NIH Just Deleted the Prior-Approval Step for Changing Your Data Sharing Plan. The Compliance Risk Moved — It Did Not Shrink.

For three years, changing an approved NIH Data Management and Sharing Plan meant filing a prior approval request in eRA Commons and waiting for a program officer to act on it. Switch repositories mid-project, push a sharing timeline, pivot the data types your aims actually generate — each one was a request, a queue, and a delay.

As of October 1, 2026, that step is gone. NIH has eliminated the prior approval requirement for modifying an approved DMS Plan and is retiring the DMS Prior Approval Request type in the eRA Commons Prior Approval Module altogether. The change arrived in NOT-OD-26-100, released July 29, 2026, under the agency's administrative burden reduction banner.

It is real relief. It is also a transfer of risk from a transaction NIH used to review in real time to a single field in an annual report that nobody reviews until something goes wrong.

What Replaced It

Changes to an approved DMS Plan now get reported in the Research Performance Progress Report, specifically Section C.5.c. The reportable categories NIH names are the ones that matter in practice:

• Changes in scientific direction that alter what data the project generates
• Changes to the data repositories you will use
• Changes to data sharing timelines

Section C.5 as a whole — introduced with the FORMS-I package — asks for the data types identified in your approved plan, the status of your sharing activities, the names of the repositories in use, and unique identifiers for data you have shared.

That last item is where most labs quietly fail. Naming a repository is easy. Producing accession numbers, DOIs, or dataset identifiers for data you claimed would be shared by now is a different exercise, and it is the field that makes a progress report falsifiable.

The requirement applies to all awards subject to the DMS Policy, with two carve-outs: training and fellowship awards, and a limited set of programs identified in their own notices of funding opportunity. Career development K awardees are explicitly required to comply — a point worth flagging given the separate wave of K award consolidation under NOT-OD-26-098, which has K applicants rereading their obligations from scratch this fall.

October 1 Is Doing Two Jobs at Once

The same date carries a second transition that is easy to conflate with the first, and the two have different consequences.

NOT-OD-26-046 introduced a revised DMS Plan format earlier in 2026. That format became mandatory for competing applications submitted on or after May 25, 2026. Applications submitted before that date and not yet awarded update their plans during Just-in-Time. Previously awarded projects transition to the revised format in their next progress report, and the practical deadline for that migration runs through the FY2027 RPPR cycle — meaning every active award generating scientific data should be carrying a plan in the current format by then.

So October 1 is simultaneously the date the prior approval gate disappeared and the date the format transition became fully operative. A principal investigator who hears only "NIH dropped a requirement" and stops reading will file an FY2027 RPPR with a legacy-format plan and an empty C.5.c, which is the exact combination that draws a reviewer's attention.

The format itself was already simplified earlier in the year into a shorter, more structured instrument — covered in our breakdown of the revised DMS Plan format. Shorter does not mean less binding. A two-page plan with six structured answers is substantially easier for a program officer to compare against an RPPR than a four-page narrative ever was.

The Enforcement Surface Did Not Go Away

NIH has been direct about what happens when DMS activities are not reported. The consequences on the table are special award conditions, enforcement actions up to and including award termination, and noncompliance being factored into future funding decisions.

Hold that next to the mechanics of what just changed, and the risk profile is clear.

Under the old system, a mid-project repository change produced a dated artifact: a request, a response, a record in eRA Commons that the change was known and accepted. The PI had a receipt. Under the new system, the same change sits unrecorded in the federal file until the next annual report — potentially ten or eleven months — and the only evidence that NIH ever agreed to it is a paragraph the PI wrote in Section C.5.c.

Nothing about that is improper. But it means the burden of proof shifted onto the recipient, in a policy area where the broader 2 CFR Part 200 rewrite has been pushing agencies toward faster, less reviewable termination authority. The prior approval record was, incidentally, a defensive document. Labs are now responsible for generating their own equivalent.

What a Lab Should Actually Do This Month

The useful response is small and administrative, which is why it tends not to get done.

Start a dated DMS change log now, one per award. Not a compliance binder — a single file with four columns: date, what changed, why, and what was already shared under the prior arrangement. When the RPPR comes due, Section C.5.c writes itself from that file, with dates that predate the report rather than being reconstructed from memory in the final week.

Reconcile repositories against reality before you write the report, not during. If the plan says a domain repository and the data went to a generalist repository, or went nowhere, that is the single most common gap. Fix it or explain it — the explanation is acceptable, the silence is not.

Collect identifiers continuously. Accession numbers and DOIs are trivially retrievable the week of deposit and genuinely painful to reconstruct eighteen months later across a lab that has turned over two graduate students.

Flag the scientific-direction cases early. A change in scientific direction that alters what data a project produces is a reportable DMS change, and it is often also a change in scope — which still requires prior approval under separate authority. NIH removed the prior approval requirement for the DMS plan specifically, not for everything the underlying change might trigger. Treating the two as one is the most likely way a PI uses this flexibility into an actual problem.

Check whether your institution kept its own internal gate. Many research offices built local DMS change review on top of the federal prior approval step. Some are retiring theirs in parallel; others are keeping it precisely because the federal receipt disappeared. Find out which, before you assume the change is as frictionless on your campus as it is at NIH.

The Pattern Behind the Notice

NOT-OD-26-100 did not arrive alone. It landed in the same stretch as NOT-OD-26-094's elimination of letters of support in favor of standardized letters of collaboration, and NOT-OD-26-101's high-risk life sciences framework, which pushes substantial new review obligations onto institutions starting in January 2027.

Read together, the direction is consistent: NIH is removing transactional steps it used to perform and relocating the corresponding responsibility to the recipient institution. Fewer forms to file, more records to keep, and an enforcement posture that assumes you kept them.

Investigators who treat this as one less thing to do will be right for about eleven months. The ones who treat it as a documentation obligation they now own will have a dated, defensible file when an RPPR question arrives — which, in a year when terminations have become an ordinary administrative action rather than a rare one, is worth more than the hours the old prior approval queue used to cost.

Granted's research tools track NIH policy notices alongside the opportunities they govern, so a change to how you report on an award shows up in the same place as the deadline for your next one.