Discussion

A TRUST FUND

The committee endorses the concept of a national GME trust fund as a mechanism that aggregates funds from diverse sources and allows support of a range of activities that might be questionable from a particular funding source. Teaching hospitals are threatened because they are becoming less able to transfer costs associated with GME to non-Medicare patients through higher prices. Adding general, non-Medicare funding to Medicare funding for a GME trust fund addresses this shortfall and appropriately supports the principles and practices identified in this report. This important concept has already been considered by Congress.

By changing from an open-ended entitlement to a closed-ended fund, a GME trust fund imposes a structure that limits available monies in a clear and predictable fashion, providing a controlled payment each year. Such a system would minimize the temptation to manage expenditures through formula changes that would be confusing and potentially disruptive. The system should be designed to provide greater local flexibility on work force matters and incentives to base the number of residents on market conditions rather than payment rules. The flaws in the current system that call for change are discussed under "Additional Issues." Briefly, these are: excessive DME payment variation; overemphasis on in-hospital and under-emphasis on ambulatory and managed care training; uncontrolled payments for additional residents; and loss of GME to managed care plans.

The crucial factors for the success of the trust fund are its adequacy and stability and the continuation of Medicare's commitment to meet its responsibility, as both a governmental program and the major national hospital payer, to contribute fairly to all the costs of the hospital system, not just the marginal costs of the least expensive providers. Monitoring should be required to ensure that the fund remains adequate to assist academic health centers. The fund



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund Discussion A TRUST FUND The committee endorses the concept of a national GME trust fund as a mechanism that aggregates funds from diverse sources and allows support of a range of activities that might be questionable from a particular funding source. Teaching hospitals are threatened because they are becoming less able to transfer costs associated with GME to non-Medicare patients through higher prices. Adding general, non-Medicare funding to Medicare funding for a GME trust fund addresses this shortfall and appropriately supports the principles and practices identified in this report. This important concept has already been considered by Congress. By changing from an open-ended entitlement to a closed-ended fund, a GME trust fund imposes a structure that limits available monies in a clear and predictable fashion, providing a controlled payment each year. Such a system would minimize the temptation to manage expenditures through formula changes that would be confusing and potentially disruptive. The system should be designed to provide greater local flexibility on work force matters and incentives to base the number of residents on market conditions rather than payment rules. The flaws in the current system that call for change are discussed under "Additional Issues." Briefly, these are: excessive DME payment variation; overemphasis on in-hospital and under-emphasis on ambulatory and managed care training; uncontrolled payments for additional residents; and loss of GME to managed care plans. The crucial factors for the success of the trust fund are its adequacy and stability and the continuation of Medicare's commitment to meet its responsibility, as both a governmental program and the major national hospital payer, to contribute fairly to all the costs of the hospital system, not just the marginal costs of the least expensive providers. Monitoring should be required to ensure that the fund remains adequate to assist academic health centers. The fund

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund could then be thought of generically as a GME Prospective Payment System (PPS) and more specifically as a mechanism to preserve the viability of academic health centers and GME. Management of expenditures in this way requires the fund to be distributed more evenly, be related to teaching intensity, and be responsive to information about the financial health of teaching hospitals. There is the risk, of course, that leveling reimbursement may also level centers of excellence. The GME Trust Fund would allow separation of at least a portion of GME payments from Medicare inpatient hospital DRG rates. The present IME formula directly links Medicare payments to the level and complexity of the Medicare caseload and the resident-to-bed ratio; payments are also affected by the annual DRG payment rate update, which reflects decisions based on the condition of the overall hospital industry. A predetermined, fixed trust fund would establish a payout controlled by a different set of policy variables. For example, the trust fund could distance IME funding from PPS considerations. The annual DRG update would have no impact; only the trust fund update would affect payment. The AAPCC includes hospital costs with associated GME costs to fee-for-service Medicare beneficiaries. These costs are included, in turn, in Medicare per capita payments to health plans or managed care organizations that enroll Medicare beneficiaries on a capitated basis. Hospital payments from these plans are negotiated separately and are not made through DRG rates (the PPS). These payments need not, and in most, if not all cases, do not include the GME subsidy. The subsidy remains within the plans. This issue of AAPCC diversion of GME payment from teaching hospitals to plans would become a moot point with a trust fund because any aggregate financial impact of capitated patients on GME hospital payment would be eliminated. Once a decision on the National GME Trust Fund amount was made, that amount would be distributed, no more, no less. The percentage of Medicare patients joining capitated plans could not affect the total distributed. As noted, a coefficient to recognize the proportion of Medicare capitated patients in a hospital could improve proportional distribution to that facility from the defined total National GME Trust Fund. The AAPCC would, of course, continue to reflect any GME payments from the fee-for-service sector, but that could not reduce the Trust Fund. Congress could decide whether this arrangement was appropriate. In summary, increasing the separation of the hospital rate and GME decisions encourages consideration of what the different policy variables should be, how they should change over time, and how to address the separate problems of teaching hospitals and their GME functions. PRINCIPLES AND GME The committee's principles guided its decisions on distribution of GME funds. These principles specified the public value of GME and of its direct and indirect components, and the important distinction between direct support of training and indirect support to maintain the teaching environment. Failure of

