FDA Advisory Committees
vs. Expert Panels
A structural analysis of what FDA's newer expert-panel format gains in speed, what it removes from the older advisory process, and where accountability thins.
FDA built a faster, lighter channel for outside expert input without eliminating its traditional advisory committees. The structural question is not whether speed is legitimate. It is what checks disappeared when the process got faster, and whether anything equivalent replaced them.
This analysis applies two Toward Reconstitution instruments, the Answerability Index and the Taxonomy of Structural Harm, to FDA's introduction of single-event expert panels alongside its FACA-governed advisory committee system. It reads the shift on its own terms as a genuine effort to broaden and accelerate expert input, then asks what happens when a thinner process begins to stand beside a more formal one.
The analysis concerns institutional process, not the scientific merits of any specific drug, treatment, or regulatory outcome. The instruments make visible where power sits, how people enter the process, whether feedback can alter outcomes, who must answer for conflicts or failure, and whether a consequential judgment can be repaired once it has moved downstream.
What changed inside FDA's review process?
FDA's traditional advisory committees operate under the Federal Advisory Committee Act and a broader framework of federal law, regulation, and agency guidance. Members hold continuing appointments, undergo formal conflict-of-interest review, deliberate collectively, and are often asked to vote on questions FDA places before them. Meetings are announced in advance and include a structured Open Public Hearing through which members of the public can speak directly into the proceeding.
The expert-panel format works differently. Participants are invited for a single event rather than serving as continuing committee members. The sessions are shorter, structured as sequential presentations or roundtable discussion, and do not culminate in a comparable collective vote. FDA has not treated them as FACA advisory committees. The public can watch livestreams and, in some cases, comment afterward, but there is no standard Open Public Hearing built into the event itself.
The shift is consequential because it has occurred while use of the older format has fallen sharply. The underlying analysis cites a 72 percent decline in drug-related advisory committee meetings between 2024 and 2025 and a separate 63 percent reduction in advisory committee meeting days. FDA has not abolished the traditional system. The structural change is that the agency now has discretion to choose between two processes with materially different levels of screening, deliberation, public participation, and recorded collective judgment.
Who is accountable for the judgment that reaches the public?
An advisory process answers that question before it answers any scientific one: through what mechanism can the judgment be checked before it becomes policy?
The traditional committee system built part of that answer into procedure itself: conflict screening before participation, public hearing before a vote, a standing roster the public can track, and a recorded collective judgment at the end. FDA still makes the final regulatory decision, but some of the checking occurs before that authority is exercised.
The expert-panel format answers the same question differently. FDA selects who is invited, what topic is discussed, how long the event lasts, and what happens with the resulting input. The agency retains final authority under either arrangement. What changes is how much of the checking that once happened before the agency acted remains structurally required at all.
What shape does the expert-panel format create?
The Answerability Index reads eight functions from dominion logic at the center to relational logic at the rim. The point is not the average. The shape shows which functions the newer format strengthens, which it thins, and what those gaps allow the process to do.
Eight functions
Select a function to see what the score is anchored to in the newer format.
The format adapts faster than it repairs. Its flexibility is real, but the checking mechanisms around that flexibility are much thinner.
The highest and lowest spokes
Adaptation scores highest at 5. FDA did build something new rather than force every question through the same older process. That is real institutional flexibility.
Agency and Repair tie for lowest at 2. Participants can speak without producing a defined collective judgment, and the format contains no built-in way to formally revisit a panel's composition or finding once the event has ended.
Innovation can make removed safeguards feel like a reasonable trade
A process that can point to its own speed and flexibility can make the absence of screening, public hearing, or formal vote look like the price of modernization rather than a structural loss. The high spoke does not erase those losses. It helps explain why they become easier to tolerate.
Feedback sits above Accountability
Feedback scores 4 while Accountability scores 2.5. The public can watch the panels and may sometimes comment through a docket. But the mechanism that would make a participant answerable for a financial conflict is weaker in the newer format. The system can receive signal while thinning the structure that checks who produced it.
The process can evolve while an earlier judgment stays untouched
Adaptation at 5 outpaces Repair at 2. FDA can revise the expert-panel format in response to criticism going forward. What the format lacks is a comparable path back to a panel already found to be thinly composed or to a consequential judgment that already moved downstream.
Who scored it matters. This shape was scored from FDA's public record and independent reporting, not from a patient affected by a labeling change or from a clinician who objected to a panel's composition. The distance between those scorings would itself be a finding.
What does the faster process still allow?
The Taxonomy of Structural Harm names six mechanisms through which ordinary, fair-looking procedure can still produce or sustain harm. Five of the six find room to operate here. The point is not that speed itself is harmful. It is that several functions removed in the name of speed were not replaced with lighter-weight equivalents.
A panel can count as legitimate without measuring whether its expertise fits the problem
Diagnostic question: What can this system's unit of account not see?
