no exemption for small or preliminary studies. The CONSORT extension for randomised pilot and feasibility trials asks for the pilot trial's own registration number and registry. Register the pilot before its first participant consents rather than relying on a later definitive trial's record.