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund public support for these principles would have negative consequences for major medical centers, for the quality of professional personnel, and ultimately for the quality of health care itself. A distribution plan should also reflect the principles of cost sharing among payers, of comparable payments across the nation, of responsiveness to changes in the health care delivery system and market, of providing gradual transitions, and of the need for stability in a system with long training lead times. PAYMENT TO DME INSTITUTIONS DME payments currently are made to teaching hospitals. A teaching hospital is a sponsoring institution, defined in the American Medical Association's Graduate Medical Education Directory (1995) as "an established formal organization having the primary purpose of providing educational and/or health care services (e.g., a university, a medical school, a hospital) that assumes final responsibility for a program of graduate medical education." The committee accepts this definition with the added conditions that the sponsoring institution: assume responsibility for the entire educational program; be the active paymaster for the residents by making assignments and schedules and paying resident stipends; assume responsibility for regulatory compliance; prevent double payments; and ensure that payments follow costs. Given changing conditions and the development of new institutions, some flexibility is desirable in making DME payments. The committee concluded that eligibility for direct DME payment should be expanded to additional institutions that meet the definition of a sponsoring institution. Teaching hospitals that receive the total per-resident amount and that currently, depending on local conditions, negotiate a pass-through of all or part of this amount or simply pay all or part of the salaries and benefits of their programs' residents at affiliated sites, could continue to do so. Additional kinds of sponsoring institutions could include independent institutional training programs that show responsibility for the requirements for a training program(s), such as those of ACGME/AOA or others. According to one source, for example, as many as five or six managed care organizations currently functioning as affiliated training sites might compete and provide quite interesting models as new institutions designing and implementing residency programs and meeting the criteria for becoming a sponsoring institution (G. Moore, personal communication, 20 February 1997). As already noted, these and other new sponsoring institutions would have the continuing responsibility for paying resident stipends during the training period and for working with the Secretary of Health and Human Services (HHS) to devise ways to avoid double payments if multiple institutions were involved. Because IME payment is made through the PPS for the support of teaching hospitals, only DME payment eligibility would be expanded; at some point, however, IME payment eligibility may need to follow DME funding out of the hospital. For IME to keep pace with the changes in the delivery system and to

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund ensure that GME equips trainees with the skills necessary to function in that new system, the committee considered whether IME support, as well as DME support, should be distributed more widely. For the time being, the committee suggests that a study be undertaken to explore the concept and perhaps define ways to distribute IME funding to any sponsoring institution. Doubtless, teaching hospitals would remain the usual sponsoring institutions for DME, but medical and osteopathic schools, health maintenance and managed care organizations, group practices, ambulatory centers, universities, and consortia might also qualify to receive payment directly. Consortia could actually be encouraged if they made a case for enhancing national policy objectives and it was deemed desirable, after careful consideration of the implications, to enlist them in pursuing additional GME or work force goals through the payment system. The AOA's plan to structure osteopathic GME into mandatory consortia deserves careful observation in this regard. These consortia could have a positive effect on work force and educational quality (Wickless, 1997), depending on the AOA's policy and the vigor of its implementation. A STANDARD PAYMENT The committee sought to promote consistency. The distribution of DME funds should continue on a per-resident basis, but an argument can be made for standardizing the payment. This argument depends on the view that current exceptional payment variability is, prima facie, difficult to justify. As noted earlier, the 1984 DME cost base reflected allocations and program variables that may be obsolete or were somewhat idiosyncratic at the outset (Anderson, 1996; COGME, 1988; PPRC, 1993). If existing nonteaching hospitals or entirely new training facilities are to be added, it makes sense to have a standardized payment. The question of what the nonstipend, non-fringe-benefit part of DME payment should be in newly eligible, nonhospital settings is best resolved through a standard payment. Such a standardized payment could be based on the national Medicare share of current national DME cost (or, more simply, on current Medicare national payment) divided by the total number of residents. Because the payment would be really no more than the annual Medicare DME fund divided by the numbers of residents, Congress could decide whether to enrich or reduce this amount by controlling the annual fund amount, depending on the availability of funds and the status of teaching hospitals, but it would be the basis for transition. A general, non-Medicare DME fund (if available) would be distributed in the same way. Averaging is not a new concept. Regulations now provide for payment to new hospitals of the mean per-resident amount in the hospital's PPS wage area (42 C.F.R. 413.86 et seq.).