If the unit of legitimacy is simply that a panel occurred, was livestreamed, and included named experts, it cannot register whether the panel's composition matched the topic. A ten-person panel with only one specialist in the population most directly affected still satisfies that thinner ruler.
This analysis does not locate the main risk here in a cliff mechanism
Diagnostic question: Where does improving your position cost you?
The other five mechanisms carry the structural concern in this format. The analysis does not identify a separate, material Cliff mechanism as the source of the accountability problem.
A thin front-end process has no thicker downstream process to catch what it missed
Diagnostic question: Does being in trouble generate more of it?
Once a short panel without a formal collective vote feeds into a real regulatory decision, there is no comparable public process built into the format for revisiting the judgment if the panel's balance, evidence review, or composition is later contested.
Entry depends on being selected by the agency
Diagnostic question: Does entry require prior entry?
Traditional committee membership is built around standing rosters, nominations, and published membership structures. Expert-panel participation is by direct invitation for a single event, with no comparable published pathway explaining how someone becomes eligible to be invited.
Civil society performs checks the process no longer performs before the meeting
Diagnostic question: What unlogged labor holds this up?
The organizations that reviewed the SSRI-pregnancy panel's composition and formally objected afterward performed, unpaid and after the fact, part of the balance-and-expertise check that a more structured process would ideally perform before the proceeding.
The same conflict is treated differently depending on which format FDA chooses
Diagnostic question: Do identical acts get judged by who performed them?
Financial interests are handled through mandatory disclosure and review under the traditional advisory-committee framework. In the expert-panel format, comparable interests may depend on voluntary self-disclosure. The same underlying fact is subject to materially different scrutiny based on the process the agency selected.
Speed is not the harm.
The structural problem is what speed quietly removed: conflict screening, guaranteed public hearing, and formal collective judgment, without replacing those functions with something calibrated to the newer format's pace.
Which relationship does the shift actually improve?
FDA ↔ outside expertise
The expert-panel format lets the agency convene people quickly, address broad or emerging questions, and add a second channel of outside input rather than requiring every subject to move through the same longer process.
Public ↔ regulatory judgment
The people who live with the consequences of FDA decisions can see the discussion, but the newer format gives them less assurance that participants were screened, that the relevant expertise was balanced, that the public could speak inside the proceeding, or that the group's judgment was formally recorded.
The traditional advisory committee system did not disappear, and the analysis does not argue that every question belongs there. The new format responds to a real institutional problem: some questions may benefit from a faster, more flexible way of convening expertise.
But speed and accountability are different axes. The shift moved decisively on the first without building an equivalent architecture on the second. The relation improved most clearly is FDA's access to rapid expert input. The relation left closer to reform than repair is the one between the public affected by regulatory judgments and the process that makes those judgments trustworthy before they become consequential.
What did the new format remove without replacing?
No lighter-weight conflict screening
The newer format does not substitute a faster conflict-of-interest review calibrated to a two-hour panel. The heavier screening process is simply absent.
No structured public hearing
Livestreaming lets the public observe. It does not replace a guaranteed procedural opportunity to speak into the proceeding itself.
No built-in path to reconsideration
Once a panel ends, the format contains no comparable mechanism for formally revisiting a disputed composition, collective judgment, or downstream decision.
The expert-panel format adds voices and may surface questions that otherwise would not receive formal attention. What it does not carry forward is a substitute for the accountability functions it removes. Public trust in a regulatory judgment depends on more than seeing that a discussion occurred. It depends on being able to see how the process was screened, balanced, heard, recorded, and made answerable before its conclusions became consequential.
What would stronger answerability require?
Is FDA developing any conflict-of-interest screening specific to the expert-panel format, rather than relying on voluntary self-disclosure by participants?
Following criticism of the SSRI-pregnancy panel's composition, what standard, if any, now governs how panelists are matched to the specific population a topic affects?
Is there a mechanism by which a panel already found to be thinly composed, or a labeling decision that followed from one, can be formally revisited, distinct from convening an entirely new panel?
Scope
This page analyzes FDA's expert-panel format, introduced in May 2025, alongside the agency's traditional FACA-governed advisory committee system.
It is a structural analysis of process and accountability. It does not assess the scientific merits of any particular treatment, medical claim, or regulatory decision.
Instruments used
The Answerability Index maps how eight functions are organized across a system: Power, Access, Feedback, Accountability, Agency, Care, Repair, and Adaptation.
The Taxonomy of Structural Harm traces six mechanisms through which harm can be produced or sustained by ordinary process: the Ruler, Cliff, Trap, Door, Ledger, and Scale.
Primary and supporting material
Advisory Committee laws, regulations & guidance →
FDA SSRI & pregnancy expert panel →
114-organization open letter →
Analysis by Melissa Cosgrove for Toward Reconstitution. The Answerability Index™ and the Dynamic Fractal Theorem of Relation © 2026 Melissa Cosgrove. This analysis is an interpretive framework for structural evaluation and should not be treated as medical or legal advice, or as evidence in any formal proceeding.