Executive summary
The United States is projected to be short as many as 86,000 physicians by 2036, and the federal workforce model puts the 2038 gap at 141,160.01, 02 Almost every plan to close it starts with training more doctors. That is necessary, and it is slow: a physician who matriculates today needs four years of medical school and three to nine years of residency before practicing independently.03 The faster variable is the one health systems already control. In a survey of 20,665 health workers, approximately one in three clinicians intended to reduce their work hours and one in five physicians intended to leave practice altogether.06 Burnout is not a wellbeing line item on the way to a workforce plan. It is the workforce plan's largest uncontrolled variable.
Demand is rising as the population ages while more than two in five clinically active physicians are already 55 or older.01
Pipeline expansion is the slowest lever available, and it cannot reach the hours being withdrawn right now by physicians who are still on the roster.
Treat burnout-driven attrition as a supply-side intervention with a measurable retention return, delivered as confidential clinical care rather than as a wellbeing campaign.
Hours that would otherwise be withdrawn stay in the schedule, and every retained physician removes a replacement cycle of seven to thirteen years from the workforce plan.
The problemThe shortage beyond the pipeline§
Workforce planning in American medicine is dominated by one number and one lever. The number is the projected shortfall: up to 86,000 physicians by 2036 in the Association of American Medical Colleges' latest projection,01 and 141,160 by 2038 in the Health Resources and Services Administration's simulation model, of which 70,610 are primary care physicians.02 The lever is the training pipeline. Both are real. Neither, on its own, explains why the gap keeps behaving worse than demographics alone predict.
The Bureau of Labor Statistics counts about 839,000 employed physicians and surgeons and projects roughly 23,600 openings a year through 2034, most of them created by people leaving the occupation rather than by growth.03 That is the part of the shortage nobody trains their way out of. More than two in five clinically active physicians are already 55 or older,01 and the decision about when they stop, and how many hours they deliver before then, is made one physician at a time. The Surgeon General's advisory was written against an earlier projection of between 54,100 and 139,000 physicians short by 2033, with the sharpest gaps in primary care and rural communities.07 Burnout moves every one of those decisions forward, and it is the only variable in the supply equation a health system can change inside a single fiscal year.
A physician who cuts from five clinical days to four has not left the workforce. Five of them together have. CEREVITY, 2026
The evidenceWhat the research shows§
Four bodies of evidence bear on the supply question: national prevalence surveys of physician burnout, federal workforce projections, occupational data on how physicians actually leave, and survey work on what burned-out clinicians say they intend to do next. Read separately they describe a morale problem. Read together they describe a capacity problem, and it is a capacity problem with a much shorter fuse than the training pipeline has a response to.
45.2%
US physicians reporting at least one symptom of burnout, 2023
AMA, Mayo Clinic and Stanford, 2024
1 in 3
Clinicians who intended to reduce their work hours
US Surgeon General, 2022
141,160
Projected US physician shortfall by 2038
HRSA, 2025
839,000
Physicians and surgeons employed in the US, 2024
Bureau of Labor Statistics, 2025
The pattern those four figures make is the argument of this paper. Burnout among US physicians peaked at 62.8 percent in 2021 and fell back to 45.2 percent by 2023, which is close to where it sat in 2011.04 That is real improvement, and it still describes nearly half the profession. A separate national comparison run by the American Medical Association across 81 organizations in 31 states, covering more than 12,400 physician responses, put the 2023 figure at 48.2 percent.05 Set either reading against a workforce of about 839,000 physicians and surgeons03 and a projected 2038 shortfall of 141,160,02 and the question stops being about morale. If even a fraction of the stated intent to reduce hours is realized, the clinical time withdrawn from inside the current workforce is the same order of magnitude as the gap the pipeline is being asked to close.
04 American Medical Association reporting on the triennial national physician burnout survey conducted with Mayo Clinic, Stanford Medicine and the University of Colorado School of Medicine, readings at three-year intervals 2011 to 2023.
