A specialty society president is putting together the panel that will write the next iteration of a guideline that hundreds of thousands of physicians will follow, that payers will use to set coverage policy, and that will appear as the standard of care in the next decade's malpractice arguments.
She starts, as every chair before her has started, with a mental list. The methodologist she has co-authored with twice. The academic center colleague who chaired the last version and did a good job. Two or three names from the conference circuit, people whose talks she has watched at annual meetings for a decade. She emails them. They accept. The panel is seven people deep before anyone outside her own professional circle has been considered, because there was never a step in the process where anyone outside her circle could apply.
Nobody on the resulting panel practices in a community hospital seeing forty patients with this condition a week. Nobody on it is early-career. Two of them have received industry payments related to products the guideline will evaluate, disclosed in a form nobody on the committee will cross-check against the public database that could confirm or contradict it in minutes.
This is not a failure of any individual's judgment. It is what happens, reliably, when a selection process has never included a mechanism for anyone to be considered who was not already inside the room.
The guideline panel is chosen by acquaintance, because acquaintance is the only candidate-identification mechanism that has ever existed for this job.
The financial-entanglement numbers are not ambiguous
Start with what has been directly measured, because the scale of the conflict-of-interest problem in guideline authorship is not a matter of interpretation.
Among 270 US physician authors of 2020 clinical practice guidelines, 199, or 73.7 percent, had received industry payments, with a median of $27,451 over three years, according to a 2023 study in BMJ Open. That alone would be a significant finding. What makes it a crisis of process rather than merely a crisis of money is the second number: only 4 of those 270 authors, or 1.5 percent, disclosed their conflicts accurately.
The panel chairs, the people setting the tone and the agenda for the entire guideline, were not better. 18 of 22 chairs, or 81.8 percent, had received industry payments, and none of them disclosed accurately.
This is not a story about a handful of bad actors slipping through a mostly functional system. A 98.5 percent inaccuracy rate on self-reported conflict disclosure describes a system where the disclosure form itself is not doing meaningful work, regardless of how carefully any individual guideline developer administers it.
The composition problem is the same problem wearing a different face
Run the same lens over who gets onto these panels in the first place, not just what they later fail to disclose, and the pattern repeats with striking consistency.
Across guideline panels convened by the American Urological Association and the European Association of Urology, 21.6 percent included no women at all, and voting panel members overall were 81.8 percent male and 86.8 percent white, according to a 2024 study in Urology. That figure describes not a slow drift toward better representation but a substantial share of panels that, as recently as the years studied, contained no women in any voting role whatsoever.
It is worth being precise about why this and the conflict-of-interest data are the same finding rather than two separate problems that happen to coexist. Both are downstream of an identical mechanism: a chair invites people from inside a small, pre-existing circle, and a circle built through conference attendance, prior panel service, and academic co-authorship systematically narrows toward people who look like, are paid like, and have practiced in the same settings as the people already inside it. Nobody has to intend either outcome for both to result reliably from the same closed process.
"Lack of transparency" is the field's own diagnosis
This is not an outside critique imposed on guideline development. It is close to the field's own stated assessment.
A Society of Critical Care Medicine team, reflecting on panel selection generally, wrote plainly that "processes for panel selection lack transparency." That sentence, from people who run this process themselves, is the clearest possible admission that the acquaintance-based system is not merely imperfect at the margins but structurally opaque, even to the professionals operating inside it.
Self-reported disclosure fails for a specific, fixable reason
It is worth being precise about why disclosure accuracy sits at 1.5 percent, because the failure is not primarily one of dishonesty, and understanding that changes what the fix should look like.
Self-reported conflict disclosure asks a busy physician, often years into a series of industry relationships, speaking fees, consulting arrangements, research funding, to accurately recall and characterize all of it on a form, with no independent verification step built into the process. CMS Open Payments, the federal database created by the Sunshine Act specifically to make this information public and checkable, already holds the answer for the overwhelming majority of these relationships. 826,313 of 1,445,944 US physicians, or 57.1 percent, received industry payments from 2013 to 2022, with a median payment of $48 and the top 0.1 percent averaging nearly $2 million, according to a 2024 JAMA analysis of the same database.
The tool to verify every guideline author's disclosure form against ground truth in minutes has existed publicly since the Sunshine Act took effect. Almost nobody uses it for this purpose. The 1.5 percent accuracy figure is not evidence that physicians are unusually dishonest; it is evidence that a verification step that already exists, cheaply and publicly, has simply never been built into the guideline-authorship workflow.
The counterfactual already exists, and it worked
Here is the finding that turns this from a complaint into a demonstrated alternative, and it is the single most important piece of evidence in this article.
A guideline panel convened using an intentional open-call process, publishing a defined expertise requirement in advance, issuing an open invitation to the full society membership rather than a chair's personal list, allowing voluntary diversity disclosure, and running an independent scored review of applicants, produced a 33-member panel spanning six countries that was 45 percent women and 79 percent historically excluded people and groups, including non-physician professionals and patient advocates, according to a 2024 study in Critical Care Medicine.
