Estimating the cost to departments of medicine of training residents and fellows: A collaborative analysis

We seek to define what it actually costs departments to provide an educational program that trains physicians effectively and fulfills the relevant requirements of regulatory bodies. This calculation bears no relation to what the hospital receives from the Medicare program or other payers (DME or IME) or what it saves by not having to pay in other ways for work that physicians-in-training perform. Hospital revenue for GME comes primarily from Medicare, which provides DME payments and IME adjustments to reimbursement rates to hospitals. The formula that determines the amount of this income to the hospital depends on the ratio of physicians-in-training to Medicare beds as well as other historical factors. In addition, the savings to the hospital for having physicians-in-training onsite depend on the costs of physician extenders, the willingness of private practice and full-time faculty physicians to cover patients without extenders or residents, and other local factors. Because hospital revenue and savings related to GME vary widely by region and circumstance, revenue and savings cannot serve as a basis for reimbursing departments for their expenses in providing training. The costs that departments incur in providing training are unrelated to payments hospitals receive for training. To determine whether to have GME programs and how large they should be, hospital executives need to examine true revenues and savings from these programs together with their true costs, including appropriate payments to departments for administering training programs and teaching the residents and fellows as well as the added cost of patient care provided by physicians-in-training.

The method of estimating departmental costs of teaching must be readily understandable to hospital administrators and department faculty, allowing them to identify the rationale behind the calculations and apply them locally with minimal effort and controversy. To achieve this end, wherever possible we seek to tie departmental costs of teaching directly to regulatory requirements. This approach should reduce disagreements between departments and hospital executives because hospital administrators are accustomed to defining expenditures needed to comply with regulatory mandates, such as state hospital certification, Joint Commission for Accreditation of Healthcare Organization inspections, and the like.

Departmental expenses must include all components of cost associated with administration of and teaching in the programs. For example, the cost of faculty time should include not only salary and fringe benefits, but also institutional (university or departmental) overhead expenses, office space rental, and costs of faculty staff support; these costs are all prorated to the proportion of a faculty member’s effort devoted to GME. Additional costs include administrative expenses such as staffing to coordinate recruitment, orientation, schedules, documentation of evaluations, and space rental for the education office. These administrative costs must include direct salaries and fringe benefits as well as all of the appropriate overheads.

Many training programs include a major component based at one or more Department of Veterans Affairs (VA) facilities. The costs of supervising and teaching physicians-in-training at VA medical centers should not be included in negotiations with non-VA hospitals because the VA central office provides funding to individual medical centers to pay for educating the physicians-in-training working there. However, because many GME programs that include VA training positions are administered from non-VA teaching hospitals, the program must expend resources on recruitment of physicians-in-training who will serve at the VA for part of their training as well as on oversight of the VA curriculum and evaluation of physician-in-training performance at VA sites. Thus, some of the administrative costs involved in VA training will need to be supported by non-VA hospitals.

A prior study of costs of GME developed sophisticated estimates of administrative costs per physician-in-training, which were found to vary inversely with program size. Although these published estimates may serve as useful guidelines, it appears likely that administrative costs may be difficult to standardize from place to place. Recruitment costs will vary widely among programs. Some programs will need to invest a great deal of time and effort helping international medical graduates with visa issues. Programs with residents located in a single institution may have lower administrative costs than programs that manage resident and fellow rotations across multiple institutions. Oversight of large numbers of residents rotating through many different community-based sites may require more administrative effort than oversight of physicians-in-training at only a few such sites. As discussed above, programs with a large VA component may need to cover central program administrative costs for physicians-in-training while they are at the VA. As a practical matter, if the hospital administration seeks to contest the administrative costs, the budget for this function can be reviewed locally, line by line.

All costs reported are for fiscal year 2003, which, for the University of Pittsburgh School of Medicine, is July 2002 through June 2003. Using these principles, costs are categorized as follows.