Two features of that series matter for planning. The swing is large, more than 24 percentage points between the 2020 and 2021 readings, and the measurement interval is short relative to the pipeline: national prevalence is re-read every three years, while a replacement physician takes seven to thirteen years to produce from matriculation.03 That asymmetry is the case for treating the current state of physician wellbeing as a leading indicator of clinical capacity rather than a lagging measure of morale.
08 Centers for Disease Control and Prevention, Vital Signs, Morbidity and Mortality Weekly Report 72(44), 2023. General Social Survey Quality of Worklife module, 226 health workers in 2018 and 325 in 2022.
The federal picture points the same way. Between 2018 and 2022 the share of health workers reporting burnout very often rose from 11.6 to 19.0 percent, the share saying they were very likely to look for a new job with another employer rose from 11.1 to 16.5 percent, and reported workplace harassment more than doubled, from 6.4 to 13.4 percent.08 Those are small samples of health workers generally rather than physicians specifically, but the direction is unambiguous and it is the direction that empties a schedule. The pattern is established earliest in training, where 45 to 60 percent of medical students and residents report substantial symptoms of burnout,09 which is why what burnout looks like in the years just after residency is a workforce question and not only a wellbeing one.
| Measure | Finding | Population and sample | Source and year |
|---|---|---|---|
| Projected physician shortfall, 2036 | Up to 86,000 | US, all specialties | AAMC, 2024 |
| Projected physician shortfall, 2038 | 141,160, of which 70,610 primary care | US, all specialties | HRSA, 2025 |
| Clinically active physicians aged 55 or older | 42 percent (20 percent aged 65+, 22 percent aged 55 to 64) | US clinical physician workforce | AAMC, 2024 |
| Physicians with at least one burnout symptom | 45.2 percent in 2023, down from 62.8 percent in 2021 | Triennial national physician survey, 2011 to 2023 | AMA, Mayo Clinic, Stanford, 2024 |
| Intent to reduce clinical work hours | Approximately 1 in 3 | 20,665 health workers | US Surgeon General, 2022 |
| Intent to leave practice altogether | 1 in 5 physicians, 2 in 5 nurses | 20,665 health workers | US Surgeon General, 2022 |
| Health workers very likely to seek a new employer | 16.5 percent in 2022, up from 11.1 percent in 2018 | 325 health workers, General Social Survey module | CDC, 2023 |
The frameworkA model you can name and own§
Workforce dashboards record departures. They do not record the eighteen to thirty-six months before a departure, which is where most of the lost clinical time actually accrues. A named model helps because it gives a health system something to look for while the physician is still on the roster. What follows describes how clinical capacity leaves an organization in four stages, each with a different intervention attached to it, and only the last of which a turnover report can see. It is a clinical description rather than a diagnostic instrument, and the trajectory changes when what sits underneath it is treated, which is why structured clinical treatment for burnout belongs in a retention plan and not only in an assistance-program brochure.
CEREVITY model
The Capacity Withdrawal Model
A four-stage description of how a health system loses clinical hours before it loses a physician. Each stage names something observable in a schedule, a roster or a conversation, and each one is reversible at a different cost.
Absorption
The physician takes on the gap. Extra shifts, notes finished after hours, coverage for a vacancy nobody has filled. Output is unchanged or higher, so nothing in the operational data registers a problem.
Withdrawal
Discretionary work goes first. Committee seats, teaching, mentoring, research time. The clinical schedule still looks intact while the institutional work that holds a department together quietly stops.
Contraction
Clinical hours themselves are reduced. A five-day week becomes four, call is dropped, a partial retirement is negotiated. This is the stage the shortage arithmetic actually feels, and it is almost never coded as attrition.
Exit
The physician leaves the organization, the specialty, or clinical practice altogether. It is the only stage a turnover report can see, and by the time it is visible the capacity has been gone for a year or more.
SCHEMATIC Schematic, not measured data.
Illustrative shape of the four-stage model described in section 03. No underlying measurement; the stage descriptions are qualitative network observations.
