Minimum Instructional and Program-Specific Administrative Costs of Educating Residents in Internal Medicine

Background  The cost associated with education of residents is of interest from an educational as well as a political perspective. Most studies report a single institution's actual incurred costs, based on traditional cost accounting methods. We quantified the minimum instructional and program-specific administrative costs for residency training in internal medicine.

Methods  Using the Accreditation Council for Graduate Medical Education program requirements for internal medicine as minimum standards for teaching and administrative effort, we quantified the minimum instructional and administrative costs for sponsorship of an accredited residency program in internal medicine. We also analyzed the impact of resident complement and program curricular emphasis (outpatient, inpatient, or traditional) on the per-resident cost. The main outcome measure was the minimum annual per-resident cost of instruction and program-specific administration.

Results  Using the assumptions in this model, we estimated the annual cost per resident of implementing the program requirements to be $50 648, $35 477, $28 517, and $26 197 for inpatient intensive residency programs with resident complements of 21, 42, 84, and 126, respectively. For outpatient intensive residency programs of identical resident complements, we estimated the annual per-resident cost to be $58 025, $42 853, $35 894, and $33 574 for similar resident complements. Fixed costs mandated by the program requirements, which did not vary across program size or configuration, were estimated to be $640 737.

Conclusions  There are fixed and variable costs associated with sponsorship of accredited internal medicine residency programs. The minimum cost per resident of education and departmental administration varies inversely with program size within the sizes examined.

THE PAYMENT for costs associated with graduate medical education (GME) has historically been linked to payment for patient care. During the era of cost-based reimbursement, the allowable direct costs of GME were paid to institutions that sponsored GME programs. Methods for the identification and itemized reimbursement of direct costs of GME by Medicare were refined in the prospective payment system implemented in 1984. Reimbursement for the higher costs associated with care provided in teaching hospitals was provided in cost-based reimbursement systems, as well as in the prospective payment system, with the implementation of the indirect medical education payment method. This method, often interpreted as reimbursement for the inefficiencies of care provided in the teaching setting, reflects the higher technology mix, higher level of severity of illness, and higher level of uncompensated care seen in most major teaching institutions.¹ During the past 10 years, teaching hospitals have seen erosion of the support for GME from all payers. In 1999, most explicit GME reimbursement for teaching hospitals was derived from the federal government, and the single largest component of that support derived from the traditional Medicare system.¹ The Balanced Budget Act placed substantial limitations on expansion of resident numbers through funding caps and mandated a phased reduction in indirect medical education reimbursement over 5 years.¹

Despite these and other changes in reimbursement, the Medicare program faces long-term challenges in the control of growth of expenditures.² While the rate of growth of Medicare expenditures has diminished, pressure to redefine the Medicare program has included reexamination of the role of Medicare in the payment of GME costs.¹ Concern for the long-term viability of the Medicare trust fund has resulted in exploration of removal of direct costs of GME reimbursement from the Medicare program by the Bipartisan Commission on the Future of Medicare. Although a final recommendation was not forthcoming from this commission, the range of options examined by both the commission and MedPac include major restructuring of GME reimbursement.

This and other threats to stable and secure funding have raised questions concerning the stability of the GME enterprise. These factors, coupled with the proposal to fund resident salaries and fringe benefits through an appropriations process, motivated us to reexplore a commonly asked, but as yet unanswered, question. How much should it cost to educate a physician at the residency level?

The institution-specific direct costs of educating residents have been studied.^3-6 Authors of these studies attempted to quantify components of the total cost of education of residents in individual programs or components of programs, usually in single institutions. Extrapolation of these findings to project a national figure or a range of cost per resident has been problematic for many reasons. These reasons include variation in (1) the structure of programs within a given specialty among institutions; (2) the elements and methods of education in different specialties; (3) the payment mechanisms for faculty and other participants; (4) assumptions regarding the allocation of faculty effort; (5) administrative overhead and allocated institutional overhead; (6) the costs of the noninstructional components of support of educational programs (for example, financial support of clinics); and (7) the expense of medical malpractice insurance.⁷

Further evidence of the difficulty in comparing existing incurred costs may be inferred by the range of direct costs of GME provided by Medicare to teaching hospitals. The range is very wide,⁸ suggesting the likelihood of substantial accounting differences in the base year across the country, as well as a wide range of expenditures incurred by institutions⁹ in direct support of GME.

