FDA single IRB requirement: what a research site prepares

FDA's single IRB requirement is not in force yet, but the date is close enough to plan against. The latest Unified Agenda lists the final rule for December 2026, and FDA proposed that it take effect one year after publication. Once it does, most FDA-regulated multisite studies first approved after that date will be reviewed by one IRB of record instead of one per site. For a research site with its own local IRB, three things change: the local IRB stops approving those studies, the site writes and keeps a reliance record for each, and the site becomes responsible for sending local context to an IRB that has never seen its building. None of it applies to studies already open.

In short

• December 2026 is the final rule date on the latest Unified Agenda for RIN 0910-AI08; the Spring 2025 agenda had January 2026 (OIRA).
• The rule would take effect one year after publication and apply only to studies initially approved on or after that date (FDA, 2022).
• FDA proposed four exceptions, including IND-exempt drug studies and exempt device studies (FDA, 2022).
• FDA estimated 15 hours per reliance agreement, 10 agreements per IRB and 378,000 hours a year across 2,520 IRBs (FDA, 2022).

When does the FDA single IRB requirement take effect?

If the rule publishes in December 2026 as listed, it takes effect in December 2027. The rule was proposed on September 28, 2022 at 87 FR 58752, and its comment period closed on December 28, 2022 after an extension. The Office of Information and Regulatory Affairs lists it at the final rule stage with a date of "12/00/2026", meaning December 2026 with no day set. FDA proposed that the final rule "become effective 1 year after the final rule is published in the Federal Register."

Treat December 2026 as a target, not a date. The Spring 2025 agenda listed the same rule for January 2026. It has slipped before and may again. What does not slip is the arithmetic: whenever the rule publishes, a site has twelve months to get ready, and the pages still telling sites to expect it in 2025 are out of date.

What the rule actually changes

The proposal is short on the site's side, which is why it is easy to underestimate.

Scope. It applies to "FDA-regulated cooperative research initially approved by an IRB on or after the proposed effective date." A study that opened at your site before then is not pulled in. Ongoing studies may move to a single IRB but would not be required to.

Exceptions. FDA proposed four. Research where review by more than one IRB is required by law, including tribal law. Research involving "a highly specialized FDA-regulated medical product for which unique, localized expertise is required." Drug studies exempt from an IND under 21 CFR 312.2(b). Device studies that meet 812.2(b) or 812.2(c). An investigator-initiated drug study that runs under an IND exemption stays outside the rule.

A new record. Proposed 21 CFR 56.115(a)(8) requires, for research overseen by an IRB the institution does not operate, "documentation specifying the institution's reliance on the IRB for oversight of the research and the responsibilities that each entity will undertake." It can be a written agreement, an institution-wide policy or the protocol.

Local context. The proposal expects "the local site providing the single IRB of record with information on local context and updates." FDA's 2006 guidance on centralised review says the same thing from the other side: the process should include "mechanisms to ensure meaningful consideration of these relevant local factors."

Why sites that already use central IRBs are not finished

Many sites already work with a central IRB on industry studies, and NIH-funded multisite research has run under a single IRB policy since applications submitted on or after January 25, 2018, with the Common Rule's cooperative research provision applying from January 20, 2020. NIH is blunt about it: "Exceptions to use of single IRB are rare."

The FDA rule is different in two ways. It reaches studies with no NIH money and no sponsor that chose a central IRB, which includes investigator-initiated device and drug work at hospital-based sites. And it writes the reliance documentation into FDA's IRB recordkeeping rule, so for every study overseen by an outside IRB there is a required record specifying who is responsible for what. A site that cannot find its copy of that record will struggle to answer who approved what.

Seven things a site can prepare before the rule publishes

1. List your open multisite studies by IRB of record. Local, central, or another institution's. This is the baseline and it takes an afternoon.
2. Mark the ones that would fall under an exception. IND-exempt drug studies and exempt device studies stay outside the rule. Knowing which is which now saves a question at every startup later.
3. Write one local context summary for the site. Michigan state law requirements that apply to consent, your population's languages, your pharmacy and imaging arrangements, your ancillary committees. An IRB of record in another state needs this in writing, and writing it once is faster than writing it for each study.
4. Decide who signs reliance agreements. The research office, the local IRB chair, the site director. FDA estimated 15 hours per agreement; that is time someone owns, and the owner should be named before the first one arrives.
5. Keep the reliance record where the regulatory binder lives. The agreement, the institution-wide policy or the protocol section that documents reliance, filed per study. Our regulatory binder guide covers the documents that expire; the reliance record sits beside them.
6. Map the reporting lines that change. Unanticipated problems, deviations and continuing review go to the IRB of record, while the local institution may still require its own notices. Write down which goes where for each study, or a report goes to the wrong board.
7. Put amendment approvals on one clock. Under a single IRB, amendment approval comes from outside, and the site's cutover still happens inside. The protocol amendment implementation guide covers that cutover.

The mistake most sites make at step 3

They treat local context as something the central IRB will ask for. It will ask, but it asks per study, in its own form, on its own deadline, usually during startup when the coordinator is also chasing the budget and the 1572. A site with a standing local context document answers in minutes. A site without one rebuilds it from memory every time, and the answers drift between studies. An FDA inspector reading two studies' files will see the drift.

Where the time actually goes

FDA's own burden estimate is the useful number here. It assumed 2,520 IRBs, each entering about 10 reliance agreements, at 15 hours each, for 378,000 hours a year. The estimate sits on the IRB side, but every agreement has a site signature on it and a site copy in a binder. The site-side work is not the agreement itself. It is the startup tracking around it: which studies are waiting on reliance paperwork, which on the IRB of record's approval, which on the site's local ancillary reviews.

Inspection readiness changes too. The FDA inspection guide for research sites covers what investigators are cited for; under a single IRB, "where is the IRB approval for this version" becomes a question with two places to look.

What we build for this

For a site, the single IRB rule adds a dated record to every new multisite study. Our clinical research site systems include regulatory document tracking, which puts the reliance record on the same list as the licences, training and 1572s with a date and an owner, and a study startup tracker that shows which studies are waiting on reliance paperwork or the IRB of record's approval. We track and remind; the site, the IRB of record and the sponsor still decide what is approved.

The rule has slipped before. The one-year runway after publication will not.