On Implementing a National Graduate Medical Education Trust Fund (1997)

The committee's GME distribution proposal simplifies the decisions needed to structure support given to new facilities. The uniform DME per-resident payment could be paid. The current IME formula could be applied to new teaching hospitals using values for the variables from the most recent experience, then divided into halves in the same way as is done for existing teaching hospital IME funding.

Merged or acquired institutions with DME training programs could be paid the uniform per-resident amount or, if available, the existing per-resident amount, whichever is less. The committee considered and rejected a transition payment as a DME incentive to reduce or eliminate resident positions or discrete training programs. Payment would be reduced as the number of residents were reduced without any kind of hold-harmless provision.

IME formula payments could be calculated using the new resident and bed counts, applying the formula as usual. IME historical payments could be based on historical resident numbers transferred from the merged institutions and used, with the new bed numbers, to calculate IME payment for the new entity if it were deemed advisable to provide an even greater incentive to reduce the number of residencies and system overcapitalization through institutional restructuring. Although this incentive might be a real consideration in encouraging institutional consolidation and the development of training economies by reorganization, this kind of activity should be monitored or perhaps transitioned. Rebasing in the future might be advisable if payments became disproportionate to the size of the new programs and institutions.

The committee's plan for distribution of a National GME Trust Fund noted the desirability of including some general, non-Medicare contribution. This is

consistent with the precedent of the 1995 Balanced Budget Act. It is also consistent with the committee's principles, and with the use of some source of funds to address the non-Medicare shortfall in support of GME and related functions. In addition, it makes for a better plan in, for example, leveling the variability due to Medicare caseload in IME payments.

If non-Medicare funds are not included, the committee's plan is less faithful to its principles, but is still an implementable improvement. The approach to uniformity for DME funding will provide support for more sponsoring institutions and is still reasonable in theory, though less so. The expanded eligibility for DME funding provides a GME system that is more open to allowing training opportunities in new and innovative settings. The incentives to continually expand residencies will be diminished by an annually defined fund and the DME changes. The problem of capitated patients outside of the PPS will be eliminated, and AAPCC decisions and GME decisions about the amounts of funding will be separated. Importantly, a controlled academic hospital GME payment system will have been devised for Medicare that will allow consideration of academic health center issues separate from overall hospital or health care services sector issues.

Although the committee did not analyze every past legislative initiative, there was a sense that some effort should be devoted to reviewing the GME provisions of the Balanced Budget Act of 1995. These provisions were enacted by the 104th Congress but were vetoed by President Clinton. Further, the request for advice from the House Ways and Means Committee implied consideration of a similar trust fund presumably reflecting some of the thinking if not the precise provisions of the 1995 act.

The act established five funds using Medicare and general revenue sources. Two funds would continue present Medicare DME and IME distributions, but at reduced levels. Two general revenue funds would allocate a predetermined amount based on the proportion of an institution's payout during an historical period (1992–1994). A fifth fund would pay hospitals in proportion to their service to Medicare managed care (''MedicarePlus'') patients. DME support of consortia would be permitted, but other expansion of DME eligibility was not included. The general funds would not fluctuate with changing resident or bed numbers or institutional Medicare or non-Medicare caseload. Resident number would be capped. The act is more prescriptive for the work force in other ways as well. The MedicarePlus fund would vary only with respect to Medicare managed care and would be unrelated to the size of a training program. This could allow substantial payments to hospitals in regions with high numbers of capitated patients, a small training effort, and little of the public value associated with a developed teaching program or support of unsponsored clinical research and innovation. The increasing amount of revenue for the fifth fund assumed a

rather dramatic increase in the proportion of Medicare beneficiaries joining private managed care plans (Reuter, 1996b). The fund was intended to help institutions that lost money because of this shift. The committee expressed serious concern that such a reallocation would reduce support for academic medical centers and major teaching hospitals, and increase funding for those institutions with minor programs and lesser costs.

The committee questioned the wisdom of a general revenue fund continuing to reward an institution based on its Medicare rather than non-Medicare caseload. The use of the non-Medicare caseload would level overall distribution. It would supply some government educational support for training institutions that currently receive very little help, such as children's hospitals. The committee also thinks that an important justification for the use of general revenues is that private managed care payments to teaching institutions in the future are less likely to reflect the added costs of such teaching. General revenue funding would help compensate for this loss, though only for a portion of the previous extra payments teaching institutions received from private fee-for-service patients.