We attempted to construct a theoretical standard annual minimum cost for components of the total direct costs of educating residents. Rather than using existing cost report data or surveys of faculty effort, we approached this question by using an external standard for the structure of educational programs. This standard is used both by GME-sponsoring institutions as acceptable definitions of minimum standards for offering an accredited residency program, and by the Health Care Financing Administration as minimal standards for reimbursement of educational costs by Medicare. These minimal standards are the program requirements promulgated by the Accreditation Council for Graduate Medical Education (ACGME) and the respective residency review committees. The standards should provide information that would permit deduction of the minimal resources required to achieve accreditation of educational programs. If these standards are considered a threshold for reimbursement by Medicare, then it appears reasonable to assume that costs required for their satisfaction should establish a base for compensation. These costs should inform the debate regarding the appropriate level of support to be provided to institutions sponsoring GME programs, should the direct costs of GME be moved to an appropriations process by Congress. Finally, were an all-payer pool to be considered for GME programs, a base payment for instructional costs that is standardized would be helpful in establishment of equitable payment.

In light of the limitations of a standardized assessment of the components of GME costs, which vary widely across institutions (eg, overhead allocation, infrastructure and nonfaculty costs associated with ambulatory clinics, and opportunity costs born of the inefficiencies of residents), we chose to assess the annual economic commitment required to satisfy the direct costs of organization and instruction (called the minimum instructional and program-specific administrative costs [MIPSACs]) mandated by accreditation standards. We analyzed the MIPSACs in internal medicine residency training for the following reasons. First, the cohort enrolled in general internal medicine residency training is the largest single cohort of residents, with 21 130 (21.7%) of the 97 383 residents and fellows on duty in the United States.¹⁰ Second, the program requirements in internal medicine are explicit in many dimensions (eg, numbers and types of rotations, mandatory minimum teaching requirements, and specific ratios of faculty to trainees), permitting the estimation of cost per unit of educational function per trainee.¹¹

On the basis of the program requirements and the assumptions outlined in the Methods, we calculated an annual MIPSAC per resident and examined the impact of number of trainees and ranges of program design on these costs. Finally, we projected the cost per trainee for instruction and resident salary and fringe benefits by means of mean salary figures from a national sample.

We used the current program requirements for residency programs in internal medicine¹¹ (effective July 1998) to design a spreadsheet that modeled the fixed and variable direct costs of a residency program of 3 years' duration. The model contained the following categories of rotations or services: (1) inpatient internal medicine rotations (ward or floor rotations); (2) inpatient critical care rotations (coronary care and intensive care); (3) inpatient consultation rotations; (4) emergency medicine rotations; (5) ambulatory block rotations in a physician's office (this rotation may be with a generalist or subspecialist); and (6) ambulatory continuity clinic (this function is integrated into all rotations other than emergency medicine and critical care).

For each rotation category, the number of house officers and the ratio of faculty to resident was assigned. Formal teaching time (time defined as instructional time with no other source of compensation, ie, patient care revenue) was established by reference to the program requirements. In the few instances where the program requirements are not explicit, we made assumptions on the basis of our experience. For example, during critical care rotations, teaching rounds were budgeted to occur daily rather than the minimal standard of 3 days per week. The specifics of these assumptions are enumerated in Table 1.

The unit faculty cost of provision of each educational service was calculated with a blended medicine faculty rate (Table 2), on the basis of a faculty mix of grade and salary for the United States.¹² A fringe benefit rate of 25% was used for estimation of fully absorbed faculty salary costs for all faculty-related expenses.