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund DISSOCIATION OF PAYMENT FROM INDIVIDUAL CASELOAD Currently, a teaching hospital's DME payment fluctuates with the proportion of that hospital's inpatient days that are paid for by Medicare. This introduces a variable that bears little relationship to the cost of training, its effectiveness, or quality. ''[DME] factors . . . [are] really based on objectives quite apart from the care of particular patients in a particular hospital'' (Schweiker, 1982). It also penalizes or rewards teaching hospitals participating in a training system according to local caseload and population characteristics. Further, Medicare admissions to teaching hospitals would not be relevant to the costs and benefits of newly eligible nonhospital sponsoring institutions. Arguments favoring local caseload adjustments include precedent, reluctance to disturb current payment patterns, concerns about any change causing substantial redistribution, the incentive to provide access to services to Medicare patients, and the concept of a direct linkage of Medicare payment to the Medicare beneficiary. A uniform payment is recommended by its effect on sites with low Medicare caseloads that are valuable training settings or important for generalists' experiences, such as children's hospitals (which are losing state and federal GME support as a result of conversion of fee-for-service Medicaid patients to capitated plans). As discussed earlier, if DME and associated teaching hospital activities are more a general, societal benefit, to a degree separate from services to individual Medicare beneficiaries, then Medicare and non-Medicare payers could contribute a proportional share, distributed evenly. This suggests very strongly the inclusion of payments from non-Medicare payers, or from general revenue. The committee thinks a uniform payment and a leveling of DME payment variation is a desirable objective. Reaching this payment is a two-step process; standardizing the per-resident amount and then paying a uniform, Medicare portion of that amount. In practice, of course, with a defined fund the payment is simply the fund divided by resident numbers, as noted. Medicare expenditures are the same whether Medicare (a) pays a percentage of the costs of each resident at an institution based on Medicare's proportion of the institution's total caseload or (b) sums all percentages and pays the national average percent of national teaching hospital caseload. The same would be true for funds applied to non-Medicare caseloads. In the former case, the estimated amount is distributed unevenly. In the latter case, once the rationale of a uniform national percentage and a standardized payment dissociated from the 1984 cost base is accepted, the practical effect is simply a uniform national per-resident payment. Payment would be composed of a uniform Medicare and general non-Medicare component. This payment can be dissociated from cost and considered to be paying for a benefit prospectively. It could be adjusted when appropriate after the first phase of the transition, to accommodate regional differences in training cost, such as costs of inputs, or to encourage programs responding to special needs. Reductions or enrichments can be achieved by reducing or raising the total appropriation to the trust funds, which has the effect of changing the per-resident

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund payment amount. The effect is reminiscent of the change in hospital payment caused by the introduction of the PPS (Fishman, 1993); it could require no midyear adjustment for DME if the DME funds are simply divided up based on a tally of resident numbers. The Secretary of HHS should be given flexibility in timing and precision for determining how residents are counted and payments reconciled so that added per-resident payments during a payment year will not be made after the year's standard per-resident amount is calculated. Residents added at a time appropriate to be counted should generate payments at the institution's current transition amount or the current uniform amount, whichever is less, as described below. The committee's plan moves over time, with the transition discussed, to eliminate the caseload variable (and the 1984 cost base), thus creating a uniform payment. The cost of a resident to an institution does not depend on who the payers are. A TRANSITION Sudden shifts in payment amounts are disruptive. Legitimate differences in the cost of doing business exist for institutions in different regions. However, the committee is aware that adjustments as a result of regional differences, even if applied generously are unlikely to substantially mitigate DME payment redistribution effects (Anderson, 1996; COGME, 1996). A gradual movement from historical payments, which bear some relationship to costs, toward a uniform payment over a 5-year period should provide the best protection against undue disruption of institutions. The proposed method requires first adjusting each institution's current DME per-resident payment proportionally to fit the sum of payments within the determined fund amount for the year in question, and then calculating the uniform per-resident payment amount and moving each institution's per-resident payment up or down toward that amount by 5% each year (or a lesser percent if it reaches the uniform amount). The objective of the transition is to prevent severe losses or windfall gains. This initial transition should continue for 5 years, during which time a DME payment system should be developed for implementation after the transition. This DME payment system should continue the movement to uniformity across institutions and to a more standard national price, but at the same time reflect legitimate differences in costs of training residents in various institutions and parts of the country. Adjustments for these differences were not included in the original transition because movement from historical payments which, in theory at least, reflect total DME costs is slow and costs of inputs explain very little of current variation (Anderson, 1996). In the future, however, fairness requires that the differences be accounted for. In addition, it is desirable and may be possible to design a DME payment system that recognizes locations with special needs. During the transition, new programs, new sponsoring institutions, or residents added to programs at existing sponsoring institutions should be