The reason to name the middle two stages is that they are where the intervention is cheap and the recovery is fast. A physician in Contraction is still licensed, still credentialed and still known to the patients on the panel. Restoring that person's hours costs a fraction of recruiting a replacement and none of the seven to thirteen years the pipeline needs.03 A physician in Exit has to be replaced by someone who does not exist yet.
By professionHow it presents across roles§
The shortage is a national number. The withdrawal that produces it happens in three quite different settings, and an intervention that works in one is close to useless in another.
Primary care physicians
Primary care carries the largest single share of the projected gap: 70,610 of the 141,160 physicians the federal model expects the country to be short by 2038.02 It is also where the geography of the shortage is most extreme, with nonmetro areas projected to face a 58 percent shortfall against demand while metro areas face 5 percent.02 A rural primary care physician is therefore not one clinician among many. In a two-physician county practice, a single contraction from five clinical days to four removes a tenth of that county's primary care capacity, and removes it without generating a vacancy, a job posting or a line in a turnover report.
The pattern reported in this group is Absorption running for years. Panels grow because the clinic down the road closed. Documentation moves to the evening because the day is full. The work dropped first is the work with no billing code attached, which in primary care means continuity, teaching and the informal case discussion that keeps a small practice safe. Physicians in this position rarely describe themselves as burned out. They describe themselves as behind.
Two things make help-seeking harder here than prevalence figures alone suggest. Licensure and credentialing questions about mental health treatment are a documented reason physicians hesitate to seek care,14 and in a small community the nearest clinician may be a colleague, a patient's relative or someone who shares an admitting privilege. Confidentiality is not a preference in that setting. It is the condition under which treatment happens at all.
Individual confidential psychotherapy built around a physician's schedule
Organizational the clinician wellbeing partnership for medical groups and systems
Physician leaders and health system executives
Chief medical officers, department chairs and service line leaders sit on both sides of this problem. They own the retention number and they are inside the population it measures. That dual position produces a specific pattern: the person who chairs the wellbeing committee is often the person in the organization least able to use anything the committee builds.
The structural reason is visibility. An internal program, however well designed, is staffed by colleagues, reports into a structure the leader sits on, and creates a record inside an organization the leader may be quietly negotiating an exit from. Distress at this level therefore tends to present late, and to present as a resignation rather than as a request for help. What happens when the organizational side is taken seriously is measurable: among health systems recognized for physician wellbeing work, one reported cutting physician turnover to 3 percent and another reported a 64 percent reduction in burnout.16 Those were organizational results, produced by leadership attention rather than by resilience training.
The clinical work at this level is rarely about workload alone. It is about carrying a decision that harmed someone, holding a workforce plan the leader privately believes is undeliverable, and having no peer inside the organization with whom either can be said out loud. The World Health Organization's framing is useful here: burn-out is classified as an occupational phenomenon arising from chronic workplace stress that has not been successfully managed, not as an individual medical condition.11 Read that way, it belongs in the same register as any other occupational exposure a system manages deliberately.
Individual clinical support for the CEO seat
Organizational a confidential mental health benefit for leadership teams
Physician owners in independent and small-group practice
The share of US physicians working in private practice fell from 60.1 percent in 2012 to 42.2 percent in 2024.13 Every point of that decline is a workforce event as well as an ownership one, because a physician who sells or closes a practice frequently reduces clinical hours in the same transaction, and sometimes retires a decade earlier than planned.
Owners occupy the hardest position in the model above, because Absorption is almost indistinguishable from ownership. Covering a nurse's shift, taking the call nobody else will take and deferring one's own leave are all rational responses to running a small business, right up to the point where they are the clinical presentation. There is no HR function to notice, no chief wellness officer and no colleague whose job it is to ask. The person who would normally escalate the problem is the person who signed the lease.