Fixed direct costs of program-specific administration of the educational program were calculated with the assumptions contained in Table 3. Of note, no attempt was made to calculate overhead costs, such as rent for administrative offices, allocated institutional administrative overhead, or administrative costs associated with the operation of resident clinics or office practices.

The cost of faculty instruction in continuity clinics and other ambulatory settings was included in this calculation of instructional costs. For example, in the continuity clinic, a resident-faculty ratio of 4:1 was used under the assumption that the faculty members do not see their own patients independently during sessions involving residents. Therefore, all faculty time spent in clinic is considered instructional time. However, in an office setting during an ambulatory block rotation where the faculty and residents provide service to patients, the faculty receive compensation for these clinical services. Therefore, only specific, mandated instruction time, rather than total contact time, is considered teaching time. In this fashion, patient care services provided by the educational team (patient–trainee–faculty member) are compensated through billing for physician services in the teaching setting, and only mandated instructional time is compensated as a direct instructional cost of GME. No attempt was made to quantify the complex balance between the opportunity cost of faculty time devoted to teaching residents in the office setting and the benefit to the faculty physicians of having residents assist with the care of their patients.

The program requirements stipulate an affiliation with a medical school accredited by the Liaison Committee on Medical Education or demonstration of equivalent facilities and programs. Whether for a university-based or community-based program, this level of linkage usually involves the education of third- and fourth-year medical students. In modeling the teaching faculty units of service, the teaching team on the inpatient internal medicine teams were assumed to include 1 supervising resident, 2 postgraduate year 1 house officers (interns), 2 third-year medical students (1 per intern), and a fourth-year medical student (subintern or acting intern). No faculty time or expense has been added to the model for exclusive support of undergraduate medical education.

Variable instructional costs were established through calculation of the number of units of each type of faculty service (general internal medicine teaching rounds, critical care unit rounds, continuity clinic supervision, etc) required on the basis of the assumed distribution of rotations. Three models of program structure are presented. The first, called inpatient intensive, has the minimum amount of ambulatory clinic exposure (total, 33% of time; equal to 1 year of training), with the maximum inpatient and critical care unit rotations permitted. A single half-day continuity clinic per week is present in this model. The second, called ambulatory intensive, has the minimum amount of inpatient experience (33% of time; equal to 1 year of training), coupled with the maximum allowable number of ambulatory rotations. This includes 2 half-days per week of continuity clinic and the maximum of 2 years (total, 67% of time) of outpatient educational experience. A minimum of 3 months of critical care was used in this model. The third is called a traditional model. In this model, approximately 40% of the resident's time is spent in the outpatient setting and 60% in the inpatient environment. Allocation of time in each rotation category is contained in Table 4.

The rotational schedule is defined in 13 four-week blocks per year, to standardize the number of weeks per month (4) and teaching hours per rotation. Vacation time was allocated as 1 month (4 weeks) per year. Thus, each trainee is allocated 12 four-week rotations per year, 4 weeks of vacation per year, for 36 four-week "months" of actual training during the 3 years of residency. Resident research rotations were not separately identified. The models were examined for the impact of program size by examining the per-resident costs in programs with total resident complements of 21, 42, 84, and 126 (7, 14, 28, and 42 residents at each level). These are multiples of the number of rotations per year (plus 1 for vacation) and approximate the range of minimal size (the minimum size permitted by the program requirements is 6 residents per year, or 18 total residents¹¹) and small, moderate, and large program size.

Faculty salary assumptions are presented in Table 2. Of note is the slightly higher salary for "ambulatory" rotation preceptors. This figure is derived from the mean salary for practicing internists, which includes generalists and subspecialists.¹³ Since our ambulatory rotations (noncontinuity clinic) are assumed to be a mix of electives and mandatory rotations in both general internal medicine and the subspecialties, this figure provides an estimate of average salary. Critical care and inpatient subspecialty rotations were assigned the mean of the procedure-oriented specialties available in the 1999 AAMC Databook.¹² The inpatient teaching rounds (ward rounds) salary was assigned the mean salary of all 4 faculty grades (instructor, assistant, associate, and professor).

