Despite its vast wealth, the United States has fewer physicians per capita than most other developed nations. According to the OECD, the U.S. has about 2.7 physicians per 1,000 residents, ranking 28th out of 31 member countries. This shortage is partly due to the way graduate medical education (GME) is funded. 

Each year, the federal government allocates over $20 billion dollars to teaching hospitals through formulas designed to encourage hospitals to train medical residents and to guide trainees toward specific geographic areas and specialties. However, this system is underperforming. The U.S. graduates only about 8 new physicians per 100,000 residents annually, whereas other OECD countries train two to three times as many doctors per capita.

The impact of this limited supply is dire. Fields like primary care and psychiatry face chronic shortages. Rural areas suffer from inadequate access to care, leading to longer wait times, reduced preventive services, and worsening health outcomes. 

There are several reform options to improve the current system. The Senate GME Working Group, led by U.S. Senators Bill Cassidy, M.D. (R-LA), Catherine Cortez Masto (D-NV), John Cornyn (R-TX), and Michael Bennet (D-CO), recently proposed draft legislation that includes a mix of structural and distributional changes and increases the amount of slots on offer. These changes are an important first step to increase doctor supply, modernize the program, and ensure better distribution of slots. 

To build a healthcare workforce training model capable of meeting the nation’s evolving long-term needs, comprehensive reform of how physician training is subsidized will be essential. Congress should explore structural changes to the Medicare funding of residency programs, including streamlining funding streams and introducing greater flexibility. Alternatively, policymakers could consider fundamentally rethinking the role of federal funding in GME to better align with the nation’s healthcare priorities.

Steps in the right direction

The bipartisan proposal from the Medicare GME Working Group would fund an additional 5,000 residency slots to be distributed to teaching hospitals over 5 years (between 2027 and 2031). This would be five times more than the most recent addition of 1,000 slots (200 per year through 2026) as a part of the Consolidated Appropriations Act of 2021 and represents a 12 percent increase in the total number of residency positions. Each year, thousands of medical school graduates fail to secure a residency position due to a shortage of available slots.  As a result, they are forced to make alternative plans for the following year, such as pursuing research opportunities, enrolling in fellowship programs, or, in some cases, abandoning their aspirations of practicing medicine altogether.

If enacted, the proposal would be just the second time Congress has added slots since a funding cap put in place by the 1997 Balanced Budget Act. Because Medicare covers over two-thirds of GME subsidies, the cap has served as a significant limiting factor to producing enough residency slots to keep up with demand. Residency programs are able to fund slots above the cap, but without financial support from Medicare. In this way, GME funding has a crowd-out effect on teaching hospitals – without residency funding to cover their costs, programs lack sufficient incentive to create enough slots independently to address demand.

The proposal introduces structural reforms by applying a nationally uniform per-resident amount to the newly created 5,000 slots. Medicare subsidizes residencies in two categories: 1) Direct Graduate Medical Education (DGME) which covers the direct costs of running a residency program and makes up 18% of all GME funding, and 2) Indirect Medical Education (IME) which covers the ancillary costs like patient care and makes up 45 percent of GME funding. The draft legislation pertains exclusively to DGME funding. Currently, DGME support is calculated based on each hospital’s reported costs of operating a residency program (which determines the per-resident amount), the number of residents they train, and their Medicare patient load.

The formula for calculating each hospital’s per-resident amount is outdated, relying on cost reporting from 1983 that no longer reflects the actual expenses of operating residency programs today. Since each hospital has a unique per-resident amount, the current formula exacerbates already wide geographic disparities in program funding and does not adjust to population trends and changing demographics. This design rigidity, combined with the capping of GME funding in 1997, has effectively frozen the funding disparities across institutions. The draft legislation seeks to address this issue by standardizing a portion of the cost calculations, ensuring all hospitals receive the same per-resident amount. Additionally, it introduces potential funding boosts for hospitals in rural and underserved areas—a modernization that is much-needed and long-overdue.

