Central vs. Local IRB: What a Reliance Agreement Moves, and the GCP Duties It Can Never Transfer

At a glance

• A central IRB is an ethics review body that is not operated by the enrolling institution; choosing one is an oversight-allocation decision, not a vendor-shopping exercise.
• For US-based cooperative (multisite) research under the Common Rule, a single IRB of record is the default mandate, with two narrow carve-outs (45 CFR 46.114). FDA-regulated multi-institution studies are permitted, not required, to use reliance arrangements (21 CFR 56.114).
• A reliance agreement reassigns who performs the ethical review. It does not transfer the investigator’s GCP duties (21 CFR 312.66) or the sponsor’s oversight duties (ICH E6(R3) 10.1).
• The investigator still owns prompt reporting to the IRB of record and still cannot enrol a single participant before that IRB’s documented approval (21 CFR 312.66; ICH E6(R3) 2.4.2).
• The sponsor still owns confirming documented IRB approval and overseeing every activity it transferred, including ethics review (ICH E6(R3) 3.8.2; 3.6.9).
• Local context, state and local law, and institutional policy do not move to the central IRB and are never preempted by signing one (21 CFR 56.103(c); 45 CFR 46.114(a)).

Teams stand up a multisite study, pick a well-known central IRB, sign a reliance agreement, and quietly assume the accountability moved with the review. It did not. A reliance agreement is a reassignment of one task, ethical review, between two bodies. It is not a transfer of the legal and GCP duties that sit on the investigator and the sponsor. This guide names who owns what after you sign, when a single IRB is mandatory versus optional versus carved out, and where the handoff actually bites in practice.

Central vs. local IRB: the one distinction that changes your workflow

A local IRB is operated by the institution where the research happens. A central IRB (the navigational searches that land here, “western irb / WCG,” “commercial irb,” “list of central irbs,” point at commercial central IRBs) is an ethics review body that is not operated by the enrolling institution. The Common Rule recognizes exactly this arrangement: it describes research at an institution “in which IRB oversight is conducted by an IRB that is not operated by the institution” and requires that the institution and the IRB document the reliance and the responsibilities each will undertake (45 CFR 46.103(e)).

That is the distinction that changes your workflow. With a local IRB, submission, approval letters, and continuing review flow through a body inside your own walls. With a central IRB of record, those same artifacts flow to and from an external body, and a written reliance agreement (or institution-wide policy directive, or a provision in the protocol) becomes the document that allocates responsibilities between you and that IRB. Everything else about who is accountable for the conduct of the trial stays where the regulations already put it.

The single-IRB (sIRB) mandate: required, optional, or carved out

Whether you may choose a central IRB or must use one depends on who funds the study and which framework governs it. The two big frameworks pull in different directions, so state the tension plainly rather than smoothing it over.

Under the Common Rule, sIRB review is a default mandate for cooperative research. Any US institution engaged in cooperative research (a project involving more than one institution) must rely upon approval by a single IRB for the portion conducted in the United States (45 CFR 46.114(b)(1)). There are two carve-outs: research for which more than single-IRB review is required by law, and research for which the supporting federal department or agency determines and documents that a single IRB is not appropriate for the particular context (45 CFR 46.114(b)(2)).

Under FDA’s IRB regulations, the posture is different. For FDA-regulated multi-institutional studies, institutions may use joint review, reliance upon another qualified IRB, or similar arrangements to avoid duplication of effort (21 CFR 56.114). That is permissive language: a reliance arrangement is allowed, not commanded. This is the tension to flag for any study under both frameworks at once: the Common Rule says you generally must use a single IRB for cooperative federally supported research, while FDA’s Part 56 only says you may. Do not reconcile these into one rule. A federally funded, FDA-regulated multisite drug trial sits under both, so you follow the stricter sIRB mandate from 45 CFR 46.114(b)(1) while reading FDA’s permission in 21 CFR 56.114 as the floor it sets, not a ceiling.

sIRB decision table

Funder / framework	Study type	sIRB status
NIH / federal department or agency (Common Rule)	US cooperative (multisite) research	Required by default (45 CFR 46.114(b)(1))
Common Rule	Cooperative research where more law requires multi-IRB review	Carved out; single IRB not the rule (45 CFR 46.114(b)(2)(i))
Common Rule	Cooperative research the agency documents as unsuitable for sIRB	Carved out by agency determination (45 CFR 46.114(b)(2)(ii))
FDA-regulated (Part 56)	Multi-institutional study	Optional / permitted, not mandated (21 CFR 56.114)
Common Rule	Single-site research	Not a cooperative project; sIRB mandate does not apply (45 CFR 46.114(a))

The table answers the most common planning question, but it does not relieve anyone of a single duty in the sections that follow.