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund supported at the uniform per-resident payment or, when available, the institution's actual per-resident amount, whichever is less. PAYMENT POLICY AND WORK FORCE COMPOSITION The committee decided not to disturb the current weighting of residents for payment purposes based on duration of residency or specialty type. These differences should be continued because they may encourage teaching hospitals to continue residencies in primary care that are more aligned with current work force policy. They may also discourage allowing trainees to change from short residencies to longer ones. However, recent analyses (Cohen and Todd, 1994; Goodman et al., 1996; Mullan et al., 1994; PPRC, 1997; Seifer et al., 1995) reinforce the committee's uncertainty about future work force requirements and thus about the case for attempting additional work force restructuring through payment changes. Even though there is considerable support for some work force policies regarding primary care and the ratio of generalists to specialists, there are differences of opinion and uncertainties caused by the variability of predictive models (AAMC, 1996; Burg et al., 1994; Cooper, 1994; Meyer et al., 1996; Rivo et al., 1993). There remains, as well, a legitimate question as to whether weighted DME payments will influence applicants as effectively as the market does (PPRC, 1993). Some have also raised objections to a governmental prescription concerning the work force (Dranove and White, 1994). The shape of the U.S. health care delivery system, which has shown the capacity for considerable change recently, is probably not reliably predictable a decade from today when medical students now entering the training pipeline will emerge. Further, in a fixed-cost system, additional payment adjustments to influence work force would have no aggregate cost effect and, unless they were quite severe, probably only modest distribution or behavior effects consistent with past experiences in reaction to payment weights. The committee appreciates the value of primary care providers and of an emphasis on ambulatory and preventive medicine (IOM, 1996b) but prefers to allow change to occur through a system that is relatively permissive, rather than attempting to impose policy and limit local options through payment restrictions. Almost all observers agree that the United States faces an overall surplus of physicians (IOM, 1996a; COGME, 1996), although at the moment there are some conflicting indicators (PPRC, 1997). The committee does not think that this argues against GME and teaching hospital support. It does require that a payment system at least set a total expenditure level and not pay above that for additional residents who will contribute to oversupply. Nonetheless, in some major regions the number of physicians is significantly below the national average (Kindig and Libby, 1994), suggesting that there may occasionally be local reasons for the introduction of new programs or the expansion of old ones. A fixed (and possibly reduced) pool of total GME funds, a lower high-end DME payment, payment for new positions at the lesser of the uniform or actual price,

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund and changes in IME to dissociate resident numbers from payment, will diminish the GME payment stimulus on work force. The committee rejected, therefore, a freeze on total numbers. The committee appreciates that a more forceful approach than eliminating incentives, such as strong incentives (or penalties) to reduce the number of resident positions could be implemented. Given that, with the proposed plan, resident numbers would impose no aggregate financial cost to the National GME Trust Fund, the committee preferred to adopt a less intrusive posture and allow the system time to change in response to current market conditions. This less intrusive posture also resolves the committee's uncertainty as to the merit of creating a separate payment program for other professions, such as podiatry and dentistry, or the fairness of sweeping them into a restrictive mandate primarily in response to a physician surplus. The committee also took no steps to control the number of international medical graduates (IMG). The 46% excess in the number of residents over the number of graduates from U.S. medical schools is due to inclusion of IMGs (Iglehart, 1996; Mullan et al., 1995). About half of these international graduates are legally entitled to remain in the United States (Aronson, 1996; IOM, 1996a). The committee believes that the number of residency positions is leveling off (Jones and Smith, 1997; PPRC 1997) and will shrink, reducing opportunities for IMGs. Positions should remain open to all competitors. Some institutions rely on IMGs to deliver care, often uncompensated care; they should not be disadvantaged. The committee is confident that U.S. medical graduates will continue to match to residency programs much more successfully than IMGs, as they have historically (AAMC website, 19 March 1997), and as the number of positions decreases, that the IMG situation will evolve in a way that minimizes program turmoil. The committee also noted that training IMGs, if they return to their home country, can be an export opportunity as well as a service to the rest of the world. MAINTAINING QUALITY STANDARDS Accrediting bodies, laws, and regulations must continue to strengthen the requirements for resident supervision, the priorities for education over service, and the diversity, relevance, and quality of the training experience. Policymakers should be wary of any changes in policy or hospital behavior that shift the balance unduly toward the provision of low-cost labor or that inappropriately minimize training priorities and expenditures. These changes could reorient the financial incentives and undermine the public purposes supporting the rationale for DME (and IME) funding in general and for any distribution strategy specifically (PPRC, 1994). The committee strongly urges accrediting bodies to encourage reductions in lesser quality training programs by vigilantly enforcing educational standards and by prohibiting undue emphasis on service. Withholding accreditation should be used to enforce quality. Accrediting bodies also will have a major role in ensuring the training diversity allowed by the