The draft bill also allows flexibility for telehealth technologies that teaching physicians might use in supervising residents who are doing their training in rural areas. In many non-interventional and low-risk contexts, the use of real-time audio and video technologies is adequate and reasonable for the purposes of trainee supervision. This will ensure that rural hospitals are not excluded from GME incentives merely because they rely upon efficiency tools such as telehealth technologies.

More work to be done

To ensure residency funding is prioritizing specialties and areas of need, any new legislation should address the root causes of poor distribution built into Medicare’s funding structure. Like the 1,000 slot increase in 2021, the new draft legislation tackles the distribution problem by placing requirements on where new slots can go. Of the new slots, 25% are reserved for primary care and 15% for psychiatry–specialties facing the most significant shortages. For context, as of 2023, approximately 45% of the over 40,000 residency slots are allocated to primary care, while only 5% are dedicated to psychiatry.

Additionally, the draft legislation mandates that at least 10% of new slots go to residency programs in rural areas, shortage regions, hospitals in areas with new medical schools, and those currently training more residents than their cap allows. These guardrails will likely help to distribute residency slots to where they are most needed, but do not address Medicare’s outdated and perverse funding structure that perpetuates the unfair distribution. 

Although the working group’s proposal introduces a uniform amount for calculating DGME payments, significant challenges remain. Indirect medical education (IME) payments, which the proposal leaves unchanged, make up nearly half of GME funding and serve as the primary distortionary factor. IME funding operates like a percentage-based gratuity on Medicare-reimbursed services, akin to a restaurant tip. 

This structure disproportionately benefits hospitals in higher-cost areas and those performing more expensive procedures. Consequently, facilities in regions like the Northeast, where training costs are higher, receive significantly more GME funding than other parts of the country. For instance, New York trains 3.5 times as many residents as Georgia but receives six times more Medicare funding per resident. Moreover, because residents in lucrative subspecialties perform higher-cost procedures, the funding structure discourages programs from expanding primary care slots, exacerbating shortages in essential specialties.

Rather than attempting to rebalance payments through piecemeal restrictions and earmarks with each new increase in funding, a more effective strategy would be to streamline IME funding into uniform payments with enough flexibility to drive slots to areas and specialties of highest demand. This could be accomplished by consolidating DGME and IME payments into a geographically uniform amount, as recommended in Unmatched: Repairing the U.S. Medical Residency Pipeline. Additional adjustments based on geographic and specialty demand could be used to incentivize slot creation in areas of high demand as needed. Currently, lawmakers are considering consolidating GME funding streams into a single block grant–a proposal that the Congressional Budget Office (CBO) estimates could save approximately $100 billion over 10 years. However, it is essential that any such consolidation does not lead to limits on or reductions in the availability of residency slots.

Finally, the draft legislation creates a Medicare GME Policy Council, composed of 13 members appointed by the HHS and tasked with advising the HHS Secretary on the distribution of residency slots, in terms of both geography and area of medical specialty. This council would essentially have the power to recommend the rebalancing of GME positions every five years. The proposal specifies the makeup of this council, which is an important guard against the introduction of special interests and bias into this process. However, given the substantial funds involved, it is important to recognize that such a policy council may not be immune to external influences. To address this concern, we recommend implementing additional safeguards. For instance, a data-driven allocation requirement could ensure that any recommended changes to the allocation of slots be based on objective data on shortage areas or specialty mix. 

Bottom line

America’s approach to GME funding deserves considerable rethinking. The reforms to Medicare GME proposed by the Senate GME Working Group are a meaningful acknowledgment of the program’s shortcomings and a critical incremental step toward addressing America’s physician shortage. However, addressing the root causes of poor distribution will require broader, structural changes to GME funding to better align the supply of physicians with patient demand.

We strongly urge Congress to prioritize residency reform as a key strategy to resolve this critical bottleneck in physician training and ensure a more equitable and effective healthcare workforce for the future.