Physician Burnout Statistics (2026)

Physician burnout peaked at 62.8% in late 2021 and has fallen every year since. The American Medical Association's 2025 benchmark, built from nearly 19,000 physician responses across 106 health systems, puts it at 41.9%. The most recent national study with a general-worker comparison puts it at 45.2% for 2023, roughly where it sat in 2017. Both figures are real, they come from different samples, and the pages that rank for this question splice them into one line without saying so.

This page keeps the series apart, gives the sample and fielding date next to every number, and dates the cost estimates that get quoted as if they were new. It covers prevalence and trend, specialty, hours and electronic health record time, intent to leave, what burnout costs health systems, how it differs by career stage and gender, and what the evidence says reduces it.

Prevalence and trend: two series, not one

Two measurements of physician burnout get reported in the US, and they do not share a sample or a method. The AMA's Organizational Biopsy collects responses from physicians at health systems that choose to take part, using a validated short burnout item. The triennial national study led by Tait Shanafelt and colleagues draws a sample from the AMA Physician Masterfile, uses the Maslach Burnout Inventory, and compares physicians with other US workers. The AMA series shows the recent trend. The national study shows where physicians sit against everyone else.

The AMA figures come from its April 2026 release and the two before it. The 2021 and 2023 national figures are from Mayo Clinic Proceedings 2022 and the 2025 follow-up in the same journal, which found that the 2023 rate was not statistically different from 2017. The 2021 study had a 5.6% response rate, which the authors address with demographic weighting, and which is worth remembering when the 62.8% peak is quoted as a precise figure.

Three other measures appear in the ranking pages and measure different things. The Physicians Foundation's 2025 survey of about 1,000 physicians found 54% "often" feel burnout, down from 60% in 2024. Doximity's June 2026 poll of over 600 physicians found 82% say they are overworked, which is not a burnout measure. And the CDC's 2023 MMWR analysis found 19.0% of health workers of all kinds felt burned out "very often" in 2022, up from 11.6% in 2018, with 45.6% saying often or very often. That population is nurses, aides and technicians as well as physicians, and the sample was 325 people.

Alongside burnout, the AMA tracks job stress and satisfaction in the same sample. Job stress fell from 55.6% in 2022 to 42.9% in 2025. Job satisfaction rose from about 68% to 77%. The share who feel valued by their organization rose from 46.3% to 56.2%. The direction is consistent across all four indicators.

Burnout by specialty

The AMA's 2025 data, published in April 2026, gives a spread of more than 26 points between the highest and lowest specialties.

Emergency medicine has led every recent measurement. In the 2021 national study it had the highest adjusted odds of burnout of any specialty, at 4.59 times the reference group, with general pediatrics at 2.44 and family medicine at 1.57. Grouped by setting, the 2025 AMA data put hospital-based specialties (emergency medicine, radiology, anesthesiology) worst on three of five indicators, with the lowest job satisfaction at 74.8% against 83.0% for psychiatry.

Residents and fellows are measured separately. The AMA's 2025 resident report, from more than 3,000 trainees at 29 organizations, found 28.6% burnout, down 5.9 points on the year, with 90.1% satisfied with their program.

Hours, EHR time and administrative load

Physicians in the AMA's 2024 sample worked an average of 57.8 hours a week: 27.2 in direct patient care, 13.0 in indirect care and 7.3 on administration, down from 59.0 hours in 2023. Hospital medicine averaged 62.8 hours and emergency medicine 52.3. The same AMA analysis found that 22.5% of physicians spent more than eight hours a week in the electronic health record outside working hours, defined as 5:30 pm to 7 am on weekdays. That share was 20.9% in both 2022 and 2023. Hours fell and burnout fell, and after-hours EHR time rose.

The EHR figures that circulate most widely are older than they look. The claim that doctors spend two hours on the computer for every hour with patients comes from a 2016 Annals of Internal Medicine study that observed 57 physicians in four specialties for 430 hours in 2015, and found 49.2% of the office day on EHR and desk work against 27.0% of direct clinical face time, plus one to two hours of evening work in the diaries of the 21 who kept them. It is a careful study of a small, self-selected group of high-performing practices.

Larger log-based measurements exist. A 2021 JAMA Internal Medicine analysis of 351 ambulatory organizations on Epic, covering January to August 2019, found primary care clinicians averaged 115.0 minutes of active EHR time a day, with 29.8 minutes of it after hours; medical specialties averaged 85.7 minutes and surgical specialties 45.6. A later study of 307 primary care physicians at Mass General Brigham, reported by the AMA in January 2024, found a 30-minute visit generated a median 36.2 minutes of EHR work, including 6.2 minutes after hours and 7.8 minutes in the inbox.

The CDC's 2023 analysis quantified what protects health workers. Trusting management reduced the odds of burnout to 0.40 of the baseline, having a supervisor who helps to 0.26, having enough time to do the job to 0.33. Harassment raised the odds 5.83 times, and the share of health workers reporting harassment rose from 6.4% in 2018 to 13.4% in 2022.

Intent to leave, cut hours or retire

"One in five doctors plans to leave" is the most quoted statistic in this field, and it is 2020 data. The source is a 2021 Mayo Clinic Proceedings study of 20,665 health care workers at 124 institutions surveyed between July and December 2020. Among the 9,266 physicians, 23.8% intended to leave their practice within two years and 31.4% intended to reduce clinical hours within a year. Burnout raised the odds of intending to leave 2.57 times; feeling valued cut them to 0.40. Physician burnout in that sample was 47.9%. Nurses were at 63.1%, with 40.0% intending to leave.

