A completed trial. Clean data. A real finding. And a manuscript that cannot be published, because the trial was registered three months after the first patient was enrolled.
This happens constantly, and it is almost always a genuine oversight — the team knew registration was required and treated it as administration to be completed before submission rather than before enrolment. Under ICMJE policy, that timing difference is the entire requirement.
What registration is actually for
Registration exists to solve two specific problems, and understanding them explains why the deadline is where it is.
Publication bias. Trials with disappointing results were historically less likely to be written up and published. The evidence base therefore drifted optimistic — not through fraud, but through selective visibility. A public register created before results exist means an unpublished trial leaves a permanent, findable trace.
Outcome switching. A trial specifies a primary outcome. If it disappoints while a secondary outcome looks impressive, there is an obvious temptation to present the secondary as though it had been the target all along. A registered protocol makes this checkable: anyone can compare the registered primary outcome with the published one.
Both defences only work if registration happens before the results are known. This is why retrospective registration is not a lesser version of the same thing — it is a different thing, which provides neither protection.
What counts as a clinical trial
Broader than most authors expect. The ICMJE definition covers any study that prospectively assigns human participants to an intervention to evaluate a health outcome.
That includes:
- Drug and biologic trials
- Device and implant trials
- Surgical procedures and techniques
- Behavioural and psychological interventions
- Physiotherapy and rehabilitation protocols
- Dietary and nutritional interventions
- Educational and training interventions with a health outcome
- Diagnostic strategies where assignment affects management
Randomisation is not required by the definition. A single-arm prospective interventional study still qualifies. The common misreading — "this was not randomised, so it is not a trial" — is the second most frequent route to this problem after simple lateness.
Genuinely observational studies, in which you record what happens without assigning anything, are not required to register.
Where to register
Any primary registry in the WHO International Clinical Trials Registry Platform network, or ClinicalTrials.gov. Widely used options include:
- ClinicalTrials.gov — United States, used internationally
- CTRI — Clinical Trials Registry, India
- ISRCTN — United Kingdom, international scope
- ANZCTR — Australia and New Zealand
- ChiCTR — China
- CTIS — European Union
- IRCT — Iran; UMIN-CTR — Japan; PACTR — Pan-African
Registration is free in most registries. There is no advantage in choosing one over another beyond convenience and local requirements, and no requirement to use a registry in your own country.
An institutional list of ongoing projects is not a registry. Neither is a departmental website, an ethics committee reference number, or a funder's grant database. The registry must be publicly accessible, free to search, and part of the WHO network.
What the registration record must contain
The WHO Trial Registration Data Set specifies the minimum. Two fields deserve particular care because they are the ones later compared against your paper:
The primary outcome, stated specifically. "Improvement in symptoms" is not adequate. "Change in VAS pain score from baseline to 12 weeks" is. A vague registered outcome provides no protection against outcome switching, which is why reviewers treat it as a red flag rather than a technicality.
The enrolment start date, which is what determines whether your registration was prospective.
Also required: public and scientific title, sponsor, funding source, contact, health condition, intervention description, key eligibility criteria, study type and design, target sample size, secondary outcomes, and recruitment status. Keep the record updated as the trial proceeds — a registry entry showing "recruiting" three years after completion signals a team that registered and then forgot.
Reporting it in your manuscript
Include the registry name, the full registration number, and the registration date. Most journals want it in the abstract as well as the Methods.
This trial was registered with ClinicalTrials.gov (NCT0XXXXXXX) on 14 March 2024, before the enrolment of the first participant.
Where the trial was registered late, say so explicitly:
This trial was registered with CTRI (CTRI/20XX/XX/XXXXXX) on 2 August 2024. Enrolment began on 15 May 2024; registration was therefore retrospective, owing to [reason].
The second statement may cost you a journal. Omitting it and being discovered will cost you considerably more.
If you are already late
- Register now. A late registration is better than none, and some journals will consider retrospectively registered trials on disclosure.
- Record the dates honestly in the registry, including the true enrolment start date.
- Disclose in the cover letter before you submit. Give the editor the facts and the reason up front; do not let them find it.
- Never describe it as prospective. This is the line between an administrative failure and a misrepresentation, and the two are treated very differently.
- Check journal policy first. Policies vary from outright refusal to acceptance with a disclosure statement. Ask the editorial office before investing in a submission.
Preventing it next time
The fix is procedural rather than clever: make registration part of the ethics approval workflow, not the publication workflow.
The trigger to register is ethics approval granted — not "study finished", not "results look interesting", not "we are ready to write". Registration typically takes a few days to be processed and can usually be done while final logistics are still being arranged. There is no reason for it to be on the critical path, and every reason to have it done before the first participant is approached.
One habit worth adopting: registration is also a check on your own protocol. Being forced to state a single primary outcome in public, in advance, is a discipline that improves trials as well as documenting them.
Related reading: our guides to registered reports and study preregistration and to ethics approval and informed consent. Our requirements are set out in the Research Integrity Statement and Author Guidelines.