A third-year general surgery resident at a community-based program is sitting in her call room at 9:40 p.m. with a laptop open to a blank abstract template.
Her program, like essentially every one of the more than 13,000 ACGME-accredited residency and fellowship programs in the country, requires scholarly activity to graduate and to stay accredited. She has an idea: a retrospective look at a technique variation her attendings use that she has not seen written up anywhere. It is a genuinely good idea. What she does not have is anyone to take it to.
Her program is affiliated with a community hospital, not a university medical center. There is no research office down the hall, no biostatistician she can grab for twenty minutes, no senior faculty member with a standing lab and a stable of trainees rotating through it. The one attending who publishes occasionally is stretched between three hospitals and has said, kindly and honestly, that he does not have time to mentor a project from scratch this year.
She emails a researcher two states away whose paper on a related topic she admired. No response. She checks her specialty society's mentor-matching page, which runs once a year and is already closed for this cycle. She considers, seriously, just writing up a case report instead, something small enough to finish alone, because at least that satisfies the requirement on the page even though it will do nothing for the research career she actually wants.
Eighty miles from her call room, at the region's academic medical center, a professor with three funded grants and a wait list of eager trainees is turning down a fourth mentee this month because he is already at capacity, and does not know a single resident like her exists.
The requirement to produce scholarship is universal. The infrastructure to meet it is not, and nothing routes the surplus that exists at one end of that gap to the deficit at the other.
The size of the gap, measured directly
This is not an impression. It has been measured, and the number is stark.
A 2026 survey of 93 surgeons published in the American Journal of Surgery found that community-based practice was associated with an odds ratio of 0.09 for having research mentor access, compared with academic peers (95 percent confidence interval 0.01 to 0.69, p equals 0.02). An odds ratio of 0.09 means, in plain terms, roughly a 90 percent reduction in the odds of having a mentor at all simply by virtue of practicing outside an academic center.
That figure describes surgeons already in practice, not just trainees. The gap does not close when residency ends. It follows a physician into the rest of her career, at exactly the moment promotion, credibility and grant competitiveness increasingly depend on a publication record she has structurally less access to build.
A separate systematic review of 33 studies, drawn from 2,351 screened, names "lack of strong research mentorship and sponsorship" and "unsupportive institutional cultures that lack resources or infrastructure" as the dominant drivers of scholarship delay specifically for early-career and underrepresented faculty. The geographic gap and a demographic one compound each other rather than operating separately.
People are already leaving because of this
The consequence is not abstract. It is attrition happening now, among physician-scientists who are the products of years of federal training investment.
A national survey of 230 physician-scientists found that nearly half reported considering leaving their research career within the next two years. The top reasons: burnout and unhappiness (35 percent), stress (35 percent), and lack of funding (30 percent). Sixty-three percent cited difficulty balancing clinical and educational responsibilities, and 53 percent cited work-life balance.
Read those reasons carefully. Funding is real, but it sits third, and mentorship shortage is not a separate line item in that list, it is the connective tissue underneath most of the others. A physician-scientist with a present, engaged mentor has someone helping triage which projects are worth the balancing act, which grants are realistic, and how to say no to the clinical load that is crowding out the research time. A physician-scientist without one is absorbing all of that judgment alone, on top of the actual scientific work.
A 2025 national forum convened by the American Society for Clinical Investigation, the Association of American Independent Medical Departments, and the Burroughs Wellcome Fund explicitly flagged "the unclear path afterward" for physician-scientist trainees as a workforce-sustainability problem. That framing matters: the field's own leadership is now naming the pipeline itself, not just the funding pool, as the binding constraint.
A K-award, the standard federal bridge to research independence, commonly carries $75,000 to $100,000 a year in salary support for three to five years. Every physician-scientist who exits early after receiving one represents a multi-year federal and institutional investment that produces no return. Extrapolated across even a fraction of the near-half of surveyed physician-scientists reporting intent to leave, that is a recurring, multi-million-dollar national loss, on top of whatever science never gets done.
Why the requirement exists everywhere but the support does not
Follow the logic of how scholarly-activity requirements and research infrastructure came to be distributed so unevenly, because the mismatch is not an accident of underfunding, it is structural by design.
ACGME's scholarly-activity requirement applies to essentially every accredited program, community-based or academic, because the accreditation standard is written at the level of the trainee's development, not the institution's research capacity. That is reasonable on its own terms: a resident anywhere should learn to think critically about evidence and contribute to it.
Research infrastructure, by contrast, is concentrated by design. Grant overhead, protected faculty time, core laboratory facilities and biostatistics support are expensive, and they accumulate at a small number of institutions that have spent decades building them, chiefly R1 research universities and their affiliated academic medical centers. That concentration is not a flaw in the research system. It is how large, capital-intensive science gets funded and sustained.