This did not happen by lowering the bar. It happened by widening who was allowed to apply, and then applying a defined, scored process to the resulting pool rather than a chair's personal recollection of who deserves consideration. The methodology paper itself states the problem in the same words already quoted: "processes for panel selection lack transparency." The team that wrote that sentence then went and ran a process that was, by design, transparent, and the resulting panel looked nothing like the acquaintance-selected norm.
The existence of this single documented counterexample changes the terms of the conversation for every society that has not done the same. It is no longer possible to argue that a more open, more representative panel is a nice aspiration incompatible with running a rigorous guideline process. It has been run, it is published, and it worked.
Why the guideline conflict-of-interest debate has spent fifteen years on the wrong lever
This is the structural diagnosis, and it explains why so much institutional effort has gone into disclosure rules and recusal thresholds with so little apparent effect on the underlying numbers.
The dominant reform strategy for guideline conflicts of interest, since roughly the Institute of Medicine's 2011 "Clinical Practice Guidelines We Can Trust" standards, has focused overwhelmingly on disclosure requirements and recusal thresholds: how much industry money triggers a mandatory recusal, how detailed a disclosure form needs to be, whether a chair can hold financial relationships at all. Fifteen years of that focus produced a system where 98.5 percent of disclosures are still inaccurate.
The BMJ Open data suggests the reform effort has been aimed downstream of the actual point of failure. The disclosure rule only matters once someone is already on the panel. The binding constraint sits earlier: who was ever considered a candidate in the first place. A closed, acquaintance-based candidate pool will keep producing panels with high, poorly-disclosed industry entanglement no matter how the recusal threshold is tuned, because the same small circle that generates the composition problem also generates the conflict problem, through repeated selection of people whose industry relationships accumulate over successive panel appointments.
Fix the candidate pool, verify payments automatically against a database that already exists rather than asking people to remember, and a meaningful share of the conflict problem shrinks without writing a single new recusal rule.
Why societies control both the disease and the only available cure
Run through why this has persisted despite being, in a real sense, an easy technical fix.
Societies control both the guidelines and the candidate pool that feeds them, and face no external pressure to widen either, because the people who would have to authorize an open-call process are the same incumbents who currently benefit from the closed one.
Journals check whether a disclosure form was submitted, not whether it was accurate. Editorial policy has converged on requiring disclosure, not verifying it, which is precisely the gap the 1.5 percent accuracy figure exposes.
No vendor can sell payment-verification services to an organization that would be visibly embarrassed by the results. The commercial incentive runs backward: a society has more to lose from publishing an accurate audit of its own panels than from continuing not to look too closely.
The people harmed by this, the community clinicians never considered, the early-career physicians with no path to nomination, the patients whose guideline was written by a panel that never included anyone who looked after someone like them, are diffuse and unorganized. There is no lobby demanding an open call the way there is a professional apparatus defending the status quo.
What would actually work
Publish the expertise definition and the selection criteria before recruiting begins. The Critical Care Medicine open-call panel worked in part because applicants knew, in advance, what qualified them to apply, rather than discovering after the fact that selection depended on a relationship they did not have.
Open the invitation to the full relevant membership, not a chair's list. This single change, moving from personal invitation to open call, is the mechanism that produced the 45 percent women, 79 percent historically excluded composition in the one documented case where it was tried.
Verify every disclosed conflict against CMS Open Payments automatically, not as a spot check. The database already exists, is already public, and already contains the ground truth for the overwhelming majority of industry financial relationships. There is no remaining technical excuse for a 1.5 percent accuracy rate.
Score applicants on defined criteria, independently reviewed, not by chair discretion. The open-call model's independent scored review is what made the wider invitation actually convert into a different panel rather than simply into a larger pool the chair could still select from by habit.
Include practice-setting representation as an explicit criterion, not an afterthought. A guideline written entirely by academic medical center physicians for a condition managed overwhelmingly in community settings is a mismatch this article's evidence suggests is at least as large as the demographic gap, and it is almost entirely unmeasured by current society reporting.
Track and publish recurrence. A count of how often the same individuals appear across a specialty's guideline panels over a decade would make the acquaintance pattern visible in a way no single guideline's author list currently reveals on its own.
Make retired and 55-plus physicians an explicit part of the eligible pool. Deep expertise combined with expired industry relationships is, structurally, one of the strongest conflict-of-interest profiles available, and current selection practices generally do not seek this group out.
What you can do now
If you are a community or early-career clinician
Ask your specialty society directly whether an open call exists for its next guideline panel, and if not, why not. The Critical Care Medicine example gives you a concrete, published precedent to point to rather than an abstract request.
Volunteer for committee and review work even when it is unglamorous. Prior panel and committee service is one of the few credentials the current acquaintance-based system does recognize, and it is a legitimate way to enter the circle that selection currently favors, even while that circle itself needs opening.
If you are asked to serve on a panel, ask who else was considered. A direct, non-confrontational question about the selection process is one of the few pressure points an individual invitee actually has, and chairs increasingly need to be able to answer it.
If you chair or lead a specialty society
Read the Critical Care Medicine open-call methodology before your next panel convenes. It is a published, working template, not a theoretical proposal, and it demonstrates the approach did not slow the process down in a way that compromised the guideline's quality.
