During an institutional review board (IRB) meeting, any member who may have a potential conflict of interest (COI) with a study under review should disclose the conflict to the convened board, withdraw from both deliberation and voting on that protocol, and leave the room during the final discussion and vote. The member may answer factual questions before withdrawing, but must not participate in the decision. Federal rules require this recusal, and the meeting minutes must record it.
The mechanism: why recusal is structural, not optional
The IRB's authority rests on independent judgment. A conflicted member cannot supply that judgment, regardless of personal integrity. The federal regulation at 45 CFR 46.107(e) states that no IRB member may participate in the initial or continuing review of any project in which the member has a conflicting interest, except to provide information requested by the IRB. The FDA's parallel rule at 21 CFR 56.107(e) uses nearly identical language for drug, device, and biologic reviews. This is a bright-line rule, not a balancing test.
The mechanism is institutional, not personal. An IRB is a deliberative body whose legitimacy depends on the appearance and reality of impartiality. If a member with a financial stake in a sponsor, a supervisory role over the investigator, or a competing research program helps shape approval conditions, the board's decision becomes vulnerable to challenge. Recusal removes that vulnerability at the source.
Two procedural steps follow from the rule. First, the member discloses the conflict before discussion of that protocol begins, typically to the chair and the convened board. Second, the member physically leaves the room during deliberation and voting so that their presence cannot influence others. The minutes then document the conflict, the recusal, and the vote tally on that protocol and on any affected quorum determination.
The word "potential" matters. It doesn't require proof of actual bias. A reasonable appearance of a conflicting interest triggers the same disclosure and withdrawal obligation (45 CFR 46.107(e); 21 CFR 56.107(e)). Waiting for certainty before disclosing defeats the rule's purpose.
Quantified context: how real boards look in practice
Most IRBs are small. The federal minimum is five members with varying backgrounds, at least one nonscientist, and at least one member unaffiliated with the institution (45 CFR 46.107). The 2020 AAHRPP accreditation standards, which many U.S. boards adopt voluntarily, require written COI policies covering financial interests, institutional roles, and investigator relationships. AAHRPP also requires documented training and defined disclosure timelines, and pre-meeting conflicts of interest.
Academic medical centers frequently use convened panels of roughly 10 to 20 voting members, so a single recusal can shift quorum math. If a quorum requires a majority of members, losing one or two members to recusal on a multi-site protocol can stall a vote until alternates are seated. That is why experienced chairs screen agendas for conflicts before the meeting, not during it.
The regulatory consequence is concrete. An approval invalidated by an undisclosed conflict can force re-review, delay enrollment, and jeopardize federal funding through Office for Human Research Protections (OHRP) compliance actions. The cost of a five-minute recusal is trivial by comparison. OHRP determinations and FDA inspection findings commonly cite failure to follow written IRB procedures, including COI handling, as a repeat deficiency.
Edge cases: where simple recusal rules break down
Conflicts of interest are not only financial. A member may supervise the principal investigator, share a department with a co-investigator, or hold equity in a sponsor. Intangible conflicts count. An investigator who is also an IRB member presents a common case: that member must recuse from reviewing their own study entirely, even though they can answer questions.
The rules diverge on one detail. The FDA's member conflict rule allows a conflicted member to remain to provide information requested by the IRB (21 CFR 56.107(e)); the Common Rule language is similar, permitting participation only to supply requested information. In practice, chairs should let the member answer factual questions, then ask them to leave before deliberation and voting. In a public meeting, particularly under FDA rules that permit nonvoting industry representatives to attend, the conflicted member should still not vote or lobby.
Consultants present a separate trap. An IRB may invite a consultant with special expertise to assist review, but that consultant must also be screened for conflicts and excluded from voting; the 21 CFR 56.107(f) pathway covers this narrowly. A consultant with a stake in the outcome cannot be used to launder a conflicted vote.
Proportionality is a real disagreement among chairs. Some boards require recusal for any consultant relationship with a sponsor in the last 12 months; others use dollar thresholds. The regulatory floor is qualitative (no participation), but institutional policies often set stricter dollar trigger points. When a chair applies a looser threshold than the institution's written policy, the chair becomes the variance.
The conflicted member must not vote, and must leave during deliberation. Failing to disclose a known potential conflict is itself a compliance finding. A consultant or guest with a sponsor relationship must be pre-screened. Minutes must record every disclosed conflict and its resolution. "Potential" conflicts trigger the same duty as actual conflicts.
Misconceptions that undermine compliance
Misconception one: abstaining is enough. Abstaining from the vote while remaining in the room still violates the rule because the member can shape deliberation. Recusal requires physical withdrawal during discussion.
Misconception two: financial conflicts are the only conflicts. Institutional, academic, and personal conflicts count. A member whose department competes for the same grant has a conflict even with no money in the study.
