The Data Plan Now Has to Survive Publication

A small yes-or-no question in NIH's current Data Management and Sharing instructions should be getting more attention from journal offices: will the scientific data underlying peer-reviewed publications be shared by the time of publication? NIH's writing guidance now says the 2026 DMS Plan format is required for all competing and non-competing awards: https://grants.nih.gov/policy-and-compliance/policy-topics/sharing-policies/dms/writing-dms-plan.

That question sounds as if it belongs to grant administrators. In practice, it lands on editors, production teams, data editors, repository staff, and authors at the worst possible moment: after a paper is accepted, when everyone wants the article online and no one wants to reopen the data plan.

The change is not that NIH suddenly invented data sharing. The 2023 DMS Policy remains in place. The operational shift is that the 2026 format is simpler, more explicit, and harder to hide behind general prose. It asks whether publication-linked data will be shared on time, whether repository and retention expectations are met, and whether limits on sharing have a specific ethical, legal, or technical reason. For journals, that creates a useful prompt: stop treating data availability statements as decorative paragraphs and start treating them as publication controls.

The Grant Promise Lands Late

NIH Notice NOT-OD-26-046, released February 25, 2026, updated the elements of a DMS Plan and required the new simpler format for applications submitted for due dates on or after May 25, 2026: https://grants.nih.gov/grants/guide/notice-files/NOT-OD-26-046.html. NIH said the revision was meant to clarify common confusion and aid compliance monitoring after institutes and centers reviewed more than 1,100 plans.

The plan is assessed in the funding workflow, not automatically imported into the journal workflow. NIH guidance says program staff assess DMS Plans and that peer reviewers are generally not asked to comment on them unless data sharing is integral to the project design. It also notes that the plan is stored as a separate DMS Plan document in the grant folder, with the applicant institution responsible for it along with the rest of the application.

That separation creates a gap. By the time a manuscript reaches a journal, the authors may remember that a plan exists but not the exact repository, timing, retention expectation, access-control condition, or exception language. The institution may track compliance in a grants system the journal never sees. The editor may see only a sentence saying data are available on reasonable request. Everyone can be acting in good faith while the article package still contradicts the funded project record.

Repository Choice Is Part of the Article Record

NIH's repository guidance, updated August 20, 2026, makes clear that repository choice is not a last-minute upload decision: https://grants.nih.gov/policy-and-compliance/policy-topics/sharing-policies/dms/selecting-a-data-repository. It encourages established repositories, gives priority to discipline or data-type specific repositories where available, and lists characteristics such as persistent identifiers, metadata, curation, long-term sustainability, free and timely access, clear use guidance, security, confidentiality, provenance, and retention policies.

Those characteristics are also article quality signals. A dataset with a stable accession number or DOI, usable metadata, and clear access terms is much easier for readers, reviewers, funders, and integrity teams to evaluate than a promise to email the corresponding author. The journal does not need to certify every repository. It does need a workflow that catches obvious mismatches between the data statement, the funder obligation, the repository record, and the published article.

This is especially important for journals that publish biomedical, behavioral, computational, clinical, and interdisciplinary work. The article may be one visible output of a larger funded project. The underlying scientific data may live in several places: a domain repository, a controlled-access genomic database, a generalist repository, a supplementary file, an institutional repository, or a preservation system managed by a consortium. If the journal sees only the final manuscript text, it sees the weakest version of the data story.

Make the Data Statement Earn Its Space

A useful data availability statement is not a compliance poem. It is a crosswalk between the article and the evidence readers need. For NIH-funded work, the statement should help a reader understand what data support the publication, where those data are or will be available, when access begins, what limits apply, and whom to contact when access is controlled.

That does not require a journal to publish the full grant plan. It requires a structured editorial check before acceptance. The journal can ask authors to identify NIH funding, confirm whether the work generated scientific data under the policy, list major data types, name the repository or planned repository, state whether sharing will occur by publication, and explain any ethical, legal, technical, consent, privacy, Tribal sovereignty, commercial, or security constraint that limits sharing.

• At submission, ask whether the manuscript reports NIH-supported work subject to a DMS Plan and whether the data statement is intended to satisfy that plan.
• At revision, require repository names, accession numbers, DOIs, or clear controlled-access instructions for datasets that support the main findings.
• Before acceptance, separate unresolved data deposit issues from ordinary copyediting queries so editors know what still affects publishability.
• During production, check that the HTML, PDF, XML, supplementary files, and metadata feeds point to the same data locations and restrictions.
• After publication, retain the evidence used to approve exceptions, embargoes, access controls, or statements that data cannot be shared.

The goal is not to slow every paper. It is to keep the journal from discovering, after publication, that an author promised NIH one timing rule, told peer reviewers another story, and gave readers a third version in the published data statement.

The Edge Cases Need Names

Most data-sharing failures are not dramatic. They begin with cases that were too awkward for the standard form. A clinical paper has small-cell counts that raise reidentification concerns. A genomics project needs controlled access. A qualitative study involves consent language that did not anticipate open deposit. A software-heavy article depends on intermediate files that are too large for supplement handling. A null-result study has data that authors consider uninteresting but still necessary to interpret the finding.

Journals need named exception lanes for these cases. "Available on request" should not be a default escape hatch. It should be the result of a documented judgment about why repository deposit is not appropriate, what request process exists, who controls access, what criteria apply, and how long the arrangement will be maintained. If data cannot be shared, the statement should explain the class of limitation without exposing participants, confidential agreements, security details, or other protected information.

NIH's 2024 Public Access Policy adds another timing lesson for journal teams. NIH defines the Official Date of Publication as the date the Final Published Article is first available in final edited form, whether online or in print, and the policy requires in-scope Author Accepted Manuscripts to be submitted to PubMed Central upon acceptance for public availability without embargo upon that official date: https://grants.nih.gov/grants/guide/notice-files/NOT-OD-25-047.html. In other words, article timing is now a compliance trigger in multiple workflows, not merely a production milestone.

What Journal Leaders Should Change This Month

Start with author instructions. If they still describe data sharing as a broad value, rewrite them around the decisions authors must make before acceptance. Tell NIH-funded authors that the journal expects the data availability statement, repository deposit, access controls, and publication timing to be consistent with the applicable DMS Plan. Tell editors which unresolved data questions should block acceptance and which can move safely to production.

Next, audit ten recent articles with NIH funding acknowledgements. Compare the funding statement, data availability statement, supplementary files, repository links, dataset identifiers, publication date, and any correction history. Look for vague requests, broken links, missing accessions, statements that promise future deposit with no date, and data locations that appear in the PDF but not in HTML or metadata. The sample will show whether the problem is policy language, system design, staff ownership, or author behavior.

Finally, assign ownership. Data availability cannot sit nowhere between editorial, production, research integrity, and author support. A small journal may need only a checklist and a named editor. A larger publisher may need data editors, repository integrations, exception review, and dashboards for funder-linked manuscripts. Either way, someone must be able to answer whether a paper can be accepted today without creating a compliance problem tomorrow.

Practical Takeaway for Journal Leaders

Treat the NIH DMS Plan as upstream evidence that should survive into the article package. Do not ask authors for a polished data statement only after acceptance. Ask earlier whether the work is subject to a plan, what data underlie the publication, where those data will live, when access begins, what limits apply, and how the published article will make that record visible.

The journals that handle this well will not be the ones with the longest policy pages. They will be the ones whose submission, acceptance, production, and correction workflows all tell the same data story before the article becomes part of the public scholarly record.