January 1, 1970

How University Hospitals Train Future Physicians: Inside the Pipeline

Medical students in a lecture hall during physician training

Every year, roughly 40,000 medical school graduates compete for about 37,000 residency slots through an algorithm that runs on ranked preference lists submitted by both applicants and programs. If you've puzzled over why the U.S. keeps producing medical school graduates while also projecting an 86,000-physician shortfall by 2036, the answer lives inside that gap — and inside the $21.2 billion Medicare spent on physician training in FY2023 alone.

University hospitals sit at the center of this system. They're where newly credentialed MDs learn to actually practice medicine. Understanding how they do it explains a lot about why American healthcare looks the way it does.

The Training Pipeline From Start to Finish

Medical education is a long relay race, and the hospital's role starts before graduation.

Third and fourth-year medical students arrive on hospital floors for clinical clerkships — rotations through internal medicine, surgery, pediatrics, ob-gyn, psychiatry, and emergency medicine. Students take patient histories, perform physical exams, present cases on morning rounds, and begin absorbing the rhythms of real clinical decision-making. No anatomy lab or lecture hall replicates what happens at a patient bedside at 3 a.m.

After graduation comes the Match (more on that shortly), followed by residency. The first year — intern year, or PGY-1 — tends to be the sharpest learning curve most physicians ever experience. Real patients, real responsibility, and supervisors who are watching but not always arm's length away.

Residency length varies by specialty. Family medicine and internal medicine run 3 years. General surgery takes 5. Neurosurgery extends to 7. After residency, physicians who want to subspecialize complete fellowships lasting 1 to 3 more years. The full path from the first day of medical school to independent subspecialty practice commonly stretches to 15 years.

Here's the sequence in order:

Years 1–2: Medical school preclinicals — anatomy, physiology, pharmacology, pathology
Years 3–4: Clinical clerkships at affiliated teaching hospitals
Year 5+: ACGME-accredited residency training in a chosen specialty
Optional: Fellowship for subspecialty certification (1–3 additional years)

The Match — One Night That Changes Everything

The National Resident Matching Program (NRMP) runs the algorithm that determines where each graduate will spend the next several years of their professional life. Both applicants and programs submit ranked preference lists. The matching logic derives from the Gale-Shapley stable matching theory, developed by economists David Gale and Lloyd Shapley in 1962 — the same framework that eventually earned Shapley a Nobel Prize in Economics in 2012.

Match Day falls on the third Friday of March each year. Thousands of medical students open envelopes simultaneously at schools across the country. Some match their top choice. Some match their fifth. Some don't match at all — landing in the Supplemental Offer and Acceptance Program (SOAP), where they scramble for unfilled slots in the days that follow.

In 2024–2025, there were 13,762 ACGME-accredited programs with 167,083 active residents and fellows in training. Of those, 82.3% were in specialty programs and 17.7% in subspecialty fellowship programs. That's a substantial training infrastructure — but still not enough to keep pace with the physician demand a growing, aging population generates.

What Life Inside a Residency Actually Looks Like

Short version: long hours, genuine stakes, and nothing resembling the TV version of medicine.

Residents work under attending physicians — fully licensed doctors who carry legal and professional responsibility for patient outcomes. Early in training, significant clinical decisions get reviewed before they're executed. By year three, residents operate with considerably more independence, preparing for the day they sit on the other side of that supervisory relationship.

A typical internal medicine resident might spend five days managing a 15-patient hospital service, rotate to outpatient clinic for a week, then head to the ICU for a month. The breadth is deliberate. Before physicians narrow their focus, programs expose them to the full clinical spectrum of their specialty.

Teaching rounds provide daily structure. Each morning, a team — attending physicians, residents, interns, and medical students — moves bed to bed reviewing each patient. The senior resident presents; the attending probes clinical reasoning with questions rather than just delivering answers. It's Socratic in design. Messy in practice.