The committee supports much of the structural change in the GME provisions included in the Balanced Budget Act. Although it reduced Medicare support, the act added a new source of general revenue. In the budget climate of the 104th Congress, and given the problems of the Medicare trust fund, the act was supportive of federal funding for GME and the educational mission of teaching hospitals. If Congress were to decide to how as closely as possible to the act, the committee suggests modifications which are more consistent with its principles and which combine a less prescriptive approach with the elimination of inappropriate incentives and variation. Elimination of the MedicarePlus fund, inclusion of capitated Medicare patients in the Medicare IME formula, and making payment from the general revenue IME and DME funds proportional to an institution's non-Medicare portion of the caseload would improve the effect of the Balanced Budget Act. On the plus side, the act would still include general revenues, have an extended time horizon, limit resident numbers, eliminate some variables, and support consortia.

The committee also considered whether funds could be allocated using the current formulas. However, major flaws have been identified in the present system that, as noted earlier, support consideration of alternatives such as a trust fund. Today's DME distribution draws heavily on historical hospital expenses from as long ago as 1984, which may not be relevant in today's rapidly changing health care system; maintains marked variations in funding among teaching hospitals (ProPAC, 1996a); and, at a time when excess residency positions are a source of concern, encourages increases rather than decreases in the number of residents (Dunn and Miller, 1996; Shine, 1995). The current system limits funding for new educational sites and new types of training institutions. There is

also concern that the growth of Medicare managed care and the loss of payments from capitated patients will result in a substantial reduction of the amount of support available for GME under Medicare. Additional criticisms of present arrangements focus on work force implications: whether the right number and types of generalists and specialists are being trained; whether the sites of training should be located primarily in inpatient hospitals (Eisenberg, 1990); whether responsibility for work force distribution issues should be left primarily to the discretion of individual training institutions; and whether federal funding for the graduate training of IMGs should be continued or controlled. The current system has also been criticized for its overpayment of the Medicare share (ProPAC, 1997) and for its inability to limit uncontrolled, rapid increases in cost.

Leaving the current system unchanged or only slightly modifying it means that the problems identified above would be difficult to remedy or remain unaddressed. The committee discussed possible work force policies that could be encouraged through modifications in the present system. Given the current volatility in the health care marketplace, the importance of the market in defining areas of excess and need, and previous experience with payment incentives, the committee concluded it would be undesirable to further manipulate the current GME payment system as a mechanism for restructuring the medical work force.

A voucher system was suggested to the committee as an alternative distribution approach. Such a system could be designed to give each trainee flexibility in training site selection. It could also fundamentally shift the locus of control for the training of the nation's future health professionals from the training institutions to the students. Trainees could be empowered to select those sites that reflect their assessment of the appropriate balance between quality of program, cost of training, and regional location. It could also work in conjunction with the current matching program which controls resident assignments to training programs. In this case, it could fragment payment to over 7,500 individual residency programs. This would be a substantial change which could present administrative difficulties.

While a voucher system might increase site options such as loosely affiliated or unaffiliated ambulatory care centers, it could pose challenges to the present quality control mechanisms for residency training sites. It could add an unknown number of new sites that would have to be examined and accredited. The challenge would be in expeditiously determining and enforcing just which institutions would be eligible to receive vouchers. The quality, oversight, and supervisory considerations at ambulatory sites are, according to Kassirer (1996) "formidable." These considerations obtain, of course, whenever options are expanded.

If the trainee were to have significant freedom in structuring his or her own training, the voucher system would have implications for program coherence and

ensuring a reasonably consistent national standard that would need to be examined. Imposing some structure seems reasonable, though it likely would result in less competition among programs. Determining new payment amounts for multiple, changing sites would be an additional challenge as would be determining how to continue IME teaching hospital support for residents based in non-hospital programs. There would also be a potential for program disruption if residents could shift locations without appropriate coordination.

The dispensing of GME vouchers by the government could be a direct, powerful tool to influence work force composition. The committee feels that this could be the major effect of a voucher system. Dispensing of vouchers could be an open process, distributing payment differently, but without much more of an effect on the work force than the present national resident matching program. The government could also strictly limit vouchers, basing their distribution on decisions about the total number, educational backgrounds (i.e., IMG) or specialty of trainees. In the latter case, the government could precisely dictate work force composition, for example, creating specified numbers of specialists and generalists, or eliminating IMGs. The support for vouchers by some organizations seems to lead, at least in part, in this direction (Dunn, 1997).

The committee preferred a less interventionist approach to work force issues; the voucher system has some potential benefits in theory, but they may be difficult to realize in practice.