The problem is that nobody reconciled the two facts. A universal requirement was layered onto a deliberately non-universal infrastructure, and the assumption, never stated outright but embedded in how programs are accredited, was that trainees would simply be wherever the infrastructure already was. Most are not. A resident or attending in a community or rural setting is handed the identical requirement with a small fraction of the access to fulfill it meaningfully.
Why nobody owns fixing it
R1 institutions have no incentive to export their mentors' time. A professor's mentoring capacity is, functionally, an internal asset used to develop that institution's own trainees, publish under that institution's name, and support that institution's grant competitiveness. There is no reward structure that asks a busy researcher to spend an hour on a community resident he will never see again.
ACGME sets the requirement but supplies no mentors. Accreditation bodies are built to define standards and audit compliance, not to build or fund the workforce that meets those standards. That is true across this entire series and it is true here.
Specialty societies run matching programs that cannot scale. Where they exist, these are typically once-a-year, low-capacity pairing exercises, closed for the cycle by the time most trainees realize they need one, and structurally unable to serve more than a small fraction of the community and rural population who need them.
Commercial research consultants exist, and price out exactly the people most affected. A consultant can, for a fee, provide protocol review or statistical support. That fee is precisely the barrier a community-based trainee or an under-resourced rural attending is least equipped to pay, which means the commercial substitute serves the population that needed it least.
The result is a gap that persists not because no institution has thought about it, but because every institution that could plausibly close it has a structural reason not to, and the people who most need access have no lever to compel anyone to provide it.
What is currently substituted, and why it falls short
The workaround, described directly by residents and community attendings, is some combination of a cold email to a stranger whose paper they admired, a once-a-year society matching form with limited capacity, or simply satisfying the requirement with the smallest, least career-building project available, typically a case report or a quality-improvement summary that checks the accreditation box without building the skills or track record a real research career requires.
Physician-scientist bridge programs, including K12 institutional research-track fellowships, do exist and are genuinely valuable. They are concentrated at the same R1 institutions that already have a surplus of mentors, which means they add depth exactly where depth already exists rather than reaching the deficit everywhere else.
None of this is a failure of ambition on the trainee's part. It is a failure of routing. The 0.09 odds ratio does not describe a population that stopped trying. It describes a population trying against structural odds that make success unlikely regardless of effort.
What would actually work
Route existing surplus rather than building new infrastructure from scratch. The fix does not require every community and rural program to somehow build its own version of an R1 research office. It requires connecting the mentorship capacity that already exists at research-intensive institutions to the deficit that already exists everywhere else.
Match on a specific, narrow deliverable, not an open-ended relationship. A community trainee does not need a permanent lab appointment. She needs an hour of expert review on an IRB protocol, a grant aim, a dataset question, or a draft abstract, delivered by someone with relevant expertise and logged capacity.
Verify the mentor's actual availability, not just their willingness in principle. A mentor-matching system fails the moment supply is nominal rather than real; capacity needs to be logged and current, the same discipline this series has argued for in every expertise-routing problem it examines.
Track the downstream outcome, not just the introduction. Whether the protocol was submitted, the abstract accepted, the grant funded. Without outcome tracking, nobody can tell whether the matching mechanism is actually closing the gap or just producing well-intentioned conversations that go nowhere.
Recruit senior and retired researchers deliberately as a supply source. A retired or reduced-hours academic researcher, past the pressure of grant competition and tenure clocks, is precisely the underused capacity this series has repeatedly identified elsewhere in medicine, and research mentorship is one of the highest-leverage uses of that time.
Build authorship and credit norms into the structure from the start. Cross-institution mentorship raises real questions about who gets credited when a mentor from outside the trainee's institution contributes substantively; leaving this ambiguous discourages exactly the engagement the system needs.
Make it geography- and specialty-indexed, not a single national pool. A useful match requires overlap in both methodology and specialty; a general "mentor available" listing without that indexing produces noise, not usable connections.
What you can do now
If you are a community-based or rural trainee or attending
Ask explicitly, and ask more than once. The literature on this gap describes silence and structural absence, not researchers who actively refuse to help; a specific, narrow ask, "would you spend twenty minutes reviewing my aim," succeeds more often than trainees expect, especially from senior or retired researchers.
Start with the smallest deliverable that moves your project forward, not the whole relationship. Asking for a single review of a protocol or an abstract is a request most researchers can say yes to; asking to be taken on as an ongoing mentee is a much larger ask that is easier to decline.
Use your specialty society's matching program, but do not wait on it alone. It runs once a year and has limited capacity; treat it as one channel among several rather than the whole plan.
If you are an R1-based or academic researcher
Consider that twenty minutes of review is not the same commitment as taking on a mentee. Much of the gap this article describes is not a lack of willing researchers, it is a lack of any channel connecting willing researchers to people outside their own institution who have a specific, bounded ask.
Say yes to the narrow request even when you cannot say yes to the ongoing relationship. A single structured hour, logged and outcome-tracked, is a meaningful contribution that does not require the ongoing time commitment that makes most researchers decline.