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund expanded eligibility for DME funding. Accrediting bodies should develop approaches to further the objective of providing new GME settings that are responsive to the changing health care delivery system. DME SUMMARY The committee proposes the creation of an annually defined Medicare fund distributed in uniform payments per resident (with weighting and transition) to institutions that accept responsibility and meet the standards for a quality GME program. A second, general, non-Medicare fund is desirable and would be distributed in the same way in the form of an additional payment. A reasonable transition period is proposed during which an improved DME funding system should be designed. The committee's proposed payment system would be more neutral in terms of its work force effects. The proposal would also allow new and different institutions to participate in training, and the system could respond to market changes in the organization and delivery of health care. To the extent that numbers of residents increase, the per-resident payment would decrease. To the extent that numbers of residents decrease, the per-resident payment would increase. Institutions with higher DME payments could add residents only at the lower uniform payment level. This system will give teaching institutions stronger incentives to base the size of their resident work forces on market conditions. IME ISSUES IME payments also are made to teaching hospitals and are related to GME. They differ from DME payments in important ways, however. IME payment is part of Medicare's PPS. Because the IME payment depends on the DRGs, it is provided only in connection with inpatient hospital services. As previously noted, no specific costs are identified. Instead, payment is a percentage increase in the DRG payment for individual Medicare beneficiaries; the amount of that increase is dependent on a formula whose variable is the ratio of residents to beds in each teaching hospital—in other words to the density of students and the intensity of teaching in the facility (Fishman, 1992). Such a formula is consistent with the concept of IME as support for the hospital and support for the clinical research and patient services that accompany hospital-based training. The committee's first two principles recognize the distinct value of IME funding to support the teaching environment and subsidize the costs of residents. IME fund distribution also should be guided by the same principles that inform DME fund distribution. An IME system should work in conjunction with DME payment, coordinating the incentives and disincentives to reinforce the influences of the whole GME payment system. The resulting arrangement should

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund Summary Plan: Direct Medical Education (DME) The Health Care Financing Administration (HCFA) counts the number of full-time-equivalent (FTE) residents. HCFA divides the Medicare and (if available) general, non-Medicare DME funds by the number of FTE residents to get a standard per-resident payment amount. HCFA adjusts current DME payments to fit within the DME fund limits. HCFA pays the uniform Medicare percentage of each sponsoring institution's historical, weighted, Medicare per-resident payment, reduced or increased toward the standard by 5% each year and adds (if available) a uniform general, non-Medicare payment. New sponsoring institutions are paid the uniform Medicare percentage of the weighted, standard Medicare amount and (if available) the uniform general, non-Medicare amount for each resident. The addition of residents at existing sponsoring institutions generates the standard or the current actual Medicare per-resident payment, whichever is less, and (if available) a uniform general, non-Medicare payment. HCFA designates new non-teaching-hospital sponsoring institutions based on accrediting body approval and formal status as a health care service or educational institution and pays the uniform Medicare and (if available) general, non-Medicare amounts. be consistent with national policy goals and improve GME and the health care system. Unfortunately, as noted earlier, some current incentives, such as tying the payment to the number of residents, are associated with an oversupply of residencies and ultimately of physicians (Dunn and Miller, 1996). The IME payment system also discourages training residents in community-based ambulatory settings (Eisenberg, 1990). Service in such settings does not affect throughput of DRG patients in the hospital, nor does nonhospital working time count toward the resident-to-bed ratio. On the plus side, funding that flows with the DRG rate encourages Medicare admissions by enhancing the attractiveness of Medicare patients relative to other payers and improving Medicare beneficiaries' access to high-quality teaching facilities. MODIFICATION OF MEDICARE IME FUNDING To modify these incentives, the committee recommends dividing the IME fund into halves. One half of IME payment would be based on historical IME payments, trended forward by some inflation or medical market basket factors, or as Congress decides. Whatever the amount in this half of the IME trust fund, these payments would depend on the share a hospital's IME payment was of total national IME payment for a base period. The hospital would receive that same share of the current half of the IME fund. The other half would be paid