More recent figures are narrower in scope. The AMA's 2024 data by career stage found 36.5% of physicians more than 20 years out of training were likely to leave their organization within two years, down from 40.8% a year earlier, and 25.4% of those 16 to 20 years out. Doximity's June 2026 poll found 46% considering early retirement, up from 34% a year earlier, and 44% somewhat or very likely to leave clinical practice within two years, while 81% said they would choose medicine again. The CDC found 16.5% of health workers "very likely" to look for a new job in the next year in 2022, up from 11.1% in 2018.

What burnout costs, and when that number was calculated

The $4.6 billion figure that appears on every page about this topic comes from a cost-consequence model published in Annals of Internal Medicine in May 2019 by Han, Shanafelt, Sinsky and colleagues. It estimates the national cost of physician burnout attributable to turnover and reduced clinical hours only, with a sensitivity range of $2.6 billion to $6.3 billion, or about $7,600 per employed physician per year. It excludes the costs of quality, safety and patient satisfaction, it is not inflation adjusted, and the authors called it conservative. It is also built on inputs from the middle of the last decade. Quote it with the date.

The per-physician replacement cost of $500,000 to $1 million is usually attributed to a 2017 JAMA Internal Medicine paper by Shanafelt, Goh and Sinsky. That paper cites earlier recruitment-cost estimates rather than measuring them, so the honest attribution is "estimates cited by," not "found by."

The supply picture gives the cost its context. The AAMC's March 2024 projection puts the US physician shortage in 2036 at between 13,500 and 86,000, with primary care accounting for 20,200 to 40,400 of it. The report body states that 17% of active physicians were 65 or older in 2021 and a further 25% were 55 to 64, and that more than a third of currently active physicians are likely to retire within a decade. The 2021 edition had projected a shortage of up to 124,000 by 2034; the lower 2024 range assumes 1% annual growth in residency positions.

Career stage and gender

Burnout peaks in mid career. The AMA's 2024 breakdown by years since training shows 39.3% at one to five years, 48.4% at six to ten, 49.4% at 11 to 15, 46.5% at 16 to 20 and 38.3% beyond 20, with every band lower than in 2023. Feeling valued rose in every band too; among physicians in their first five years it went from 50.5% to 57.8%.

Women physicians report feeling valued less often than men, at 53.3% against 59.6% in the AMA's 2025 data. The national study's 2023 sample was 39.6% women, close to the 37.9% share of practicing US physicians, and a January 2026 research letter from the same group, reported by the AMA, found women physicians had higher odds of burnout than men at four of five measurement points since 2011 and 33% lower odds of being satisfied with work-life integration. The research letter itself was not opened for this page; the AMA's account of it is the source.

Against the rest of the workforce, physicians in 2023 had 1.82 times the odds of burnout of other US workers and 0.59 times the odds of being satisfied with work-life integration, after adjustment. The "twice as likely" line that still circulates is from the 2014 data.

What the evidence says helps

The strongest signal in the recent data is organisational rather than individual. As AMA burnout fell 11 points between 2022 and 2025, the share of physicians who feel valued rose 10 points and job stress fell 12.7 points. The 2020 intent-to-leave study put a number on the mechanism: feeling valued cut the odds of planning to leave by 60%. The AMA's Joy in Medicine program recognized 164 organizations across 40 states in 2025, and 34 of the 106 systems in the 2025 benchmark were recognized in both 2024 and 2025.

Documentation tools are the newest claim. A JAMA Network Open study published in October 2025 followed 263 ambulatory physicians and advanced practice providers at six US health systems who adopted an ambient AI scribe. Burnout fell from 51.9% before to 38.8% after 30 days, and after-hours note writing fell by 0.9 hours a week. The design was pre and post with no control group, the participants were voluntary early adopters, and some authors had ties to the vendor. It is a promising signal and not yet a measured effect. Vendors' own pages present it as the latter.

For managers outside medicine, the pattern transfers. Our employee burnout statistics page shows the same split by psychological safety across the general workforce, and the guide to spotting burnout in workload data covers the upstream signals, which in medicine are the hours and the after-hours EHR time above.

Where these figures stop being useful

Three cautions. The AMA sample is not a probability sample; health systems opt in, and systems that measure burnout may differ from those that do not. The national study's 2021 wave had a 5.6% response rate. And every survey here measures a self-report on a scale, which is why a single physician can be 41.9% burned out in one series, 54% in another and 82% overworked in a third on the same afternoon. Use the AMA series for direction, the national study for comparison with other workers, and the cost figures with their dates attached.

Key takeaways

• Physician burnout was 41.9% in the AMA's 2025 data, the fourth consecutive annual decline from 53% in 2022. The national study found 45.2% in 2023, back to 2017 levels after a 62.8% peak in 2021.
• The two series come from different samples and instruments. The 2021 peak belongs to the national study, not the AMA benchmark.
• Emergency medicine has the highest rate at 49.8%, infectious diseases the lowest at 23.3%. Hospital-based specialties score worst on most indicators.
• Physicians averaged 57.8 hours a week in 2024. After-hours EHR time rose to 22.5% spending more than eight hours a week even as burnout fell.
• The widely quoted 'one in five plan to leave' is 2020 data. Doximity's 2026 poll found 46% considering early retirement.
• The $4.6 billion cost estimate is a 2019 model covering turnover and reduced hours only, with a range of $2.6 to $6.3 billion.
• Feeling valued cut the odds of intending to leave by 60% in the 2020 study, and the share who feel valued rose from 46.3% to 56.2% as burnout fell.
• Physicians had 1.82 times the odds of burnout of other US workers in 2023, not twice as the 2014 figure suggested.