Health Affairs, 22, no. 6 (2003): 112-122 doi: 10.1377/hlthaff.22.6.112 © 2003 by Project HOPE	New Online   VP Candidates On Health Reform   Cutler On McCain's Plan   Scoring Obama's Plan   Roundtable: Politics, Reform   Number Of Uninsured   Delivery System Overhaul

This Article Abstract Figures Only Reprint (PDF) Submit a response to this article Alert me when this article is cited Alert me when eLetters are posted Alert me if a correction is posted Services E-mail this article to a friend Similar articles in this journal Similar articles in ISI Web of Science Similar articles in PubMed Alert me to new issues of the journal Add to My Personal Archive Download to Citation Manager Reprints & Permissions Citing Articles Citing Articles via HighWire Citing Articles via ISI Web of Science (6) Citing Articles via Google Scholar Google Scholar Articles by Koenig, L. Articles by Weissman, J. S. Search for Related Content PubMed PubMed Citation Articles by Koenig, L. Articles by Weissman, J. S. Related Collections Hospitals Research And Technology Safety-Net Systems Health Spending

Teaching Hospitals

Estimating The Mission-Related Costs Of Teaching Hospitals

Abstract

 
Academic health centers and other teaching hospitals face higher patient care costs than nonteaching community hospitals face, because of their missions of graduate medical education (GME), biomedical research, and the maintenance of standby capacity for medically complex patients. We estimate that total mission-related costs were $27 billion in 2002 for all teaching hospitals, with GME (including indirect and direct GME) and standby capacity accounting for roughly 60 and 35 percent of these costs, respectively. To assure their continued ability to perform important social missions in a competitive environment, it may be necessary to reassess the way in which these activities are financed.

Since its inception in 1983, the Medicare inpatient hospital prospective payment system (PPS) has made separate payments to teaching hospitals to cover the "direct" (for example, resident and faculty salaries) and "indirect" costs associated with GME. Medicare policy establishing payments for indirect GME costs was made in recognition of the relatively higher patient care cost at teaching hospitals. One study found that after case-mix differences were adjusted for, average Medicare costs per case at teaching hospitals were 27–39 percent higher than for nonteaching hospitals in 1999, depending on the number of medical students per bed.³

Unlike direct medical education costs, indirect medical education (IME) costs cannot be obtained directly from financial or accounting records but must be estimated using statistical techniques. To estimate these indirect costs, researchers have historically used teaching intensity as measured by a hospital’s ratio of interns and residents to beds (IRB).⁴ This measure is assumed to be correlated with academic activities, such as increased use of diagnostic or ancillary services and the availability of "state-of-the-art" treatment technologies.⁵

Initial work conducted in preparation for the Medicare PPS showed that Medicare inpatient costs per case would rise by approximately 5.7 percent for every 0.1 increase in the IRB ratio, after controlling for case-mix, area wages, bed size, and the size of the geographic area in which the hospital was located.⁶ More recent estimates of IME costs reported by the Medicare Payment Advisory Commission (MedPAC) suggest that the marginal contribution of teaching to Medicare costs per case has fallen to approximately 2.7 percent.⁷

In its January 2003 meeting, MedPAC took up the issue of whether the IME adjustment should be reduced to its empirical level of 2.7 percent. For fiscal year 2003 the IME adjustment was set at 5.5 percent, twice the empirically estimated level of GME’s contribution to costs. Despite this large difference, the commissioners narrowly voted against recommending that Congress reduce the IME adjustment. This decision was based, at least in part, on concerns about the impact a reduction in the IME adjustment would have on teaching hospitals’ financial status and ability to pursue their missions.

Because the ratio of interns and residents to beds is correlated with other teaching hospital missions, estimates of the relationship between teaching intensity and patient care costs by MedPAC and others likely overstate the indirect costs associated with the pure teaching mission and understate the full cost of all the missions pursued by teaching hospitals. Understanding how each of the missions contributes to hospitals’ costs is an important aspect missing from the discussion of funding teaching hospitals. Without a better understanding of the full range of teaching hospital missions and their associated costs, policymakers are likely to underestimate the benefits from public expenditures to teaching hospitals.