What a reliance agreement transfers, and the four things it never does

A reliance agreement transfers the performance of ethical review to the IRB of record and documents the division of responsibilities (45 CFR 46.103(e)). Here is what it never moves.

It never moves the investigator’s assurance duty. The investigator (not the central IRB, not the CRO) must assure that an IRB meeting the regulatory requirements is responsible for initial and continuing review, must promptly report to that IRB all changes in research activity and all unanticipated problems involving risk to subjects, and must not make changes without IRB approval except to eliminate apparent immediate hazards (21 CFR 312.66). Naming a central IRB satisfies the first clause; it does not delegate the reporting that follows.

It never moves the no-enrollment-before-approval line. Before initiating the trial, the investigator or institution must hold a documented and dated approval or favourable opinion from the IRB of record covering the protocol, the consent materials, and the recruitment procedures (ICH E6(R3) 2.4.2). The trial itself must be conducted in compliance with the protocol that received prior IRB/IEC approval (ICH E6(R3) 3.1). A central IRB issues that letter, but the duty to possess it before first enrollment stays with the site.

It never moves the sponsor’s oversight. Under ICH E6(R3) 10.1, the sponsor may transfer tasks and the investigator may delegate them, but each retains overall responsibility for their respective activities. Where activities are transferred or delegated to service providers, responsibility for the conduct of the trial still resides with the sponsor or investigator (ICH E6(R3) 10.2), and the sponsor or investigator must maintain appropriate oversight of those activities (ICH E6(R3) 10.3). Ethical review handed to a commercial central IRB is exactly such a transferred activity, so the sponsor must oversee it (ICH E6(R3) 3.6.9).

It never moves the underlying legal floor. Compliance with the IRB regulations does not render inapplicable any pertinent federal, state, or local laws (21 CFR 56.103(c)), and the Common Rule’s sIRB provision opens by stating that in cooperative research each institution remains responsible for safeguarding subjects’ rights and welfare and for complying with the policy (45 CFR 46.114(a)).

Responsibility matrix: who owns what after you sign

Post-reliance duty	Owner	Governing provision
Hold documented IRB approval before first enrollment	Investigator / institution	ICH E6(R3) 2.4.2; 21 CFR 312.66
Promptly report changes and unanticipated problems to the IRB	Investigator	21 CFR 312.66
Run the trial in compliance with the IRB-approved protocol	Investigator	ICH E6(R3) 3.1
Perform initial and continuing ethical review	IRB of record (central)	45 CFR 46.103(e); 21 CFR 56.114
Confirm and retain documented IRB approval / favourable opinion	Sponsor	ICH E6(R3) 3.8.2(b)(i)
Oversee ethical review as a transferred activity	Sponsor	ICH E6(R3) 3.6.9; 10.3
Retain overall responsibility despite transfer/delegation	Sponsor and investigator, each for their own activities	ICH E6(R3) 10.1; 10.2
Safeguard subjects and comply with the policy at every institution	Each institution	45 CFR 46.114(a)

Read the matrix as the operational answer to “does using a central IRB reduce the investigator’s or sponsor’s responsibilities?” It does not. It relocates one row, the review itself, to the IRB of record. Every other row is unchanged.

Local context the central IRB cannot own

A central IRB reviews the protocol and consent materials, but it does not become the keeper of every local requirement. The IRB regulations are explicit that compliance with them does not displace pertinent federal, state, or local laws or regulations (21 CFR 56.103(c)). The Common Rule reinforces this from the institution side: in cooperative research, each participating institution remains responsible for complying with the policy and safeguarding its subjects (45 CFR 46.114(a)). Practically, that means state-specific consent language, institutional policy constraints, and local legal requirements still need an owner at or near the site, even when a national central IRB holds the approval letter. The reliance agreement should name that owner; the regulations will not name them for you.

