Report trial registration with the right study scope and actual dates

1. Determine what the researchers assigned

WHO’s registration definition concerns prospective assignment of people or groups to health-related interventions to evaluate health outcomes. It includes more than drugs and randomized designs. Read the protocol’s actual allocation and intervention procedures rather than infer the design from “clinical” in the title.

Use the following invented screening questions to prepare a discussion with the study’s responsible team. They are an initial design assessment, not a substitute for the requirements of a chosen registry, funder, ethics body or jurisdiction. Preserve ambiguous cases as questions; relabelling an intervention study “observational” does not change what was done.

Reference: WHO: clinical trials, registration questions and answers

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

2. Separate journal registration policy from other approvals

ICMJE seeks public trial registration no later than first patient consent for enrollment and uses the date registration materials were first submitted. It encourages, but does not require, registration of non-trial research; secondary trial analyses should identify the parent registration. Ethics approval does not replace registration.

Make a requirement sheet for the actual study: the journal’s current policy, applicable registry instructions and the determinations of the responsible institutional or regulatory offices. Record their scope and dates. A checked journal-policy box cannot establish compliance with every country’s requirements, and a protocol number issued locally may not be a public registry identifier.

Reference: ICMJE: clinical trial registration policy

3. Inspect the registry record beyond its identifier

The WHO data set distinguishes the primary registry and trial ID, official registration date, secondary identifiers, recruitment dates and study details. Match the record to the protocol and study team; a similar acronym is insufficient. Preserve the exact record URL and the version or retrieval date used in the manuscript check.

For the proposed review worksheet, compare the intervention, population, outcomes and time points with the manuscript. Record each agreement or discrepancy and the evidence that could resolve it. A number pasted into the abstract proves only that text was pasted; it does not show that the correct trial was located or that its registered information matches the report.

Reference: WHO: Trial Registration Data Set, version 1.3.1

4. Reconstruct an invented chronology without merging different dates

In fictional Paper A, the team retains a dated protocol, a registry submission acknowledgment and an enrollment log. Registration materials were first submitted on 3 March, the first patient consent was recorded on 5 March, and the public record was posted on 8 March. The official registration-date field has not yet been checked. No real trial or registry ID is represented.

The useful action is to preserve all these dates and verify the missing field, not replace them with the earliest convenient date. The three-day interval between consent and public posting also needs an accurate account. This teaching chronology alone does not establish acceptable registration: completeness, public access, registry policy and any editorial inquiry still require review.

On small screens, scroll the table horizontally, or focus it and use the arrow keys.

5. Handle a missing or late record as a reporting problem

For a second invented scenario, suppose the only retained registration submission acknowledgment is dated 10 March, after first consent on 5 March. The team cannot make this prospective by editing the manuscript date. Ask for any earlier evidence, preserve what is available and report the actual chronology if no earlier submission can be established.

ICMJE describes exceptions to inadequate prospective registration as rare and requiring transparent explanation and editorial judgment. The author cannot award that exception to the study. “Pending” should identify the unfinished action and responsible person, not stand in for a registration number or a completed policy review.

Reference: ICMJE: clinical trial registration policy

6. Make the abstract and full report agree

ICMJE recommends the trial registration number at the end of the abstract and at the first use of a trial acronym when a number is available. Apply the journal’s actual format using the verified identifier. Preserve its letters and digits when preparing Arabic text; translate explanatory wording, not the identifier.

Use an internal drafting marker such as “[verified identifier to be inserted]” only while the field is genuinely unresolved. Before submission, resolve it or provide an honest account of the missing record as required by the responsible editorial process. Do not leave a plausible invented number in a teaching example that could be copied into a real manuscript.

Reference: ICMJE: manuscript preparation, abstract registration information

7. Hand over a reconciliation record, not just a number

The final internal packet should contain the design decision, exact registry page, retained chronology, comparisons with the protocol and manuscript, unresolved discrepancies and the person responsible for each. Let a second team member follow the packet back to the original records. If the study is a secondary analysis, make its relationship to the parent explicit.

Separate the administrative check from scientific appraisal: a complete registration packet does not establish valid randomization, faithful conduct or reliable findings. Those require their own evidence. In the fictional examples, no registry update, participant enrollment or journal determination has actually occurred.

Questions for a practical final check

• Classify the actual design before applying a registration rule.
• Read the correct public record and preserve its identity.
• Keep submission, official registration, posting and consent dates separate.
• Record discrepancies without inventing missing fields.
• Use the same verified identifier across both languages.
• Leave policy exceptions to the responsible decision maker.