This study extends previous work on IME by separately estimating the costs of teaching hospitals’ primary missions. In so doing, we contribute to the literature in three important ways. First, we use "all-payer" hospital costs per case to estimate the impact of teaching on the average cost of all patients. Second, we include measures of each of the three primary missions pursued by teaching hospitals in the analysis: GME, biomedical research, and the maintenance of standby capacity needed to provide highly specialized care to medically complex patients. By controlling for these other missions, we produce a more accurate estimate of the costs associated with teaching only (that is, IME). Third, we produce separate national estimates of the costs associated with the primary missions of major teaching hospitals, following the work of Rob Mechanic and colleagues.⁸

Data And Methods

Top Data And Methods Results Discussion Editor's Notes NOTES

 
Measuring hospital inpatient costs. We used data from the Centers for Medicare and Medicaid Services (CMS) Hospital Cost Report Information System (HCRIS) to construct estimates of all-payer inpatient hospital costs. Costs per case were measured for each hospital by dividing total inpatient costs (Medicare and non-Medicare costs) by the number of discharges.⁹ We focused on acute care in-patient services and eliminated, to the extent possible, costs associated with hospital subproviders and long-term care units.¹⁰ In addition, we excluded direct GME costs from our measure of inpatient cost per case.

Measuring mission-related activities. To assist in selecting measures of mission-related activities, we used a technical advisory panel consisting of experts from the Commonwealth Fund Task Force on Academic Health Centers. We measured mission-related activities using three sets of variables. Following prior work discussed above, we used the IRB ratio for teaching. To create measures of research, we divided medical schools into four groups (quartiles) based on the amount of their National Institutes of Health (NIH) funding. Medical schools that received the most funding and that accounted for 25 percent of NIH funding dollars in aggregate were placed in the first quartile. Of the remaining schools, those that received the most funding and accounted for the next 25 percent of NIH funding were placed in the second quartile, and so on. We then created a set of dummy variables that indicated for each hospital the quartile of all its affiliated medical schools.¹¹ The dummy variable associated with the highest quartile was used in estimating the models. Other research variables included a dummy variable indicating whether a hospital has a general clinical research center, and a variable indicating the presence of a positron emission tomography (PET) scanner. We constructed measures of hospital standby capacity based on the number of specialty care beds, the availability of sophisticated clinical services, and the number of solid organ transplants.

Regression approach. We estimate two log-linear regression models, where the dependent and continuous explanatory variables are transformed using the logarithmic function.¹² The models differ in their specification of mission-related activities. Exhibit 1 lists the variables included in each model. In both models the in-patient cost per case was the dependent variable and the hospital was the unit of analysis. Teaching intensity was measured as the log of (1 +IRB ratio). All data were for federal FY 1998.

Results

Top Data And Methods Results Discussion Editor's Notes NOTES

 
Summary statistics. Exhibit 2 presents means for inpatient hospital costs per case and measures of mission-related activities. The complete database includes information for 4,389 hospitals. However, some of the models were estimated on a smaller set of hospitals because of missing data. The summary statistics contained in Exhibit 2 represent the 3,552 hospitals for which all data were available. Slightly more than 22 percent of these hospitals are teaching hospitals, eighty-three of which met our definition of an AHC hospital.¹⁶

Regression results. As expected, the size of the coefficient on the IRB ratio falls as additional mission-related variables are added to the model. According to the IME model, all-payer cost per case increase by approximately 5.1 percent for each 10 percent increase in the IRB ratio. In the multiple-missions model this relationship falls to 2.6 percent for every 10 percent change in the IRB ratio.

Of the research measures, the presence of a general clinical research center is statistically significant at the .05 level. The variable indicating whether a hospital is affiliated with a medical school in the top 25 percent of NIH funding is also statistically significant and positive. These results suggest that costs per case are roughly 10 percent higher at hospitals affiliated with a medical school in the top 25 percent of NIH funding than at hospitals that receive no NIH funding. Of the standby capacity measures, all variables are statistically significant at either the .05 or the .10 level, except burn care beds and neonatal intensive care beds.

Mission-related and other teaching hospital costs. Exhibit 3 contains separate estimates for base costs (defined as the difference between total costs and those costs associated with wages, case-mix, and mission-related activities), wages and case-mix, and IME and other mission-related costs. In Model 1, IME costs are much higher for AHC hospitals ($1,793) than for other teaching hospitals ($435) and account for roughly 21 percent of AHC hospitals’ costs per case and 7 percent of costs for other teaching hospitals. Case-mix and wage costs are also higher for AHC hospitals ($2,673) than for other teaching ($1,718) and other large, urban community hospitals ($1,246).