Beyond rounds, there are morbidity and mortality conferences. When a patient outcome goes wrong, the care team examines what happened — systematically and, at least in theory, without personal blame. These sessions are often uncomfortable. They're also how teaching hospitals turn individual errors into institutional knowledge, which is why every ACGME-accredited program requires them.

Simulation Labs: Practicing Before the Stakes Are Real

Before simulation technology matured, the only way to learn a procedure was to do it on actual patients, under supervision, with all the uncertainty that implies. That trade-off defined medical training for over a century.

Boston University's Rod Hochman Family Clinical Skills and Simulation Center spans 13,500 square feet and houses four emergency and hospital rooms, thirteen standardized patient exam rooms, and thirteen clinical skills labs. Students practice central line placements, intubations, and trauma scenarios on high-fidelity mannequins that bleed, develop cardiac arrhythmias, and respond — or fail to respond — to medications. Instructors run scenarios from an adjacent control room, adjusting variables in real time.

The real value of simulation isn't mimicking reality. It's separating skill acquisition from patient risk. By the time a resident places their first central line in a live ICU patient, they should already have failed that procedure — and recovered from it — dozens of times.

That said, simulation has limits. No mannequin replicates the cognitive load of managing five simultaneous clinical concerns on a genuinely sick person while also fielding nursing calls and keeping pace with documentation. Simulation teaches procedural skill. The hospital floor teaches everything else.

Medicare's Cap and 27 Years of Inertia

Medicare funds GME through two channels. Direct payments cover resident salaries and program administration costs. Indirect payments compensate teaching hospitals for the operational overhead of running training programs — supervision takes time, documentation is heavier, patient throughput slows. In FY2023, both channels combined for $21.2 billion in federal spending on physician training.

But in 1997, the Balanced Budget Act capped the number of Medicare-funded residency positions at each hospital, freezing slots at their 1996 baseline. The logic at the time: physician supply was adequate. That judgment aged badly. Medical school enrollment has grown roughly 30% since the cap was set, but residency capacity hasn't kept pace.

Congress has chipped at the problem. The Consolidated Appropriations Acts of 2021 and 2023 authorized 1,200 new total slots, with 135 hospitals across 37 states receiving positions in the most recent distribution. Approximately 62% of those slots went to primary care and psychiatry programs, where shortages bite hardest. The bipartisan Resident Physician Shortage Reduction Act — currently under congressional consideration — would add 14,000 slots over seven years at 2,000 per year from 2026 through 2032.

Even if that bill passes in full, it still doesn't close the AAMC's projected 86,000-physician shortfall by 2036. The writing has been on the wall since at least 2010, and every legislative response arrives a decade behind the data.

The Rural and Specialty Gap

Only 2% of Medicare-funded GME slots are located in rural areas, despite rural Americans comprising about 18% of the population. Teaching hospitals are predominantly urban. And where physicians train is a strong predictor of where they eventually practice — professional networks, geographic familiarity, and proximity to academic centers all pull in the same direction.

The specialty distribution reflects financial incentives that individual institutions can influence at the margins but can't override at scale.

Specialty	Residency Length	Match Difficulty	Physician Shortage Risk
Family Medicine	3 years	Lower	High
Internal Medicine	3 years	Moderate	High
Psychiatry	4 years	Moderate	Very High
General Surgery	5 years	High	Moderate
Orthopedic Surgery	5 years	Very High	Low
Neurosurgery	7 years	Extremely High	Low

Primary care positions go unfilled while orthopedic surgery programs receive 15 applications per available slot. The income gap is the driver — family physicians earn roughly $250,000 annually while orthopedic surgeons often clear $600,000 or more. Medical students making specialty decisions during third-year rotations are doing math, consciously or not, and teaching hospitals can't undo that math through program design alone.

What the Best Teaching Programs Get Right

Not all teaching hospitals are serious about teaching. Some run residency programs because programs supply labor and attract Medicare GME revenue. Residents notice the difference, and so do the physicians they become.