The MIPSACs of residency programs in internal medicine as estimated in this model is presented in Table 5. The fixed costs for a residency of any size are in excess of $640 000 per year under the assumptions included in this model in implementation of the internal medicine program requirements. Variable costs, the costs that are incurred on the basis of the number of trainees and the faculty time required to provide instruction, increase slightly with increasing numbers of residents. This is because of the allocation of additional administrative expense of program director and associate program director costs to support the increased number of trainees.

The annual per-resident fixed costs decrease dramatically as resident numbers increase, while variable costs per resident vary only slightly as the number of residents increases. Thus, the MIPSAC per resident varied inversely with the number of residents in the program. The annual MIPSAC per resident in a traditional model with a resident complement of 21 (7 residents per year) is $51 061. This decreases to $35 890, $28 930, and $26 610 for program complements of 42, 84, and 126 residents, respectively. The fixed, variable, and total annual MIPSAC for each model are graphically displayed in Figure 1, Figure 2, and Figure 3.

The differences calculated between inpatient intensive, outpatient intensive, and traditional models are less striking. The major factor influencing the differences in instructional component of costs is the assumption that ambulatory rotations (other than continuity clinic) have a resident-faculty ratio of 1:1. In contrast, the model assumed an inpatient consultation resident-faculty ratio of 2:1, decreasing the instructional component of cost per resident on inpatient rotations (more prevalent in the inpatient and traditional models) in comparison with outpatient rotations (more prevalent in the ambulatory model). In addition, a second continuity clinic is assumed in the outpatient intensive model, increasing faculty costs beyond those of the inpatient intensive model.

Also of note in Table 5 is the specific cost per resident when resident salary and fringe benefits are included. The range of cost per resident for programs based on a traditional model is $95 143, $79 972, $73 012, and $70 692 for resident complements of 21, 42, 84, and 126, respectively. The per-resident costs for MIPSACs plus resident salary and fringe benefit costs for an outpatient intensive residency program are $102 107, $86 935, $79 976, and $77 656 for resident complements of 21, 42, 84, and 126, respectively. Costs per resident in a traditional model are intermediate, approximately $400 more per year than the inpatient intensive structure at comparable program sizes. The ratio of the highest to lowest per-resident MIPSAC plus resident salaries and fringe benefits, based on program size, is 135%, 131%, and 135% for the inpatient intensive, outpatient intensive, and traditional models, respectively.

The federal Balanced Budget Act of 1997 imposed new limits on the amount of public financial support that will be provided in the future for residency programs under Medicare.⁸ Regulations based on the Act place caps on reimbursement for GME; they move away from full cost reimbursement and lead the way toward uniform reimbursement across all programs. These restrictions, combined with the growing resistance of private insurers to compensate teaching hospitals for their higher costs, are forcing leaders in academic medicine to confront the real cost of GME. The present study was designed to inform policymakers of the challenges that will be faced by many residency programs as they implement the current Accreditation Council for Graduate Medical Education program requirements for residencies in internal medicine in the context of restricted financial support for GME.

We modeled the minimal budget that would be needed to implement the Accreditation Council for Graduate Medical Education program requirements for residencies in internal medicine based on cost estimates of the requisite program administration staffing, educational time mandated for faculty, typical payroll costs of residents' salaries, and fringe benefits. Because the program requirements included dedicated core resources for every accredited program regardless of resident complement, the cost model distinguished the fixed cost of these core resources from the variable costs that are tied only to the number of residents in a particular program. Examples of the fixed costs that were included in the model for every program include the required time commitments of the program director, department chair, assistant directors, program administrator, secretary, and faculty participation at mandated conferences. On the other hand, variable costs included not only the resident payroll cost but also the faculty time required for supervision and direct education of every resident.

