The Case for Teaching Hospitals

In a recent article published in Academic Medicine, Ralph W. Muller, immediate past chair of the AAMC, gave the AAMC’s rationale for public support of teaching hospitals and medical schools:

• About half, overall, of teaching hospitals’ patients have public insurance. In 1998, approximately 30 percent of patients treated by teaching hospitals were Medicare beneficiaries and 20 percent were Medicaid recipients.
• Teaching hospitals “provide services that are not generally available in all hospitals.” They are the “specialty centers” to which people go for advanced specialized care. For example, in contrast to 6 percent of all hospitals, 66 percent of teaching hospitals performed organ transplants in 1998. Compared with 24 percent of all hospitals, 73 percent of teaching hospitals gave trauma care. As opposed to 12 percent of all hospitals, 65 percent had neonatal units.
• According to “mounting evidence,” teaching hospitals “do better in providing higher quality care. An analysis of care in ‘high-volume hospitals’ found that patients referred have significantly better mortality results than do those in low-volume hospitals for several major procedures and diagnoses.”
• Teaching hospitals “not only provide specialty care but also serve their communities’ need for primary care.” The settings in which the care is provided “are diverse, from hospital-based clinics to federally qualified health centers (FQHCs) located in neighborhoods that are underserved.”
• Teaching hospitals provide “complex care.” Patients tend to “be seen by several doctors, who receive consults from others, and attract the considerable interest of a team of medical students, residents, and fellows.” Medical centers are

  organic entities that have grown over decades to handle complicated tasks and complex missions…. The people who are skilled in handling the challenges of [medical] centers are not optimized as economically efficient actors but instead are organized—institutionally, culturally, and even politically—to meet complicated situations within a team of highly talented individuals.

• Teaching hospitals are social institutions. They may be the largest employers in their communities, have emergency rooms open 24 hours a day, and operate social-service units that link patients to other institutions in their communities.⁷

Challenges from Others

Others think public support of GME can be useful in achieving certain health professions education and health service goals. These include organizations (such as the National Association of Community Health Centers) that want to direct GME funds from teaching hospitals to ambulatory facilities as well as organizations (such as the American Academy of Family Physicians) that view GME funds as crucial in shaping the health workforce.

Among their goals are governing the rate of growth in the supply of physicians, providing incentives for a higher proportion of generalists to specialists, increasing the numbers of ethnic and racial minorities in the physician workforce, and assuring safety-net providers. The Council on Graduate Medical Education (COGME) included these goals in a document—the “Fifteenth Report”—that it issued on GME financing at the end of 2000.⁸

Changes in DGME over Time

For the first 20 years of the Medicare program, Medicare picked up “allowable” teaching hospital costs. However, a series of omnibus reconciliation measures limited reimbursement of the costs:

• The Consolidated Omnibus Budget Reconciliation Act of 1985 (COBRA, actually enacted in 1986) limited the costs to a capitated per-resident amount (figured on a hospital-by-hospital basis) and to a defined number of residency years for full Medicare reimbursement.
• The Balanced Budget Act of 1997 (BBA) capped the number of residents who could be counted. The BBA also added a provision permitting the Medicare program to pay certain ambulatory sites for DGME costs. These sites include Medicare-participating FQHCs, rural health clinics, Medicare+Choice organizations, and other entities as designated by the Medicare program. The conference report that accompanied the measure expressed concern over Prospective Payment Assessment Commission and Physician Payment Review Commission studies¹⁰ showing wide variation in per-resident amounts from hospital to hospital.
• The Balanced Budget Refinement Act of 1999 (BBRA) created both a minimum payment level and a maximum payment level for hospital per-resident amounts.
• The Benefits Improvement and Protection Act of 2000 (BIPA) raised the minimum level but did not change the maximum level.

Variations in GME payments in selected states are featured in a paper—commissioned by the California HealthCare Foundation—by Thomas R. Oliver, Ph.D., and Atul Grover, M.D., of Johns Hopkins University, as well as Philip R. Lee, M.D., of the University of California at San Francisco.¹¹ The paper examines variations in Medicare GME payments that occur in states that receive the highest amount of DGME and IME support. It reports the variations, tracks their growth since the implementation of PPS, and seeks to give reasons for them. For example, the paper includes a table that presents Medicare GME payments for the top five teaching hospitals in six states (see Table 1).

Table 1

Medicare GME Payments per Resident and Discharge for Top Five Recipients in Selected States, FY 1996.

Because both the BBRA and BIPA contained provisions—by establishing floors and ceilings—to narrow the variation in DGME payments, the profiles of hospitals and states may look somewhat different in the future, according to Michael Hash of Health Policy Alternatives.¹² It is difficult to estimate exactly how the distribution of payments will change. Although DGME is smaller than IME in terms of GME payments, it arguably reflects differences in payment that are the result of organizational decisions and circumstances that determine how to finance the costs of residency programs. Since residency stipends do not vary widely across the country, faculty costs and overhead expenses have generally been found to account for most of the variation. Moreover, the significance of faculty practice plans, volunteer faculty, fully salaried faculty, and specific overhead costs needs to be determined.