Although different approaches have been used at different medical centers, there is remarkable unanimity in the cost estimates. Nasca et al estimate that it costs departments of medicine $34 000 per year to train a resident, with costs rising to higher levels in smaller programs because of reduced efficiency in spreading administrative costs over many residents. The Nasca group does not provide an estimate for fellows. The Hunter Group, in conjunction with the places it has consulted, estimates about $34 000 per resident and $17 500 per fellow. In these analyses, the estimates for costs of training fellows were heavily influenced by the reduced level of DME and IME revenue provided by Medicare for subspecialty fellows as opposed to residents. These estimates are outlined in Table 6, Table 7. Interestingly, if the Hunter Group analysis were applied to the University of Pittsburgh training program (141 FTE at the non-VA hospitals multiplied by $34 000 per resident plus 118 FTE multiplied by $17 500 per fellow) together with the administrative office costs, the cost for teaching would total $8 134 000, a value that differs from the estimates from time studies by 11%. The difference in the calculated results comes from a lower estimate of the costs of fellowship training by the Hunter Group, which, as stated above, was based on estimates of GME funding as well as on actual costs of teaching. Time estimates at Creighton University came up with about $34 000 per resident and $24 000 per fellow. At Creighton, the costs for educating residents and fellows may be increased because the programs are small and there are high fixed costs. The University of Pittsburgh time studies led to an estimated cost of $35 000 per resident. The annual cost of training a fellow is also approximately $30 000, using the RRC-IM mandates as the basis for the calculation. It is interesting, in this regard, that the VA allocates $43 000 per year to its hospitals above and beyond the salary and fringe of each physician-in-training to train each resident. (The figure quoted varies by region. For VISN 4 [Pennsylvania, Delaware and parts of West Virginia] it was $42 911 this year.) In the VA system, payments for resident and for fellow education are identical.

In summary, the consensus approach for estimating the costs of providing GME developed here sums up local administrative costs, costs of fellowship training driven by RRC-IM mandates, and the 3 components of residency training costs including program leadership and key faculty, subspecialty faculty, and an estimate of the cost of providing teaching by the rest of the faculty. Because time studies at two institutions agree with those of the Hunter Group, it seems entirely reasonable to use the Hunter Group approach to estimating this component of residency training, which has the advantage of being formula-driven.

Although it is very difficult to determine exact costs to departments of internal medicine for providing GME, the values developed using several different methodologies at multiple academic medical centers agree, to a reasonable extent. Of course, the costs of some of these components will vary at different institutions. The approach outlined here is simple and related as directly as possible to regulations governing training programs and appears likely to be applicable to any U.S. academic medical center. Similar approaches along these lines can also be developed for training programs in other disciplines, with modifications that relate directly to the circumstances of and regulations governing training in these disciplines.

We thank Judith Lave and Nicole Buckley for help in obtaining national statistical data, Wishwa Kapoor and Walter O’Donohue for help with time studies, and the members of the APM consensus group (listed with the departments they chair): Jonathan Ravdin, MD, University of Minnesota School of Medicine; William Mitch, MD, University of Texas Medical Branch, Galveston; William Henrich, MD, University of Maryland School of Medicine; Lawrence Gardner, MD, University of Miami School of Medicine; Daniel Banks, MD, Louisiana State University School of Medicine at Shreveport; Donand DiPette, MD, Texas A & M University College of Medicine; Thomas DuBose, Jr., MD, Wake Forest University School of Medicine; John Feussner, MD, Medical University of South Carolina; Eric Gall, MD, Rosalind Franklin University of Medicine and Science, The Chicago Medical School; Nicholas LaRusso, MD, Mayo Clinic College of Medicine; Robert Luke, MD, University of Cincinnati College of Medicine; John O’Connell, MD, Wayne State University School of Medicine; Charles Sanders, MD, Louisiana State University School of Medicine at New Orleans; David Steward, MD, Southern Illinois University School of Medicine) for their ideas and critiques of approaches.