What has to appear, and where
The registry itself has to qualify. ICMJE accepts registries that are WHO ICTRP primary registers or ClinicalTrials.gov, and asks that a registry be publicly accessible, include the minimum twenty-four item dataset, and be electronically searchable.1 A registry that fails any of those three tests does not satisfy the requirement, however carefully the trial was recorded in it.
Then the number has to appear in the paper. Journals publish the trial registration number at the end of the abstract, and authors should give the number when first mentioning the trial acronym in the manuscript.1
Both of those are checkable in a minute by someone who knows nothing about the trial, and PLOS ONE lists the same expectation among its own submission requirements.2 The number is present at the end of the abstract or it is not. It appears with the first use of the acronym or it does not. It matches the registry record or it does not.
The data sharing statement has five required parts
ICMJE sets out what a data sharing statement must specify. Whether individual participant data will be shared. Which specific data will be shared. What supplementary documents will be available. When the data will become available. And by what access criteria, including who may request the data, for what kinds of analysis, and by what mechanism.1
One answer is ruled out. Undecided is not acceptable as a response on whether data will be shared.1
That is a five-part specification, and a statement missing a part is incomplete in a way anyone can see. It is also a statement that has to agree with what the paper says elsewhere. That second test is where it usually fails, and it fails quietly.
The three places it disagrees with itself
A data sharing statement written at submission has to hold against three other parts of the same manuscript. Each of the three was written by somebody with a different concern, and often at a different time.
The methods section describes what data exist. A statement offering to share something the methods do not describe is a promise about a dataset the paper has not accounted for.
The ethics statement describes what may be shared. A statement offering unrestricted sharing of data collected under a consent that did not contemplate it is a contradiction between two parts of one submission.
The journal's own policy describes what is acceptable. PLOS ONE, for instance, requires that a restricted statement name a data access committee, an ethics committee, or another institutional body. It also states that it is not acceptable for an author to be the sole named individual responsible for ensuring access.3
Checking those three against the statement is copy editing work, covering correctness, consistency, and completeness.4 It does not require the field, and it does require somebody to sit down with four sections at once.
Where the author's authority begins
Whether a trial was registered in time, whether data can be shared, what a consent permits, and which registry record is correct are the investigators' determinations and their institution's. An editor who adjusted any of them would be making a representation to a journal on the authors' behalf.
What an editor supplies is the reconciliation and the list of gaps. Registration number present and consistent. Data sharing statement complete against the five required parts. Statement, methods, and ethics section in agreement. Journal policy satisfied on its own terms.
Everything a client sends is treated in confidence, including protocols, unpublished data, manuscripts under review, and correspondence with a registry, a journal, or an ethics committee. We work on medical and healthcare documents and journal articles against whatever the destination publishes.
Investigators who draft the data sharing statement when the protocol is written, rather than when the paper is submitted, generally have less to reconcile. The questions are the same ones. Only the amount of memory required to answer them changes.