Because Medicare PPS was designed to be a competitive “winners and losers” program, examining state variations is bound to be controversial. The Oliver, Grover, and Lee paper raises significant questions about what is driving Medicare GME costs. It does so even though aggregating GME expenditures within states may obscure some of the characteristics or practices of individual hospitals that result in very different GME payment amounts. For example, for DGME, the hospital’s number of residents, Medicare inpatient days, and accounting practices at the start of PPS (when the level of cost per resident was established as a basis for DGME costs) are the primary determinants of payment. For IME, a specific hospital’s number of residents, staffed beds, area wage index, urban or rural location, and Medicare inpatient utilization rate are the major determinants.

Declines in the IME Factor

When Congress passed the Social Security Amendments of 1983 mandating a DRG-based PPS for Medicare inpatient services, it included report language calling the IME “a proxy to account for a number of factors that may legitimately increase costs in teaching hospitals.” (For instance, an often-cited factor is a teaching hospital’s having a more severely ill patient caseload than is accounted for in its case-mix index.) Because the IME is not an implicit educational payment, some have advocated giving it a strict definition. For example, COGME, in its Fifteenth Report, proposes that “IME payments should be set at no more than the analytically justified level for teaching activities.”¹³

Beginning with hospitals’ 1984 fiscal year, the first year of PPS, IME was an 11.59 percent add-on to each DRG rate (for a Medicare patient) for every 10 percent increase in a teaching hospital’s intern/resident-to-bed (IRB) ratio. Subsequent omnibus reconciliation measures amended the IME factor provision as follows:

• COBRA reduced the IME factor to 8.1 percent for fiscal years (FYs) 1986 and 1987 and to 8.7 percent in future years.
• The Omnibus Budget Reconciliation Act of 1987 (OBRA) reduced the IME factor to 7.7 percent, beginning in FY 1989.
• The BBA reduced the IME factor to 6.5 percent for FY 1999. It also provided for the factor to decline to 6.0 percent in FY 2000 and to 5.5 percent in FY 2001, provisions amended by the BBRA (see next bullet).
• The BBRA set the IME factor at 6.25 percent for every 10 percent increase above 0.10 in the IRB ratio and delayed for one year decreasing the factor to 5.5 percent—as provided in the BBA.
• BIPA further delayed the BBA reductions in the IME reduction, thereby restoring the IME factor to 6.5 percent until FY 2003, when it is slated to decline to 5.5 percent.

Table 2, prepared by the Division of Health Care Affairs of the Association of American Medical Colleges (AAMC), charts the IME percentage add-on to the DRG payment for illustrative intern-and-resident-tobed ratios incorporating changes mandated by the BBA, BBRA, and BIPA.

Table 2

IME Percentage Add-ons to the DRG Payments by Intern-and-Resident-to-Bed Ratios As a Result of the BBA, BBRA, and BIPA.

A Profile of the States

Tim Henderson, program manager and director of the Center for Primary Care and Workforce Analysis, National Conference of State Legislatures (NCSL), has done many of the analyses of Medicaid GME financing that are available to policymakers. According to a study that Henderson conducted for NCSL in 1998 and 1999 and reported in Health Affairs,

• A total of 45 states and the District of Columbia paid for GME at some level.
• Of the 45 states and the District of Columbia, 43 paid for GME in their fee-for-service (FFS) programs. Of these, 24 made both IME and DGME payments and 11 did not distinguish between the two. Thirty-five of the states that paid for GME under FFS did so through hospital per-case or per-diem rates.
• Of the 42 states and the District of Columbia that reported capitated Medicaid arrangements, 16 states and the District of Columbia made explicit Medicaid GME payments to teaching hospitals or teaching programs. Seventeen other states included the payments in managed care organizations’ capitated rates. Teaching hospitals were the recipients of most states’ GME payments, although, in Oklahoma and Tennessee, medical schools were the only training programs to receive them directly under managed care.
• While medical residents were predominantly eligible for Medicaid GME payments, nurses and other health professions students were eligible under managed care (or there was no distinction among the health professions) in eight states and the District of Columbia.
• Of the five states (and Puerto Rico) that did not provide Medicaid GME support, only Illinois and Puerto Rico had significant residency programs. Alaska, Idaho, and Montana, which belonged to the Washington-Wyoming-Alaska-Montana-Idaho undergraduate and graduate medical education program centered at the University of Washington, did not have academic health centers. South Dakota had one but did not report any GME payments to its teaching hospital or medical school.¹⁴