Programs that consistently produce strong physicians share recognizable traits:

Attendings have protected teaching time — clinical load doesn't systematically crowd out education
Residents receive regular, specific feedback on clinical reasoning, not generic performance assessments
Research access exists even for residents who never plan to publish, because systematic inquiry sharpens clinical thinking across every specialty
Trainee wellness is tracked and acted on, not just acknowledged in program materials

NewYork-Presbyterian runs more than 150 accredited GME programs with nearly 1,900 trainees across virtually every area of medicine. Indiana University School of Medicine sponsors 114 ACGME-accredited residencies and fellowships. At that scale, the logistics of teaching become their own operational discipline, and culture becomes harder to maintain — which is why the best large programs deliberately invest in it.

My honest read: the most important question a medical student can ask a residency program is whether the attendings actually want to teach. Not whether the program has a research track, not what the call schedule looks like. Whether the people doing the supervising regard teaching as the point. Everything else flows from that.

Bottom Line

The path from medical school to independent practice spans 11–15 years and runs through teaching hospitals at every stage — clerkships during school, then residency, then optional fellowship training.
Medicare's 1997 funding cap remains the structural constraint on physician supply. New legislation is adding slots in small batches, but the projected 86,000-physician shortfall by 2036 won't be solved at 400 positions per distribution.
Rural communities are losing the geography problem before it starts — only 2% of GME slots are in rural areas, and training location predicts practice location. This won't change without deliberate policy shifting both funding and incentives.
Simulation has meaningfully improved early skills training. Learning procedures on mannequins before patients is a genuine advance, not just a liability hedge.
When evaluating residency programs, ask residents — not program directors — how much time their attendings actually dedicate to teaching. That answer tells you more than any brochure.

Frequently Asked Questions

How long does residency training at a university hospital take?

It depends on the specialty. Primary care residencies — family medicine, internal medicine, pediatrics — run 3 years. General surgery is 5 years; neurosurgery extends to 7. Fellowship training for subspecialties like cardiology or oncology adds 1 to 3 more years after residency. Physicians who complete a fellowship after a surgical residency may not reach independent practice until their late 30s.

What's the difference between a teaching hospital and a regular hospital?

Teaching hospitals are accredited by the ACGME to run residency and fellowship programs. They have faculty physicians in formal teaching roles, receive Medicare GME funding, and operate under program requirements that mandate structured supervision and educational activity. Non-teaching hospitals may provide excellent clinical care but are not structured around training new physicians and don't receive GME payments.

Do residents get paid for their work?

Yes, though modestly relative to hours worked. First-year residents earned between roughly $58,000 and $65,000 annually as of 2024, typically working 60–80 hours per week. The ACGME caps duty hours at 80 per week averaged over four weeks — a rule introduced in 2003 after research linked resident fatigue to measurable increases in medical errors.

Myth vs. reality: Are residents making unsupervised clinical decisions?

No. Residents are active in clinical decision-making, but attending oversight is always required. The supervision level is calibrated to training stage — a PGY-1 intern has every significant order reviewed, while a senior resident operates with more clinical latitude. Attending physicians hold legal and professional responsibility for patient outcomes throughout the entire residency.

Why does a physician shortage exist if medical schools keep graduating doctors?

The bottleneck is residency capacity. Medical school enrollment has grown roughly 30% since 1997, but Medicare-funded residency slots were frozen at that year's baseline by the Balanced Budget Act. Some graduates each year fail to match and cannot begin independent practice despite completing medical school. The AAMC projects this structural mismatch will produce an 86,000-physician shortfall by 2036.

How does the National Resident Matching Program actually work?

Both applicants and residency programs submit ranked preference lists to the NRMP. The algorithm, based on Gale-Shapley stable matching theory, produces results that are "stable" — no unmatched pairing exists where both a program and an applicant would prefer each other over their current assignment. Results are released simultaneously on Match Day each March, and matched applicants are contractually bound to attend.