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund according to the present formula. (The committee does not define the base period for the historical IME share. Clearly, the implication of the choice of base period is very different if the period chosen is [a] 1988, before the most recent surge in resident numbers; or [b] the most recent year before plan implementation, which would reward more highly staffed hospitals; or [c] an average of several years, which could, depending on the years chosen, reflect either or none of the above implications, but in any case, would smooth out annual differences in Medicare caseload or a year that was unusual for some other reason.) Assuming that the total amount available for IME funding approximates an actuarial estimate of what would have been paid through the current formula, payment for the formula half would require dividing each addition to a hospital's DRG payment in half—in other words, using a coefficient of 0.5 in the formula. If the actuarial estimate indicated a less sufficient IME trust fund, say 90%, then a coefficient of 0.45 would be required, and so on. Distributing one defined, predictable fund in discrete payments and the other fund through an unpredictable number of rate adjustments while keeping the funds equal to each other and within the annual defined amount would require some additional adjustments. This could be accomplished through a reconciliation with Medicare cost reports, by adjusting the 0.5 coefficient, or the like. This difficult task might also justify a reserve fund or a less precise target such as accepting a final expenditure of a few percent plus or minus to the IME total fund. Dissociating half of IME payment from the resident-to-bed ratio allows hospitals to reduce resident numbers (and cost) at half the current penalty. Added to the limits in DME payment, this is a step toward payment neutrality in the resident market and should reduce the stimulus of payment policy on resident and physician supply. Continuing the other half of IME funding using rate adjustments preserves the added margin that each Medicare patient now brings to teaching hospitals, counteracting the potential loss of access that might ensue if GME payment were totally dissociated from each hospital's current Medicare caseload. When a formula is used to distribute a predetermined fund, the variables in the formula cannot have aggregate financial impacts on the fund. The variables become less important because they are significant only with regard to distribution effects. For example, the fact that the formula produced a 7.7% increase in rates for each 10% increase in the resident-to-bed ratio, which has always been known to be substantial overpayment, would no longer be as meaningful. Changing resident-to-bed ratios could not increase total IME payment to the teaching hospital sector at all if the fund amount did not increase. If any variable were to be adjusted to conform payment to fund amounts, it should be the 0.5 that was added to accomplish the division in halves. The rest of the formula should remain in its familiar form to have its familiar distribution effects. The inclusion of capitated patients in IME payment could not generate added cost to the National GME Trust Fund. The half of the payment based on

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund historical estimates is dissociated from rate adjustment and current and future caseload. This half of the payment is based on whatever the Medicare caseload was during the base period and is distributed accordingly. Recognizing capitated patients in the half based on the formula through some coefficient, as described below, could affect only distribution, not the total amount distributed. Inclusion of these patients is desirable as an incentive to teaching hospitals to compete for and serve them. In a capped system, the distribution effects would be magnified; an increased share for one institution means a decreased share for others. To accommodate capitated Medicare caseload, the committee considered adjusting the IME formula percentage added to each DRG payment by a factor based on the hospital's prepaid Medicare caseload as a percentage of total Medicare caseload. Each IME formula adjustment to DRG payment to a hospital would be multiplied by a coefficient that was based on the proportion of the total Medicare caseload at that hospital that was capitated. Case mix would have to be ignored in calculating this coefficient unless the capitated patients were assigned DRGs and the percentage was then calculated using imputed DRG billings. Using capitated Medicare inpatient days divided by total Medicare inpatient days should be a reasonable approximation. Another alternative considered was a "shadow" system in which the capitated patients were assigned DRGs. The payments would then be calculated based on the formula, and the adjusted amount would be billed separately. Adjustment of the DRG rate by an additional coefficient might be easier, but, in any case, the value of including capitated caseload in the IME payment distribution system justifies the administrative effort needed to design and implement an adjustment. IME GENERAL, NON-MEDICARE PAYMENTS The general, non-Medicare portion of the IME trust fund program, if enacted, would be divided in halves like the Medicare fund. The historical fund would be distributed based on the historical non-Medicare share. That is, for the base period, the Medicare IME adjustment would be calculated using the current formula, then applied to the hospital's revenues from the non-Medicare caseload for that period. (Consideration might be given to allowing an adjustment to these revenues in hospitals, such as public hospitals, with very high levels of uncompensated care.) Half this amount, the non-Medicare share of the historical portion, would then be paid and trended forward in the same way as the Medicare portion. The same process would be used for the non-Medicare caseload in the formula-driven portion of the fund. These amounts would also be paid separately, because they could not be added to rates. If the Medicare caseload was a small percentage of total caseload, the non-Medicare caseload would inevitably be larger, and vice versa. This would have the effect of leveling support and attenuating the distribution effect of the Medicare caseload, depending to some extent on the adequacy of the general, non-Medicare IME fund. Although this might weaken the attractiveness of Medicare patients, it