Examples of Innovation

Some states have experimented with GME funds in order to achieve certain goals. At an April 2001 state health officials forum held by NCSL under the sponsorship of HRSA,¹⁵ states that have taken the initiative in Medicaid GME reported on their progress:

• Since 1997, Michigan has had three pools: one that pays teaching hospitals at 1995 levels, one that rewards them for their proportions of primary care residents and indigent patients, and one for “innovations” that gives grants to GME consortia made up of a university, a hospital, and a health maintenance organization (HMO). On July 1, 2001, the state plans to implement a new formula for both the historical (1995 base) and primary care pools. The new formula will pay hospitals based on the number of full-time employees in training. It will be weighted both for Medicaid utilization and for performance factors (that physicians participate in Michigan’s Medicaid program after completing their residency and that they receive board certification).¹⁶
• Also since 1997, Minnesota has operated a Medical Education and Research Cost Fund—called “MERC”—for clinical training costs. It is currently funded by general revenues; the state’s tobacco settlement; and federal Medicaid matching funds, which are clearly earmarked for medical education. It supports the unrecovered costs of clinical training for medical students and residents, dental students and residents, pharmacists, chiropractors, advanced practice nurses, and physician assistants. It targets its funds to cover a specific percentage of clinical training costs at sites that provide care to the Medicaid population. As part of MERC, the state instituted a managed Medicaid carveout pool that became effective in 2000 but has yet to be extended to the entire state.¹⁷
• New York has a GME Reform Incentive Pool that has the following goals: to reduce the number of physician trainees, to increase the number of primary care physicians, and to promote residency training in ambulatory sites. Distribution of the funds, which may be to individual hospitals or a consortium of hospitals, is based on performance in meeting the goals (for example, number of residents, primary-care training, and number of ambulatory-care sites).¹⁸
• Tennessee eliminated Medicaid GME funding in January 1995, when it adopted TennCare, its Medicaid managed care program. After a study group had evaluated TennCare’s role in funding GME and developed a methodology, the state restored its GME funding in 1996, retroactive to July 1995. Under the new model (subject to a five-year phase-in), the goals are to increase the training of Tenn-Care physicians, retain them in underserved areas, and direct GME money to nonhospital settings. Payments go to medical schools, of which there are four in the state, rather than to teaching hospitals. Each medical school has its own primary care target and is subject to a cutback in funding if it fails to meet it. There is a set-aside for stipends to physicians who stay in the state.¹⁹
• Texas has had discretionary programs to encourage practice in rural areas and primary care training since 1979. In addition to its graduate medical residency program, it has family practice and primary care residency programs, plus a resident physician compensation program, which supports residency training in teaching hospitals. Its GME program centers on eight medical schools and 11 independent primary care residencies. In terms of Medicaid, the state opted in 1997 for a Medicaid carveout of GME funds from HMO capitation rates and directed the funds to teaching hospitals for primary care residency training.²⁰
• Based on a health professions education cost study completed in 1996, Utah created the Medical Education Council in 1997 to address the stabilization of health professions education funding and to correlate the financing with workforce goals. The council’s criteria include geographic balance, training setting, encouragement of multidisciplinary education, and stable funding for accredited programs. The council would like to see both Medicaid and Medicare funds paid to it. While the council has a Medicare waiver application in its final stages with the Health Care Financing Administration (HCFA), it has made no progress on amending the state’s Medicaid plan. Prior to last year, when the Utah legislature approved $500,000 to stabilize a family practice training program, the state had not provided any funds for GME training.²¹

Adverse Effects of Managed Care and the BBA

The growth of managed care in state Medicaid programs is assumed to have a negative effect on Medicaid GME payments. From 1996 to 2000, the Medicaid managed care population increased from 13.3 million to 18.8 million, or from 40.1 percent to 55.8 percent of the total Medicaid population.²² Even if states pay managed care organizations for GME, the organizations “are not bound to distribute these dollars to hospitals with GME programs or to provide GME themselves.”²³

Moreover, the BBA may have had a chilling effect on Medicaid GME because of a relationship between Medicaid GME and Medicaid disproportionate-share hospital (DSH) payments. The BBA reduced Medicaid DSH payments by $10.4 billion over five years, a reduction that was partly restored by the BBRA and BIPA. Hospitals may have looked at Medicaid GME as a trade-off for reduced Medicaid DSH payments, much the same way in which they see a trade-off between the Medicare DSH and IME adjustments.

Minor Changes—Tinkering around the Edges

Various changes have been made in the GME program since it was initiated. According to Richard Knapp, Ph.D., AAMC vice president,

six policy issues have dominated the Medicare DGME debate since the enactment of COBRA: payments to nonhospital entities, the duration of training for which payment is made, the limitation of payments to certain types or a defined number of residents, differential payments by specialty, payments directly to programs, and limits on the geographic variation in per-resident payments.