There is also a financial asymmetry that changes the shape of the curve. An employed physician who reduces hours loses income. An owner who reduces hours loses income and keeps carrying fixed cost, which makes Contraction feel unaffordable and pushes the trajectory straight from Absorption to Exit. In practical terms the intervention window for this group is shorter than for employed physicians, and format matters more than messaging: care has to fit around a schedule the owner cannot delegate. Moral distress is part of the picture too, with roughly 20 percent of clinicians in one body of research reporting they had considered leaving a job because of it.10
Individual therapy for owner-operators
Organizational physician wellness program for groups
The stakesThe cost of inaction§
The dollar modeling of physician burnout is done in detail in a companion paper on what a single burnout-driven departure costs a health system. This section takes the supply view instead: what an early exit costs in clinical hours, in replacement time, and in the training years the country has to spend before those hours exist again.
The hours that leave before the physician does
Contraction is the largest and least measured category. A physician moving from five clinical days to four withdraws a fifth of a full-time equivalent while remaining on the roster, so the loss never reaches a turnover rate. Five such reductions in one department remove a whole physician from the schedule and generate no vacancy to recruit against. With approximately one in three surveyed clinicians reporting an intention to reduce work hours,06 this is the category most likely to be understated in a workforce plan built from headcount.
The replacement window
Direct replacement cost for a departing physician is commonly put at 500,000 dollars to more than 1 million dollars once recruitment, sign-on, lost billings and onboarding are counted, and one academic medical center modeled a two-year loss of 15.5 million to 55.5 million dollars from roughly 60 anticipated departures.12 The supply cost sits alongside the dollar cost: the post is empty while the search runs, and that panel is absorbed by colleagues who are already the population this paper is about.
The training years
Replacing a physician from the pipeline rather than from the market takes four years of medical school and three to nine years of residency, plus one to three more for a subspecialty fellowship.03 That is seven to thirteen years from matriculation before an independent clinician exists, against a projected national shortfall of 141,160 by 2038.02 Globally, depression and anxiety already cost an estimated 12 billion working days and about 1 trillion dollars in lost productivity a year.17 Retention is the only lever in the workforce plan with a lead time of zero.
03 Bureau of Labor Statistics, Occupational Outlook Handbook, Physicians and Surgeons: four years of medical school, three to nine years of internship and residency, and one to three further years for a subspecialty fellowship. The recruitment window is shown for context and is illustrative rather than measured.
The solutionWhat effective care looks like§
Effective care for this population has three requirements a general wellbeing offering does not meet. It has to be genuinely confidential, because licensure and credentialing questions about mental health treatment are a documented reason physicians do not seek care.14 It has to be clinical rather than motivational, because what presents as burnout in a physician frequently has a treatable depressive or anxiety disorder underneath it.09 And it has to fit a schedule that cannot be moved, which in practice means availability outside clinic hours and formats that do not assume a free weekday afternoon. Those three constraints, confidentiality, clinical depth and fit, are what CEREVITY's approach to treatment is organized around.
CEREVITY is a nationwide network of independent licensed clinicians, matched to the person rather than assigned by rota, delivered by secure video on a private-pay basis so that no insurance record and no diagnosis code is created. The work runs in three formats. Ongoing weekly sessions at the standard length carry the continuity that recovery from burnout depends on, and the 90-minute session length exists for the sessions where an hour ends exactly where the work was starting.
For a physician who cannot commit to the same hour every week for months, a single three-hour block of work covers ground that would otherwise take a quarter of the year, which matters when the alternative to fitting care into the schedule is the schedule contracting instead.
ImplementationHow to put it into practice§
A retention program that is announced but not used costs more than one that is never announced, because it converts a live problem into a solved one on the dashboard. Four steps separate the two.
- 01
Measure contraction, not only turnover
Put clinical full-time equivalents delivered on the workforce dashboard alongside headcount, and track it by department each quarter. A department losing a tenth of its clinical hours to voluntary reductions has the same capacity problem as one losing a physician, eighteen months earlier and with a far cheaper remedy.
- 02
Separate the clinical route from the employer entirely
Physicians who fear a record will not use a program that creates one. Route clinical care to an external network on a private-pay basis, with no diagnosis code, no insurance claim and no report back to the organization beyond aggregate utilization. Anything less than full separation is read, correctly, as a risk.