Exhibit 4 presents the percentage distribution of average AHC hospital costs per case by cost category based on the results from Model 2. We find that standby capacity represents the largest mission-related component for AHC hospitals; IME costs and research represent the second- and third-largest components.

View larger version (16K): [in this window] [in a new window]   EXHIBIT 4 Components Of Academic Health Center (AHC) Hospital Costs Per Case, 1998 SOURCE: Results based on Model 2 (multiple-missions model), calculated by the authors. NOTE: Estimates of cost per case exclude direct medical education (DME) costs.

Discussion

Top Data And Methods Results Discussion Editor's Notes NOTES

 
Traditionally, teaching hospitals have used special payments from public payers, such as Medicare GME payments, and higher payments from private payers to support their missions. A recent study suggests that changes in the health care marketplace and attempts by public payers to reduce spending through reductions in the IME adjustment, for example, are challenging these hospitals’ ability to continue to fulfill their social missions.²⁰ In addition, many states are freezing or reducing Medicaid payments to hospitals to deal with state budget crises.

This study should help inform the debate on funding GME by indicating, and quantifying, the range of services associated with its provision in U.S. hospitals. Our findings suggest that reductions in payments to teaching hospitals for IME costs at the federal or state level, on the grounds that these payments only benefit the hospitals and the physicians they train, may be ill-founded. For AHC hospitals, we found that standby capacity represents the largest single component of mission-related costs (if we exclude direct GME costs) at 45 percent, with IME costs accounting for 42 percent and research accounting for 13 percent of mission-related costs. Thus, payments to teaching hospitals support these hospitals’ ability to provide costly standby capacity and conduct research, which benefit the communities that teaching hospitals serve.

Although teaching hospitals pursue multiple missions, the debate over funding teaching hospitals has historically focused on the GME function of these institutions. This practice stems from the long-established use of IRB ratios to determine the additional patient care costs associated with teaching hospitals relative to their nonteaching counterparts. Using this approach, MedPAC has estimated that Medicare costs per case rise approximately 2.7 percent for each 10 percent increase in the IRB ratio, much less than the 2003 Medicare IME adjustment of 5.5 percent. It has been assumed that the number of interns and residents was highly correlated with the other academic and social missions of teaching hospitals, but to this point no research has attempted to quantify the impact of these other missions on teaching hospital costs.

"The future of academic medicine will depend on society’s willingness to support teaching hospitals."

Given the need to limit Medicare spending, policymakers will likely consider proposals to reduce the IME adjustment closer to its "empirical" level of 2.7 percent. This paper shows that studies that rely only on the IRB ratio underestimate the contribution of other teaching hospital missions to patient care costs. We estimated that the costs associated with teaching (IME), research, and standby capacity account for 28 percent of the costs per case at the nation’s AHC hospitals. By comparison, when using only the IRB ratio, IME costs were estimated to be 21 percent of costs per case. Consequently, an IME adjustment of 2.7 percent could result in Medicare payments that are insufficient to finance Medicare’s full share of AHC hospitals’ mission-related costs.

Congress and others have proposed reforms in the financing of AHCs and other teaching hospitals to make it more accountable, predictable, flexible, and transparent. One example that has been discussed is a public trust fund that makes explicit payments to hospitals to replace many of the implicit subsidies that now exist.²¹ The methods and findings in this paper could form a basis for designing long-term reforms, whether through a trust fund or another approach. For example, additional research could explore including measures of standby capacity and research into a revised IME payment adjustment based on our estimated relationships between missions and patient care costs. Although additional research is needed to validate our measures of research and standby capacity, such an approach could improve the equity of additional payments to teaching hospitals.