A Major Change—Folding DGME into IME

For the last two years, the Medicare Payment Advisory Commission (MedPAC) has advocated replacing the existing GME system with a new approach. In 1999, the commission proposed that DGME and IME be combined into a single adjustment to DRG rates, rather than having DGME be a pass-through (albeit subject to restrictions) and having IME be an adjustment to the DRG. It referred to the adjustment as an “enhanced patient care” payment, a term that was later dropped. It also advised revising certain aspects of Medicare case-pricing, in order to achieve better correlation between expected inpatient costs and actual inpatient payments per case across various types of hospitals.²⁵

In its June 2000 Report to the Congress, the commission reiterated its positions on what it now called “a teaching adjustment” and on case-pricing. It provided an analysis of its 1999 recommendations and concluded:

The Congress should fold inpatient DGME costs into PPS payment rates through a revised teaching hospital adjustment. The new adjustment should be set such that the subsidy provided to teaching hospitals continues as under current long-run policy. This recommendation also should be implemented with a reasonable transition to limit the impact on hospitals of substantial changes in Medicare payments and to ensure that beneficiaries have continued access to the services that teaching hospitals provide.²⁶

Joseph P. Newhouse and Gail R. Wilensky, two of the MedPAC principals involved in the conception of the commission’s proposal, justify merging DGME and IME because they see “no economic reason for Medicare to distinguish” between them.

Both types of cost represent the additional cost of the patient care provided at teaching hospitals. Thus, the direct payments can be treated in the same fashion as indirect costs. Doing so would resolve one anomaly of the current payment method: direct GME costs are reimbursed at the 1984 level of costs per resident, trended forward by the change in the Consumer Price Index. Reimbursement has been held at 1984 costs plus inflation to preclude hospitals from simply moving other costs to those that are now passed through. As measured by Medicare, however, hospitals had very different levels of costs per resident in 1984, in part because some hospitals had more volunteer faculty (and thus lower costs) and in part because of differences in hospitals’ accounting practices in 1984.²⁷

Entitlement to Authorization—Taking the CHGME Model

Following the CHGME example of subjecting GME costs to the authorization and appropriation process would, of course, mean separating them from the Medicare entitlement program. Not only would going the authorization-appropriation route mean assigning the significant GME dollars to different congressional committees but it would also mean assigning the program to a different administrative unit at the Department of Health and Human Services (DHHS). The House Ways and Means Committee and the Senate Finance Committee now have Medicare GME in their jurisdictions, as does HCFA at DHHS. CHGME falls under the House Energy and Commerce Committee and Appropriations Committee, as well as under the Senate Health, Education, Labor, and Pension Committee and Appropriations Committee. While DHHS’ HCFA has Medicare GME, HRSA is in charge of CHGME.

While the National Bipartisan Commission on the Future of Medicare did not come to a final vote on its recommendations, one of its pending proposals was to convert DGME to a program in which its funds would be appropriated, while retaining IME as an HI Trust Fund expenditure. Objections were raised that doing so would make funding of DGME unpredictable and capricious, forcing it to compete with other discretionary health programs.

A Revenue Initiative—Creating an All-Payer Fund

While the idea of an all-payer fund goes back to a provision in former President Clinton’s Health Security Act proposal, Sen. Daniel Patrick Moynihan (D-N.Y.), who retired at the end of the 106th Congress, carried it forward. Along with Rep. Nita M. Lowey (R-N.Y.), he cosponsored the Medical Education Trust Fund Act of 1999, to establish separate Medicare and non-Medicare trust funds for teaching hospitals and to create a medical school account. If enacted, the Moynihan-Lowey approach would have levied a 1.5 percent tax on health insurance premiums and administrative services as well as drawn on 5 percent of federal Medicaid spending for inpatient services to establish the non-Medicare account. Sen. Jack Reed (D-R.I.) has introduced the legislation in the 107th Congress, joined by the two Democratic senators of New York, Sens. Hillary Rodham Clinton and Charles Schumer, as co-sponsors.

The rationale for an all-payer fund is that “medical education is a public good and that, because everyone benefits from a well-trained workforce, everyone should pay,” as Newhouse and Wilensky describe it. They continue:

This notion is very different from economists’ view of a public good. Loosely speaking, economists view a good as being public if the benefits are equally available to everyone—that is, if consumers cannot be excluded from consumption (for example, the light from a lighthouse).²⁸

According to COGME’s Fifteenth Report, the Pew Health Professions Commission, Commonwealth Fund, and associations representing major GME stakeholders that signed a 1997 consensus agreement²⁹ back some sort of all-payer fund. However, given the lack of Republican support to date for a new tax to mandate that private insurers and health plans contribute to GME, the outlook of such an initiative looks dim in the 107th Congress.