If you direct a residency or fellowship program, or run a GME office
Say the number out loud to your trainees and faculty. An odds ratio of 0.09 for research mentor access at community-based programs is a concrete, citable fact, and naming it changes how a program thinks about its own scholarly-activity expectations.
Do not let the requirement default to the smallest possible project. If your program's scholarly-activity compliance is quietly being satisfied entirely through case reports because nothing else is accessible, that is a measurable signal your trainees need a mentorship channel your institution cannot supply alone.
If you fund or govern physician-scientist training
Look at the near-half attrition-intent figure as the leading indicator it is. Losing physician-scientists after a K-award has already been funded is a preventable, multi-year loss, and mentorship access is one of the few levers in that equation that has not been seriously tried at scale.
Fund routing, not another curriculum. A cross-institution mentor-matching mechanism, verified and outcome-tracked, addresses the access gap directly; another scholarly-activity training module does not.
Frequently asked questions
Do community-based physicians have less access to research mentors than academic physicians? Yes, substantially. A 2026 survey of 93 surgeons published in the American Journal of Surgery found community-based practice associated with an odds ratio of 0.09 for having research mentor access compared with academic peers, roughly a 90 percent reduction in the odds of having a mentor at all.
Why do physician-scientists leave research careers early? A 2025 survey of 230 physician-scientists published in BMC Medical Education found nearly half were considering leaving their research career within two years, citing burnout and unhappiness (35 percent), stress (35 percent), lack of funding (30 percent), difficulty balancing clinical and educational duties (63 percent) and work-life balance (53 percent).
What is the ACGME scholarly activity requirement? ACGME requires residents and fellows in essentially every accredited program, more than 13,000 nationally, to engage in scholarly activity, but the requirement applies uniformly regardless of whether a program's site has research infrastructure such as protected faculty time, statistical support or lab access.
How can a rural or community physician find a research mentor? Currently, mainly through cold outreach to researchers whose published work they admire, once-a-year specialty-society matching programs with limited capacity, or paid commercial research consultants; a 2025 narrative review of 33 studies names weak mentorship access and unsupportive institutional infrastructure as the dominant driver of scholarship delay for these physicians.
What does an odds ratio of 0.09 actually mean in this context? It means the odds of a community-based surgeon reporting research mentor access were roughly 91 percent lower than for an academic-based surgeon with otherwise comparable characteristics, in a 2026 American Journal of Surgery survey of 93 surgeons (95 percent confidence interval 0.01 to 0.69). It is a measure of relative odds, not a direct percentage of physicians affected, and the sample size is modest.
Is there a way for retired academic researchers to mentor physicians outside their own institution? Not in any structured, at-scale way today. Existing K12 and institutional research-track fellowships are concentrated at the same R1 institutions that already have a surplus of mentors, and specialty-society matching programs run once a year with limited capacity, leaving retired and senior researchers' available time largely unrouted to the community and rural physicians who need it most.
The bottom line
Every one of more than 13,000 ACGME-accredited programs requires scholarly activity of its trainees. The infrastructure that makes real scholarship possible, protected faculty time, grant experience, a lab or dataset to plug into, exists almost exclusively at a small number of research-intensive institutions.
A community-based surgeon has, by measured odds, roughly a tenth the access to a research mentor that an academic peer has. Nearly half of surveyed physician-scientists, people who have already cleared years of federal training investment, are weighing whether to leave research within two years, and mentorship shortage sits underneath most of the reasons they give.
None of this is a talent problem. It is a routing problem, and a peculiarly clean one: a known surplus at one end, a known deficit at the other, and nothing built to connect them, because no institution that holds the surplus has a reason to export it and no body that sets the requirement has the mandate to supply what meeting it actually takes.
Eighty miles from that community resident's call room, a professor with unused mentoring capacity is turning down a mentee this month, unaware she exists. The distance between them is not the problem. The absence of anything that could close it is.
The requirement to produce scholarship is universal. The infrastructure to meet it is not, and nothing routes the surplus that exists at one end of that gap to the deficit at the other.
Part of a series on the missing professional infrastructure of healthcare. Previously: The Postvention Bench
Evidence note: sources include "Mapping pathways to professional support," American Journal of Surgery, 2026, based on a survey of 93 surgeons, a modest sample whose odds ratio carries a wide confidence interval (0.01 to 0.69); Farhat et al., BMC Medical Education, 2025, surveying 230 physician-scientists on attrition intent; a narrative review of 33 studies (2,351 screened) in Family Medicine, 2025; and a 2025 JCI Insight report on an ASCI/AAIM/Burroughs Wellcome Fund national forum. The surgery-specific mentor-access data should not be assumed to generalize precisely to every specialty without further study, though the attrition and mentorship-barrier findings are drawn from a broader, multi-specialty physician-scientist sample.