Independent of any changes in the Medicare IME payment formula, this work has implications for the ways in which AHC and other teaching institutions are held accountable for any additional payments they receive. IME payments are now thought of as supporting only graduate medical training. This view is demonstrated by the Medicare policy that caps IME payments for the number of residents. If we ask whether teaching hospitals are "meeting their responsibilities" while placing conditions on their receipt of IME funds, we should look at the full range of activities they perform—the number of residents trained, research productivity (especially commitment to and involvement in clinical research and technology transfer), and service to the community in the form of standby capacity. This is especially true for Level 1 trauma centers, which some hospitals are under pressure to close because of rising medical malpractice premiums. It may be short-sighted, for example, to cap IME payments based on the number of residents or to reduce state Medicaid payments based solely on workforce considerations when other responsibilities and burdens of teaching hospitals are increasing. One approach that recognizes the full range of teaching hospitals’ missions would be to associate a portion of the IME adjustment with GME activities. The remainder would be recognized as corresponding to the other missions of teaching hospitals. When trying to influence workforce policy, only the educational portion would be changed.

Future efforts to reform payments to teaching hospitals will face serious political and analytical challenges. Although we provide separate estimates of the costs associated with the missions of teaching hospitals, the results are not intended to support a payment "balkanization" strategy, whereby payers choose to pay for some missions but not others. This is because teaching hospitals’ missions are highly interrelated and mutually supportive. Ultimately, however, the future of academic medicine will depend on society’s willingness to support teaching hospitals and the extent to which their missions are seen to be in our common interest.

Editor's Notes

Top Data And Methods Results Discussion Editor's Notes NOTES

 
The authors thank the Commonwealth Fund for providing support for this research. They also thank the American Hospital Association’s Health Forum for permission to use its annual hospital survey data. They are grateful for the helpful comments provided by Melinda Abrams, Brian Biles, Eric Campbell, and participants in the Commonwealth Fund Task Force on Academic Health Centers. Kevin Coleman developed the initial cost decomposition approach used in this paper while at the Lewin Group.

Lane Koenig is a senior manager at the Lewin Group, a health care consulting company in Falls Church, Virginia. Allen Dobson is a senior vice-president and director of the Health Care Finance Practice at the Lewin Group; Jonathan Siegel is a senior associate there, and at the time this paper was written, Silver Ho was an associate. She is now a project manager at PAREXEL International. David Blumenthal directs the Institute for Health Policy, Massachusetts General Hospital/Partners HealthCare System, and is a professor of medicine and health care policy at Harvard Medical School in Cambridge, Massachusetts. Joel Weissman is an associate professor of medicine (health policy) at the Institute for Health Policy and a member of the Department of Health Care Policy at Harvard Medical School.

NOTES

Top Data And Methods Results Discussion Editor's Notes NOTES

 