Misconception three: the conflicted member can stay as a silent resource. The exception in the regulations is narrow and explicit: the member may provide information only when the IRB requests it. Proactive advocacy exceeds that exception.
Misconception four: the chair decides and carries the burden alone. Every member has an affirmative duty to disclose their own potential conflict. A member who conceals a conflict and lets the board decide has compromised the meeting, not protected it.
Misconception five: recusal weakens review quality. The opposite is true. A clean decision survives audit, appeal, and institutional review; a contaminated one does not.
Recusal in practice: a working procedure
The practical flow is short and repeatable.
Pre-meeting: the coordinator circulates the agenda to voting members with enough lead time for self-screening, commonly 3 to 7 days. Each member reviews protocols against a personal disclosure checklist covering sponsor relationships, equity, supervisory roles, and competing studies.
At the start of the meeting: the chair reads conflicts into the record for each protocol on the agenda. Any member with a potential conflict states it aloud and confirms the recusal plan.
For each conflicted protocol: the member answers any requested factual questions, then exits. The chair records the departure, the reason, and any adjusted quorum.
After the vote: the minutes list the excluded member and confirm they were absent for both deliberation and vote. A brief, dated entry is sufficient.
Audit trail: files are retained with the protocol record for the institution's retention period.
One counterintuitive detail: recusal is a member duty, not a discretionary courtesy. Chairs who "manage" conflicts informally without minutes create audit exposure even when the vote was clean. The record is the protection.
A comparison: available recusal pathways
| Approach | What it permits | Where it stops | Written record | Best use |
|---|---|---|---|---|
| Full recusal (leave room) | Answering requested factual questions only | No deliberation, no vote | Minutes note conflict and absence | Default for members with conflicting interests |
| Abstain but remain | Vote withheld | Member still present in deliberation | Vote tally only | Not compliant with IRB conflict rules |
| Use a nonvoting consultant | Expert input | No vote, screened for conflicts | Consultant record in file | Narrow technical questions |
| Seat an alternate | Quorum preserved | Alternate checked for own conflicts | Alterate appointment noted | Avoiding quorum loss |
| Chair filters agenda | Prevents conflicts from reaching the floor | Only works with timely disclosures | Agenda screening record | Routine volume management |
| Institutional COI committee referral | Separate institutional review | Not an IRB vote mechanism | Referral memo | Research misconduct and institutional COI |
A timeline: how a conflict is handled during a meeting
| Time point | Action | Who acts | Record produced |
|---|---|---|---|
| Agenda circulation (3-7 days before) | Members self-screen against disclosure checklist | Each voting member | Signed screening sheet |
| Meeting opening | Chair reads standing conflicts for each protocol | Chair | Draft minutes line |
| Raising a specific potential conflict | Member discloses conflict before discussion | Conflicted member | Disclosure entry |
| Factual questions | Conflicted member answers only if the IRB requests information | Chair and board | Q&A note |
| Final deliberation and vote | Conflicted member leaves room; quorum recalculated | Chair | Recusal notation and vote tally |
| Post-meeting minutes | Conflicts and recusals recorded in final minutes | Coordinator | Final minutes |
Practical actions and thresholds
Treat the conflict rule as a pre-meeting step, not an in-meeting discovery. Screen every agenda item 3 to 7 days out. Use a short written checklist: sponsor funding within the past 12 months, equity or stock options, royalties, consulting fees, supervisory control over the investigator, and a competing study on the same question.
Define trigger points in institutional policy and apply them consistently. A common threshold in human-subjects COI policies is reporting financial interests of USD 5,000 or more from a sponsor, plus any equity of any value; exact figures vary by institutional policy rather than federal mandate. Reconcile policy and meeting practice explicitly: if the institution's written threshold is lower than the chair's informal one, the written policy controls, and any variance should be documented, with the chair's judgment memorialized in the minutes.
Record every recusal, even when the member self-discloses a non-required borderline interest. The minutes should identify the protocol, the general type of conflict, the recusal, and quorum impact without dwelling on personal details. Brief, factual entries survive audit better than long narratives.
When a conflicted member would answer a question, the chair should ask whether the information is actually needed. If it is not, the member should leave immediately, because the requested-information exception is narrow.
For immediate procedural questions, the institution's IRB office, the research integrity officer, or counsel for the human research protection program can confirm local policy.
Limitations
Regulatory text is clear on the duty, but institutional policies differ on thresholds, reporting frequency, and training intervals. Details of quorum calculations and alternate-member seating depend on each institution's written procedures. This article describes general institutional mechanisms, not the requirements of any particular board.
This article is general educational information, not a legal opinion, compliance determination, or review of any specific IRB or protocol. Conflict-of-interest rules depend on institutional policy and the applicable regulations, including differences between the Common Rule and FDA rules. Consult your institution's human research protection program or a qualified professional for guidance on a specific study.