- 03
Fund it as workforce supply, not as a benefit
Budget the program against the replacement cost it avoids, commonly cited at 500,000 dollars to more than 1 million dollars per physician,12 rather than against a per-employee benefit benchmark. The comparison that matters is a retained physician against a replacement cycle of seven to thirteen years,03 not a cost per head against peer systems.
- 04
Fix the credentialing language before you launch
Review the mental health questions on the organization's own credentialing and privileging forms first. Physician reluctance to seek care because of licensure repercussions is documented,14 and a system that asks an intrusive question on its own paperwork has already told its physicians what using the program might cost them. Federal guidance for hospital leaders now frames staffing shortages, harassment and violence as drivers of burnout rather than as separate operational issues,15 which puts the paperwork in scope.
RecommendationsWhere to start§
Clinical
Treat what is underneath the burnout
Burn-out is classified as an occupational phenomenon rather than a medical condition,11 which means the label itself is not what gets treated. What is treatable is what sits under it, and in this population that is frequently depression that does not look like depression: performance intact, competence preserved, and everything not compulsory quietly abandoned.
Clinical
Intervene at Contraction, not at Exit
Once a resignation is submitted the clinical work is long and the usual outcome is a change of setting rather than a return to full hours. The stage to act on is the one where a physician is still practicing and has started dropping the parts of the job that are not compulsory.
Structural
Put clinical FTE on the board report
Headcount hides contraction. A board that sees only physicians employed will approve a recruitment budget for a problem recruitment cannot fix, while the hours it is trying to buy are being withdrawn quietly by physicians already on the payroll.
Structural
Buy retention on a supply-side business case
A projected national shortfall of 141,160 physicians by 203802 will not be closed by any single system's hiring plan. What a system can change is the rate at which it loses the physicians it already has, and that is the only part of the workforce equation with a lead time measured in months.
FAQCommon questions§
What causes the physician shortage?
Does burnout worsen the physician shortage?
How do health systems retain physicians?
How does private-pay billing work?
How is my privacy protected?
MethodologyHow this paper was built§
Methodology
This paper was assembled between 15 and 23 August 2026 from publicly available sources, with a deliberate bias toward federal statistical series and national professional-association surveys over single-institution studies, because the question at issue is a national supply question.
Sources searched. Federal and intergovernmental: the Health Resources and Services Administration's Health Workforce Simulation Model projections published in December 2025 covering 2023 to 2038; the Bureau of Labor Statistics Occupational Outlook Handbook entry for physicians and surgeons; the 2022 US Surgeon General's advisory on health worker burnout and its summary page; the Centers for Disease Control and Prevention's Vital Signs report on health worker mental health published in the Morbidity and Mortality Weekly Report in October 2023; the National Institute for Occupational Safety and Health's Impact Wellbeing guidance for hospital leaders; the World Health Organization's ICD-11 classification of burn-out and its 2024 fact sheet on mental health at work; and the 2019 National Academies consensus report on clinician burnout. Professional-association: the Association of American Medical Colleges' March 2024 physician supply and demand projections; American Medical Association reporting on the triennial national physician burnout survey run with Mayo Clinic, Stanford Medicine and the University of Colorado School of Medicine, on its Organizational Biopsy national comparison report, on its Physician Practice Benchmark Survey and on the cost of physician turnover; and the Federation of State Medical Boards' policy on physician wellness and burnout.
Sample sizes, where reported. The triennial national burnout survey series runs at three-year intervals from 2011 to 2023. The Organizational Biopsy national comparison covers more than 12,400 physician responses from 81 organizations across 31 states for calendar year 2023. The intent-to-leave and intent-to-reduce-hours figures come from a survey of 20,665 health workers cited in the Surgeon General's advisory. The CDC comparison rests on 226 health workers in 2018 and 325 in 2022 within the General Social Survey Quality of Worklife module.