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund would have the effect of supporting teaching hospitals with low Medicare caseloads and would be consistent with the principle of having payers contribute their share. In the case of some low Medicare caseload hospitals like independent children's hospitals, Medicare IME payments could not be made as these facilities are not part of the PPS. Any hospital that qualifies as a GME-sponsoring institution should be eligible for general, non-Medicare IME payments (if available). Their payments should be based on application of the current formula to revenues from the non-Medicare caseload. To keep within the appropriation, IME payments will need to be adjusted. AMBULATORY TRAINING Training in independent ambulatory settings currently counts toward resident time for DME payment provided resident stipends are covered by the sponsoring teaching hospital. Formula IME payment is not made for time spent by residents outside the hospital and its outpatient department. Reordering incentives, such as encouraging ambulatory training by expanding DME-eligible institutions, which might include ambulatory sites or consortia, and basing half of IME payment on historical factors, which dissociates it from resident numbers or assignments to out-of-hospital settings, would have no aggregate payment significance. Therefore, the committee recommends that resident training time in ambulatory sites should count toward IME payment in the same way it does for DME payment. This would permit expanded opportunities and enhancement of training responsiveness to current conditions. Furthermore, sponsoring institutions should be encouraged to ally themselves with additional ambulatory settings by sharing DME support in addition to paying resident stipends. The committee thinks that the GME system is still sufficiently weighted with inpatient incentives that a new balance between ambulatory and inpatient care, as described, will continue to give residents needed exposure to seriously ill inpatients as well as to changing practice patterns. Exploration of the desirability and feasibility of IME support of nonhospital sponsoring institutions was suggested earlier. The committee notes that new DME-sponsoring institutions that are not hospitals cannot qualify for IME payments. In such cases, IME payments should be made to the teaching hospital that takes ACGME/AOA-approved responsibility for each resident's hospital-based teaching and for only that time spent in hospital-controlled training areas. In such cases, the committee thinks that receipt of total DME payments is adequate reimbursement presently for nonhospital sponsoring institutions, but as noted, at some point, this policy may need to be reevaluated. The hospital should not receive IME payment for time during which the resident is the responsibility of another, separate sponsoring institution in another setting.

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund IME SUMMARY AND GME CONCLUSIONS IME fund distribution is simple conceptually, though challenging administratively. The fund is divided—one half is distributed to teaching hospitals on a historical basis, the other as rate adjustments using the current formula. The payment becomes prospective, and capitated patients are added to the formula part. General, non-Medicare funds, if included, pay their share. As a whole, these changes should improve the prospective nature and predictability of the GME payment system. The GME trust fund program allows issues concerning teaching hospitals and their financial health to be considered separately from those of hospitals in general. The fund's adequacy can be monitored and controlled by using data from teaching hospitals in ways similar to those used for the entire hospital system. This approach reduces the incentive for having an oversupply of physicians, is aligned with current views on work force composition, and generally is less restrictive and leaves work force decisions more to the market. IMGs compete with graduates of U.S. medical schools on an even playing field. In addition, GME funds in the AAPCC would no longer be an issue, since a defined annual GME fund assures GME payment cannot be lost from the Trust Fund to the AAPCC. Assuming a general, non-Medicare fund, the distribution would occur from four component funds limited by the total National GME Trust Fund amount, two for DME distributed in a uniform way, and two for IME distributed proportionally to Medicare and non-Medicare caseload. Support to institutions with low Medicare caseload would improve. Approximately the same amount would be paid for the same result. There would be some redistribution, as inevitably happens in any reorganization. The committee tried to ensure that the neutrality and fairness of the system would provide some comfort to all the participants, and that there was a give-and-take that balanced disadvantage with advantage as much as possible. DME FOR NURSING The committee recommends that nursing DME payment be for accredited graduate and postgraduate clinical nursing programs. GME funds for nursing would, as has been recommended for physicians, be paid to the entities that can qualify as sponsoring institutions and will be responsible for the graduate clinical education and incur the costs of that education. Funding targeted at the graduate level supports trainees who have completed the basic requirements for licensure. These nurses provide valuable professional services that enhance the overall quality of patient care while they pursue their advanced training (Malone, 1997). About 70% of GME payments for nursing support undergraduate, preprofessional nursing in diploma (i.e., "provider-operated") programs. These programs are vanishing; during the period Medicare has been supporting them, their numbers have decreased by about 85% (Aiken et al., 1994). Provider-

