Graduate medical education (GME), the training of resident physicians, is funded by GME payments to hospitals and health systems, largely from Medicare and Medicaid. The number, specialty, and practice locations1 of future physicians is heavily dependent on how GME positions are determined and placed. In 2015, Medicare alone provided $12.5 billion in GME payments to teaching hospitals. Yet, shortages persist in select specialties, such as primary care, and in rural and underserved areas.
Medicare provides 2 types of GME payments to hospitals: direct medical education, based on the proportional Medicare patient load and the number of resident physicians; and indirect medical education, an add-on to Inpatient Prospective Payment System reimbursements based on a resident-to-bed ratio, such that hospitals with higher Medicare patient loads, resident physician-to-bed ratio, and Medicare reimbursements get paid more.
The indirect payment supports the higher cost of patient care associated with teaching. The Medicare Payment Advisory Commission found only 40% to 45% of the indirect payment was empirically justified and recommended reallocating payments more than the justified amount to new GME programs.2 The Institute of Medicine recommended consolidating Medicare GME payments into a single payment and redirecting 30% to a transformation fund.3
We examined the growth of Medicare GME payments per resident full-time equivalent (FTE) since 2000, variation in payment rates between hospitals, and potential savings from capping Medicare GME.
We used fiscal years 2000 to 2015 Medicare hospital cost reports to calculate annual direct, indirect, and total GME (direct plus indirect) per FTE payments for each teaching hospital. We compared mean payments in 2000 and 2015, using the 2-sample t test with STATA statistical software (version 14, STATA Corp). We studied subcomponents in direct and indirect payments to identify major contributors to the observed trends. We described the variation in total GME per FTE across teaching hospitals in 2015. We excluded teaching hospitals in the bottom or top 1% of per FTE payment. Dollar amounts were inflation-adjusted to 2015 using the Consumer Price Index for all urban consumers.
We also calculated potential savings from capping hospital payments at $150 000 per FTE in 2015. The $150 000 level is based on the Teaching Health Centers GME program—the only federal GME program providing a single payment as recommended by the Institute of Medicine. Medicare hospital cost reports are publicly available, organization-level administrative data sets. This study did not include human participants research and therefore did not qualify for institutional review board review.
Among 1624 teaching hospitals, the mean per-resident GME payment increased from $117 323 in 2000 to $138 938 in 2015 (P < .001) (Figure 1), largely owing to indirect payments (nearly 75% of total payments). The mean Medicare patient load remained relatively stable during the study (0.44 in 2000 vs 0.43 in 2015). The increase in indirect payments was largely owing to inpatient costs rather than resident-to-bed ratios (Figure 2), which grew 9.6%, whereas inpatient reimbursements grew 47.5%.
In 2015, the interquartile range in payment rates across hospitals was $105 761 to $182 233 per FTE, and 573 (47%) teaching hospitals received more than $150 000 per FTE. Medicare provided an estimated $1.28 billion in payments of more than the $150 000 rate.
Medicare GME payments per resident FTE grew nearly 20% from 2000 to 2015, largely driven by increasing inpatient reimbursements. This raises the question of whether linking GME payments mechanically to inpatient reimbursements without assessing the association with teaching costs is sensible policy. The variation between hospital GME rates suggests some hospitals could support GME at lower payment rates. If Medicare GME were capped at the $150 000 rate of the Teaching Health Centers program, $1.28 billion would have been available for redistribution to address other US health workforce needs. Wide rate variation also suggests the need for caution and pacing in GME payment reform, with attention to how payment reductions may affect hospitals receiving lower payments.
Study limitations included analysis of Medicare GME support, not actual training costs, and we did not examine additional hospital characteristics that may contribute to GME payment variation or Medicaid GME payments.
Capping the Medicare GME payment rate would be a limited reform. More comprehensive approaches would involve rethinking GME payment structures, components, and accountability.
Corresponding Author: Candice Chen, MD, MPH, Milken Institute School of Public Health, The George Washington University, 2175 K St NW, Ste 250, Washington, DC (cpchen@gwu.edu).
Accepted for Publication: August 9, 2019.
Published Online: October 7, 2019. doi:10.1001/jamainternmed.2019.4429
Author Contributions: Dr Chung had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
Study concept and design: All authors.
Acquisition, analysis, or interpretation of data: All authors.
Drafting of the manuscript: Chen, Chung.
Critical revision of the manuscript for important intellectual content: Chung, Petterson, Bazemore.
Statistical analysis: Chung.
Administrative, technical, or material support: Chen, Bazemore.
Study supervision: Chen, Petterson, Bazemore.
Conflict of Interest Disclosures: None reported.