Cross-check every panelist's disclosure against CMS Open Payments before the panel's roster is finalized. This single step, using a free public database, would have caught the overwhelming majority of the inaccurate disclosures documented in the BMJ Open study, and it costs nothing beyond the staff time to run the search.
Publish your selection process alongside the guideline itself. The Society of Critical Care Medicine's own language, that "processes for panel selection lack transparency," is a standard your society can simply choose to meet by disclosing how members were chosen, not just who they are.
If you build health-technology, payer, or regulatory advisory panels
Recognize that this problem extends well beyond specialty-society guidelines. Coverage advisory panels, quality-measure developer committees, and HTA bodies draw from the same acquaintance-based candidate pools, and the same open-call, verified-disclosure fix applies directly to your own panel recruitment.
Ask your current advisors how they were selected, and whether that process could survive being published. If the honest answer is "I know them from conferences," that is the exact mechanism this article documents producing both the composition and conflict problems in specialty-society guidelines.
Frequently asked questions
Who writes clinical practice guidelines and how are they chosen? Panel members are typically selected by a chair, appointed by society leadership, who invites people from their own professional circle: prior co-authors, past panel members, and colleagues known from conferences. A Society of Critical Care Medicine team has stated plainly that "processes for panel selection lack transparency."
Do guideline authors have conflicts of interest? Very often, yes. Among 270 US physician authors of 2020 clinical practice guidelines, 199, or 73.7 percent, had received industry payments with a median of $27,451 over three years, and 18 of 22 panel chairs, or 81.8 percent, had industry payments, according to a 2023 study in BMJ Open.
What percentage of guideline authors accurately disclose their industry payments? Only 1.5 percent, or 4 of 270 studied US physician authors of 2020 guidelines, accurately disclosed their conflicts of interest when checked against CMS Open Payments records, per the same BMJ Open 2023 study.
Are guideline panels diverse? Generally not. Across American Urological Association and European Association of Urology guideline panels, 21.6 percent included no women at all, and voting panel members overall were 81.8 percent male and 86.8 percent white, according to a 2024 study in Urology.
How can a community physician join a guideline panel? Currently, there is usually no formal application route; panels are built through personal invitation from the chair. One documented exception exists: a 2024 Critical Care Medicine study describes an open-call process, publishing an expertise definition and inviting the full society membership to apply, that produced a 33-member, six-country panel that was 45 percent women and 79 percent historically excluded people and groups.
Why doesn't CMS Open Payments already catch inaccurate conflict-of-interest disclosures on guidelines? Because guideline developers have generally not built automated cross-checking against the database into their author disclosure process, relying instead on physicians' self-reported forms. The database itself is comprehensive: a 2024 JAMA analysis found 826,313 of 1,445,944 US physicians, or 57.1 percent, received industry payments between 2013 and 2022, meaning the ground truth needed to verify most guideline authors' disclosures already exists publicly.
The bottom line
Guideline panels are chosen the way they have always been chosen: a chair reaches for the people already in her professional circle, because no other mechanism for identifying candidates has ever existed. The result, measured directly, is a system where 73.7 percent of guideline authors have received industry payments, 81.8 percent of chairs have too, and only 1.5 percent of those relationships were disclosed accurately, alongside panels that were, in a documented share of cases, entirely without women in a voting role.
These are not two separate failures. They are the same acquaintance-based selection mechanism producing two visible symptoms. Fifteen years of reform effort aimed at disclosure rules and recusal thresholds has not moved the underlying number, because the point of failure sits earlier than any of those rules reach: who was ever considered a candidate in the first place.
One society has already run the alternative. An open call, a published expertise definition, and an independent scored review produced a panel that looked nothing like the acquaintance-selected norm, without anyone claiming the resulting guideline suffered for it. That single documented case removes the strongest argument for not trying it elsewhere.
That society president is going to email her seven familiar names again for the next guideline cycle, because nobody has ever asked her to do it differently, and no structure currently exists to make a wider search easier than the list she already keeps in her head. The fix here is not a mystery. It has already been run once, in public, and it worked.
Part of a series on the missing professional infrastructure of healthcare. Previously: Answered, Not Expert
Evidence note: guideline author industry payment and disclosure accuracy figures are from a 2023 study in BMJ Open analyzing 270 US physician authors of 2020 clinical practice guidelines against CMS Open Payments records. Panel composition figures (21.6 percent with no women, 81.8 percent male, 86.8 percent white voting members) are from a 2024 study in Urology covering American Urological Association and European Association of Urology guideline panels. The open-call panel case study (33 members, six countries, 45 percent women, 79 percent historically excluded people and groups) and the "processes for panel selection lack transparency" quotation are from a 2024 study in Critical Care Medicine describing a Society of Critical Care Medicine guideline process. Broader physician industry-payment figures (826,313 of 1,445,944 physicians, 57.1 percent, 2013 to 2022, median $48, top 0.1 percent averaging nearly $2 million) are from a 2024 JAMA analysis of CMS Open Payments data. This article describes documented, published findings; the opening scenario is a composite illustration built from the patterns in this evidence, not a specific reported case or society.