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund operated programs currently train about 5% of nurses, and are expected to train even fewer nurses in the future (HRSA, 1992). Hospitals with undergraduate nursing programs are often not teaching hospitals or sponsoring institutions as defined by the committee, and they, understandably, tend to provide training with a hospital focus. Medicare pays a portion of nursing training costs that totals approximately $220 million yearly (B. Wynn, personal communication, 21 February 1997). Summary Plan: Indirect Medical Education (IME) The Health Care Financing Administration (HCFA) sets the historical half of Medicare IME payments based on the same proportion of one half of the Medicare IME fund that each institution received of total Medicare IME payment in the base period and trends this forward. HCFA calculates the IME Medicare adjustment to the diagnosis-related group (DRG) rate by the current formula and makes an initial adjustment to conform the actuarially estimated annual payout to the actual amount in the formula half of the Medicare IME fund, including funds for the capitated patients' coefficient distribution. HCFA makes periodic historical Medicare IME payments, monitors IME-formula-adjusted DRG payment, and adjusts payments to conform with the target fund amount. HCFA applies a capitated patient adjustment at the time it considers to be most efficient. HCFA sets the historical half of general, non-Medicare IME payment (if available) as was done for Medicare using imputed non-Medicare calculations instead of Medicare data. HCFA calculates the formula half of the general, non-Medicare IME payment (if available) using the current formula and based on the formula's application to hospital reports of non-Medicare revenue. HCFA makes periodic historical IME general, non-Medicare payments and formula general, non-Medicare IME payments adjusted to keep within the target fund amount. Time spent by residents in ambulatory, non-hospital settings is included in the full-time-equivalent resident count for IME payment purposes. Support for stipends and for supervision of advanced clinical trainees, similar to that provided for physician DME, is lacking. The committee thinks DME funds should support advanced trainees. The establishment of a trust fund would provide a vehicle for retargeting in a budget-neutral manner funds presently going to nursing education to yield better access and quality of care than present policies. Funding could be set at the same percentage of total DME as is currently devoted to nursing. A uniform per-trainee payment would be paid to sponsoring institutions. The total payments would be kept within the

OCR for page 21
On Implementing a National Graduate Medical Education Trust Fund appropriated fund amount. To avoid adversely effecting any student currently in training current funding of hospitals with undergraduate, provider-operated nursing programs should be phased out over not more than 4 years, so that students now in these training programs have graduated. Summary Plan: Nursing The Health Care Financing Administration (HCFA) counts the number of full-time-equivalent (FTE) graduate nurses undergoing clinical training as the responsibility of the sponsoring institutions. HCFA divides the amount of direct medical education funds set aside for graduate nursing by the number of FTE graduate nurses in clinical training to get a standard per graduate nurse payment amount. All sponsoring institutions are paid the standard per graduate nurse payment for their graduate nurses in clinical training. ALLIED HEALTH The committee only briefly considered allied health programs. They constitute only 1%–1.5% of GME expenditure (B. Wynn, personal communication February 21, 1997). Allied health training is presently supported in the same way as nursing training, through partial reimbursement to hospitals for the submitted costs of accredited programs. The committee recommends that consideration be given to paying the costs of allied health programs in ways that are proportional to the limits proposed for nursing and other parts of GME payment. If GME support is distributed from an annually determined fund, the committee does not want support of allied health programs to grow at the expense of other professionals. A number of allied health professions, not currently eligible for Medicare GME support, proposed that they should become eligible. The committee cannot speak to the merits of all these applicants but notes that some allied health professionals can substitute for residents and physicians and also pursue advanced clinical training—such as physician assistants, and in some cases, perhaps clinical psychologists, for example.