There is also an institutional-review layer that survives a central IRB. Research approved by an IRB may still be subject to further review and approval or disapproval by officials of the institution, though those officials may not approve research the IRB has not approved (21 CFR 56.112). A central IRB’s green light is necessary but not always locally sufficient.

Choosing a central IRB: shortlist criteria beyond turnaround time

Turnaround time is the metric every central IRB markets and the one that matters least to your liability. Build the shortlist around whether the IRB lets you discharge the duties that stay yours.

• Reliance documentation discipline. Can the IRB produce the written reliance agreement that 45 CFR 46.103(e) requires, with the responsibility allocation spelled out, rather than a generic acceptance letter?
• GCP-compliant operation. The sponsor must be able to obtain a statement that the IRB is organised and operates according to GCP and the applicable regulatory requirements, plus documented initial and subsequent approvals (ICH E6(R3) 3.8.2(b)(i)). An IRB that cannot supply this on demand makes your sponsor-side duty harder.
• Reporting channels you can actually meet. Your investigators must promptly report changes and unanticipated problems to this IRB (21 CFR 312.66). Confirm the submission pathway and timelines before, not after, the first safety event.
• Continuing-review cadence. The IRB of record performs the initial and continuing review the investigator must assure under 21 CFR 312.66; make sure its continuing-review intervals fit your protocol’s risk profile.
• Local-context handling. Ask how the IRB ingests state-law and institutional-policy variations across sites, given that those obligations are not preempted by its review (21 CFR 56.103(c)).
• Documentation you can audit. You will be confirming and retaining its approval letters as a sponsor duty (ICH E6(R3) 3.8.2). Favor an IRB whose records you can pull cleanly into your trial master file.

Speed is a tiebreaker among IRBs that clear these bars, not a substitute for them.

The hybrid model and where it bites

Many multisite studies run a hybrid: a central IRB of record for most sites, with one or more institutions keeping their local IRB because state law, a funder carve-out, or institutional policy demands it. The Common Rule explicitly anticipates this through its carve-outs, where more law requires multi-IRB review or an agency documents single-IRB review as inappropriate (45 CFR 46.114(b)(2)). The hybrid bites when teams forget that the carve-out sites still owe the same investigator and sponsor duties as the central sites: documented approval before enrollment (ICH E6(R3) 2.4.2), prompt reporting (21 CFR 312.66), and sponsor oversight of each review body (ICH E6(R3) 3.6.9). Two IRBs of record means two streams of approval letters to confirm and two reporting pathways to keep prompt, not a lighter load.

Where teams get it wrong

The recurring failure is treating the reliance agreement as a liability transfer. Three patterns dominate. First, a site assumes the central IRB now “owns” deviation and unanticipated-problem reporting, so nobody files promptly, even though 21 CFR 312.66 keeps that squarely on the investigator. Second, a sponsor treats ethics review as fully outsourced and drops it from its oversight plan, contradicting ICH E6(R3) 10.1 and 3.6.9, which keep transferred activities under sponsor oversight. Third, a team enrols before the central IRB’s documented approval is in hand because “the central IRB is fast,” ignoring that ICH E6(R3) 2.4.2 requires the dated approval before initiation regardless of how quick the IRB is. The fix in all three cases is the same: write the reliance agreement to name the owner of each duty in the matrix above, then staff and audit against it.

A central IRB can make a multisite study faster and more consistent. It cannot make anyone compliant on its own, and it cannot absorb accountability that the regulations assign to the investigator and the sponsor. Software and process, including a well-run central IRB, enable compliance; the sponsor and investigator stay responsible. Sign the reliance agreement knowing exactly which duties you kept.

Sources

• ICH E6(R3) Good Clinical Practice — ICH, version r3 — https://www.ich.org/page/efficacy-guidelines
• 21 CFR Part 312 Investigational New Drug Application — FDA, version 2026-04
• 21 CFR Part 56 Institutional Review Boards — FDA, version 2024 — https://www.govinfo.gov/content/pkg/CFR-2024-title21-vol1/pdf/CFR-2024-title21-vol1-part56.pdf
• 45 CFR 46 Protection of Human Subjects (Common Rule) — HHS, version 2018 — https://www.govinfo.gov/content/pkg/CFR-2024-title45-vol1/pdf/CFR-2024-title45-vol1-part46.pdf