An important sub-issue for the committee was the role of the states in the future funding and direction of GME. There is considerable state activity in support of GME (T. Henderson, personal communication, 3 December 1996; Plumb, 1995; Plumb and Henderson, 1995). States have sought waivers from the current Medicare GME system to pool these funds with those from Medicaid and private sources (GNYHA, 1996; Healthcare Association of New York State, 1996; Minnesota Department of Health, 1996). They have also sought to use these combined funds in different ways, including the development of coordinated statewide educational consortia (Deloitte and Touche, 1996). Some decentralization of GME decision making could be encouraged and enhanced by awarding a limited number of Medicare GME block grants. These could be distributed in accordance with mechanisms developed by states in coordination with the federal government. The committee recognizes the important role of states in addressing different ways of doing things and thus supports using states and organizations (e.g., privately operated consortia) as laboratories for the development of workable vehicles for the distribution of GME funds and for strengthening high-quality training institutions. The design of formulas and policies for distribution, in the context of today's climate of rapid change, is subject to the uncertainties of prediction and unintended consequences. There is a great deal to be said for using pilot projects at the state level as models for

constructive change and for allowing some regional variability reflecting regional conditions. In this regard the committee noted the recently introduced New York waiver which has some conceptual similarities to the committee's plan.

The committee does not favor, however, devolution of the entire GME system to the states through a statutory grant of funding. GME is a nationwide activity and has national implications; it supports a labor market that changes frequently. The differences in state policies and capabilities raise concerns about the creation of 50 separate distribution, training, and work force policies rather than a national system with some local flexibility. The committee also worries that states might naturally favor state institutions and state budget priorities to the disadvantage of private institutions, depending on the strength of federal guidance to the contrary. Concern has been expressed in the past that some states have used Medicaid support of GME as a mechanism to maximize federal support of local academic institutions; once federal dollars were "granted" and assured, federal support might be diverted to other priorities. A stable federal program could be considered a preferable alternative to the ups and downs of multiple state programs.

Educational consortia represent an additional set of considerations. Such consortia have been described as formal partnerships composed of two or more separate institutions involved in GME, formed to reorganize or strengthen medical education, and characterized by shared and joint decision making (Cox and Dower, 1996). Some have cautioned against encouraging consortia as recipients of GME funding because it might lead to the development of redundant organizations simply to receive funds. On the other hand, an effort to structure the GME system to be more responsive to changes in the organization and delivery of health care and to national policy initiatives might justify some consortium arrangements. Consortia might help address the concern that limiting most training funds to only one or just a few sectors of the health care system could bias the focus of training outcomes. In many cases, consortia might also recognize more fully the needs of managed care, ambulatory medicine, and primary care (Van Etten, 1995).

If consortia serve to promote coordination, resource sharing, and efficiency among GME institutions, and if they are formed voluntarily to serve members' needs, then the committee thinks it makes sense to include them as recipients of GME funds. If there is a desire to promote national GME and work force policies to encourage primary care, ambulatory training, reductions in resident positions, physician distribution, and minority recruitment goals, and if consortia could be structured to make progress toward these objectives, then it might be reasonable to provide active encouragement by some incentive add-on along the lines of a 1% or so extra DME payment. Consortia with adequate authority and

control of resources could be powerful forces for quality improvement. If consortia eliminated lower quality programs and caused other programs to coordinate, adapt to change, and improve education in an accountable way, as has been suggested, enhanced payment might be a good investment (Kelly et al., 1994; Kochar and Cooper, 1996).

Alternatively, the recipients of DME payment (at present only teaching hospitals) could be required to meet certain objectives similar to those advanced for consortia as a condition of payment. If consortia are encouraged, attention should also be given to their structure and activities. Consortia might be required to consist of an equal partnership of community institutions, including a required nucleus of teaching hospitals with medical schools (except in cases of distance or lack of relevance of the medical school to the market in question). Other components should include at least some of the following: managed care organizations, ambulatory centers, community health centers, group practices, universities, or other institutions involved in graduate training. Consortia should agree to receive the funds and should have a formal arrangement for a coordinated program and distribution of payments in support of GME.

The committee urges caution in moving too quickly in expanding consortia. At present, GME funding is based on the recognition of a federal obligation to help support training and the mission of teaching hospitals. Additional policy objectives and conditions should be added with care. Consortia are something of a work in progress, though the movement of AOA to mandatory consortia may provide better information and understanding in the near future. The present evidence that they can return the effort and resources expended in their implementation may not justify a substantial shift in the GME program. Review of the activities of current consortia reveals that they are few in number, lack broad membership, and do not generally have the authority, or in many cases the inclination, to implement national goals. Nevertheless, consortia are an attractive concept and continued exploration of their value seems justified.