The total annual fixed cost for a typical program was estimated at about $641 000. This cost is fixed, regardless of the number of trainees. The variable cost per resident, including salary and benefits, ranged between $65 000 and $72 000 per year depending on the balance of inpatient and ambulatory teaching. Thus, the cost of a program with 21 residents is set at about $2 million per year. However, this analysis projected only what was referred to as the minimal instructional and program-specific administration costs (MIPSAC). In addition to residents' payroll costs, these minimal costs included program administration tasks and mandated instructional activities taken directly from the program requirements. However, true excellence in GME demands a vast array of other resources and professional activities that were not incorporated in the model presented herein. The potential cost of these resources cannot be ignored.

Figure 4 outlines 6 major components of the cost of GME. The MIPSAC developed in the present study includes only payroll costs and the cost of the absolute minimum program activity at the departmental level. However, overall institutional educational and administrative support for GME was excluded. Similarly, no provision was made for the cost of space allocated to departmental libraries, call room space, office space, and other facilities dedicated to resident teaching. Finally, no attempt was made to estimate the value of volunteer faculty time in situations where faculty effort exceeds the value of the patient care services provided by residents to their patients.

If one accepts the important distinction between a residency program's fixed and variable costs, then this model raises troubling questions about the cost-effectiveness of smaller residency programs in relation to larger programs. A program director will minimize the average fixed cost per resident by spreading a program's fixed costs over the largest possible number of trainees. If the variable cost per resident is similar across programs of different size, it is easy to see that the larger residency programs will be the most cost-effective. Their average fixed cost per resident will be lower. On the other hand, it is possible that there are other fixed costs that might be expected at some thresholds of increasing program size in certain settings, or some inefficiencies that are unique to extremely large programs. For example, very large programs may require additional staff costs for scheduling or more elaborate physical teaching facilities that are unnecessary in small programs. Large programs are more likely to be located in urban areas with higher payroll costs than programs in small cities. Large programs may require travel time for commuting or rotating among sites that reduces the productivity of faculty and residents. The resources required by programs of different size are not addressed directly by the program requirements and, therefore, were not examined in this model. While it appears that large programs are theoretically more cost-effective, further empirical analysis will be needed to confirm this.

Although the results of this study underscore the high cost of training residents, one cannot assume that every teaching hospital offering an internal medicine residency program must receive full public support for all instructional and administrative costs. The GME-related teaching and program administrative services provided by faculty are funded through multiple sources. Other funds for faculty teaching flow from the patient care revenue of faculty practice plans, Veterans Affairs hospital support for faculty, federal awards such as Title VII training grants, and private philanthropy. Furthermore, the hidden teaching time of faculty supported directly by research grants and the substantial efforts contributed by volunteer faculty are exceedingly difficult to quantify. The issues of who should reimburse the different costs of training, which organizations should receive which funds to offset which costs, how funds should be distributed among the units and individuals actually incurring the costs within the organization, and related concerns are beyond the scope of this analysis.

It is possible to debate some of the specific assumptions in this cost model. However, the overall analysis clarifies the consequences of instituting formal requirements across all programs of varying size. The goal of these requirements is to ensure the quality of GME by mandating that a minimum level of resources be allocated to teaching. However, the annual fixed costs involved in satisfying these requirements will place a greater burden on smaller programs because reimbursement for GME traditionally has been budgeted on a per-capita basis. It is essential that program directors, leaders in academic medicine, and policymakers understand the implications of this cost structure as they develop models for financing GME and negotiate reimbursement for their own residency programs.

Accepted for publication September 14, 2000.

Corresponding author and reprints: Thomas J. Nasca, MD, Room 108, College Bldg, 1025 Walnut St, Philadelphia, PA 19107 (e-mail: thomas.nasca@mail.tju.edu).