1. Commonwealth Fund, Leveling the Playing Field: Financing the Missions of Academic Health Centers, Report of the Commonwealth Fund Task Force on Academic Health Centers (New York: Commonwealth Fund, May 1997).
2. J.Newhouse and G. Wilensky, "Paying for Graduate Medical Education: The Debate Goes On," Health Affairs (Mar/Apr 2001): 136–147; and Committee on Implementing a National Graduate Medical Education Trust Fund, Institute of Medicine, "On Implementing a National Graduate Medical Education Trust Fund" (Washington: IOM, 1997).
3. A.Dobson et al., "Financial Performance of Academic Health Center Hospitals, 1994–2000" (New York: Commonwealth Fund Task Force on Academic Health Centers, September 2002).
4. J.Lave, "The Medicare Adjustment for the Indirect Costs of Medical Education: Historical Development and Current Status" (Washington: Association of American Medical Colleges, January 1985).
5. G.Anderson and J. Lave, "Financing Graduate Medical Education Using Multiple Regression to Set Payment Rates," Inquiry 23, no. 2 (1986): 191–199.[ISI][Medline]
6. J.Pettengill and J. Vertrees, "Reliability and Validity in Hospital Case-Mix Measurement," Health Care Financing Review 4, no. 2 (1982): 101–128.[Medline]
7. MedPAC controlled for case-mix, area wages, and Medicare outlier payments. The different findings between Pettengill and Vertrees and MedPAC may be explained by differences in the measurement of hospital costs, increased efficiency of teaching hospitals, or recent improvements in Medicare case-mix measures. Medicare Payment Advisory Commission, Report to Congress: Medicare Payment Policy (Washington: MedPAC, March 2003).
8. R.Mechanic, K. Coleman, and A. Dobson, "Teaching Hospital Costs: Implications for Academic Missions in a Competitive Market," Journal of the American Medical Association 280, no. 11 (1998): 1015–1019.[Abstract/Free Full Text]
9. Although we do not explicitly estimate the cost of charity care, these costs are captured in our all-payer measure, which includes costs for all patients.
10. Subproviders are distinct units within a hospital, such as rehabilitation and psychiatric units. To measure all-payer costs, we summed up inpatient costs from the following cost centers: (1) routine costs (excluding subprovider and long-term care costs); (2) ancillary costs associated with hospital inpatient care (net of subprovider ancillary costs); and (3) other reimbursable costs associated with hospital inpatient care. To identify hospital inpatient costs from the ancillary and other reimbursable cost centers, we multiplied each cost center’s total hospital costs by its associated ratio of inpatient to total charges.
11. We tried a number of different constructs of the NIH funding research variables, including dividing the medical schools into quintiles based on research dollars and then creating dummy variables for the quintiles and also continuous measures of research funding. The approach presented in this paper provided the most consistent and intuitive results.
12. In addition to the two models presented in this paper, we also estimated an "explanatory" model, which included a broader set of covariates that relate to hospital costs, including number of hospital beds, staffing, occupancy rate, and total HMO penetration. To the extent that these variables are correlated with mission, the explanatory model produces lower estimates of mission-related costs. However, the results from this model did not differ materially from the multiple-mission model described in this paper.
13. These are the costs remaining after factoring out wages, case-mix, and mission-related costs and are approximately equal to the average cost of an urban, community hospital with a wage index and case-mix index equal to 1.0.
14. Allocating a cost component first will produce a larger cost than allocating it second or third. For comparison purposes, we also report results using an approach that was neutral to the order in which mission-related variables, wages, and case-mix entered the decomposition. Details are available on request from Lane Koenig at lane.koenig{at}lewin.com.
15. For hospitals with missing data, we imputed the average (weighted by cases) value for a variable according to teaching status. This allowed us to obtain a predicted cost for each teaching hospital. We updated the all-payer cost estimates to 2002 using the CMS Prospective Payment System Market Basket.
16. We identified 124 AHC hospitals and 1,015 other teaching hospitals from the Medicare Inpatient Impact file. Although only a subset of these hospitals are included in the regression analysis because of missing data, we provide national estimates of mission-related costs for all teaching hospitals in the analysis below. Although there is no official list of AHC hospitals, we used the set of integrated teaching hospitals identified by the Association of American Medical Colleges, in which a majority of clinical chiefs of staff are also department chairs in the affiliated medical school. This is the definition used by a number of other studies of AHC hospitals. J.Reuter, "The Financing of Academic Health Centers: A Chart Book" (New York: Commonwealth Fund Task Force on Academic Health Centers, September 1997).
17. We also tried an approach that was neutral to the order in which mission-related variables, wages, and case-mix entered the decomposition. Using this approach, we found that IME and other mission-related costs account for 23 percent of AHC costs per case, compared with 28 percent when mission-related variables are entered into the decomposition first. For other teaching hospitals, IME and mission-related costs account for 10 percent of cost per case, as compared to 11 percent when mission-related variables are entered into the decomposition first.
18. Direct GME costs were estimated using data from cost reports available through HCRIS.
19. A study by Rob Mechanic and colleagues using the IRB ratio estimated total direct and indirect GME-related costs between $18.1 and $22.8 billion in 1997. Updating these figures to 2003 using the Hospital Market Basket indicates GME-related costs between $21.8 to $27.4 billion. These findings are similar to the results reported in this paper. Mechanic et al., "Teaching Hospital Costs."
20. Dobson et al., "Financial Performance of Academic Health Center Hospitals."
21. Commonwealth Fund, Envisioning the Future of Academic Health Centers, Final Report of the Commonwealth Fund Task Force on Academic Health Centers (New York: Commonwealth Fund, February 2003).

                      What's this?

This article has been cited by other articles:

				E. A. Davis, A. Escobar, J. Ehrenwerth, G. A. Watrous, G. S. Fisch, Z. N. Kain, and P. G. Barash Resident Teaching Versus the Operating Room Schedule: An Independent Observer-Based Study of 1558 Cases Anesth. Analg., October 1, 2006; 103(4): 932 - 937. [Abstract] [Full Text] [PDF]

				H. Moses III, S. O. Thier, and D. H. M. Matheson Why Have Academic Medical Centers Survived? JAMA, March 23, 2005; 293(12): 1495 - 1500. [Abstract] [Full Text] [PDF]