Limitations, stated plainly. First, the two shortfall projections quoted here are not interchangeable: they use different models, base years and demand assumptions, and they are reported as two independent estimates rather than as a range. Second, intent to leave and intent to reduce hours are stated intentions, not realized behavior, and the realized fraction is not established by the sources cited. Third, the CDC sample is small and covers health workers generally rather than physicians specifically. Fourth, the two 2023 burnout prevalence readings quoted, 45.2 percent from the triennial national survey and 48.2 percent from the Organizational Biopsy, use different instruments and different populations, and are reported separately here for that reason. Fifth, replacement cost figures are drawn from association reporting and one academic medical center's model, and they vary widely by specialty and region.
No CEREVITY internal intake data is used in this paper. Every numeric claim above is external and carries a numbered citation. The stage descriptions in sections 03 and 04 are qualitative network observations rather than measured findings, and the figure illustrating the model is labeled as a schematic.
References
- 01Association of American Medical Colleges. (2024). New AAMC Report Shows Continuing Projected Physician Shortage. AAMC. aamc.org
- 02Health Resources and Services Administration. (2025). Projecting Health Workforce Supply and Demand. US Department of Health and Human Services. bhw.hrsa.gov
- 03Bureau of Labor Statistics. (2025). Physicians and Surgeons. Occupational Outlook Handbook, US Department of Labor. bls.gov
- 04American Medical Association. (2024). National physician burnout survey: what the latest study shows. AMA. ama-assn.org
- 05American Medical Association. (2024). Physician burnout rate drops below 50 percent for first time in 4 years. AMA. ama-assn.org
- 06Office of the US Surgeon General. (2022). Addressing Health Worker Burnout: The US Surgeon General's Advisory on Building a Thriving Health Workforce. US Department of Health and Human Services. hhs.gov
- 07US Department of Health and Human Services. (2022). Health Worker Burnout: Current Priorities of the US Surgeon General. HHS. hhs.gov
- 08Centers for Disease Control and Prevention. (2023). Vital Signs: Health Worker-Perceived Working Conditions and Symptoms of Poor Mental Health, United States, 2018 to 2022. Morbidity and Mortality Weekly Report, 72(44). cdc.gov
- 09National Academies of Sciences, Engineering, and Medicine. (2019). Taking Action Against Clinician Burnout: A Systems Approach to Professional Well-Being, Summary. National Academies Press. ncbi.nlm.nih.gov
- 10National Academies of Sciences, Engineering, and Medicine. (2019). Taking Action Against Clinician Burnout: Factors Contributing to Clinician Burnout and Professional Well-Being. National Academies Press. ncbi.nlm.nih.gov
- 11World Health Organization. (2019). Burn-out an occupational phenomenon: International Classification of Diseases. WHO. who.int
- 12American Medical Association. (2018). How much is physician burnout costing your organization? AMA. ama-assn.org
- 13American Medical Association. (2025). Physician Practice Benchmark Survey. AMA. ama-assn.org
- 14Federation of State Medical Boards. (2018). Policy on Physician Wellness and Burnout. FSMB. fsmb.org
- 15National Institute for Occupational Safety and Health. (2024). Impact Wellbeing: Guidance for Hospital Leaders. Centers for Disease Control and Prevention. cdc.gov
- 16American Medical Association. (2025). Health system impact: addressing physician burnout. AMA. ama-assn.org
- 17World Health Organization. (2024). Mental health at work. WHO fact sheet, 2 September 2024. who.int
PsyD, Licensed Psychologist
Maria Gonzalez, PsyD is a clinical psychologist licensed in California, New York and Massachusetts, seeing clients by telehealth through CEREVITY's nationwide network of independent licensed clinicians. She works with executives, entrepreneurs, attorneys and accomplished professionals questioning long-held assumptions, on career and identity transitions, on burnout, anxiety and overwhelm, on the particular pressures carried by first-generation professionals, and on grief, loss and relationship strain. Her clinical work draws on psychodynamic therapy, narrative work and acceptance and commitment therapy.
Talk to someone who gets it.
If this paper describes something you recognize, a confidential conversation is the next step. CEREVITY matches you to an independent licensed clinician who works with people in your position.
Schedule consultationPrivate-pay, telehealth, nationwide. Questions: (562) 295-6650