Abstract
BACKGROUND We wanted to evaluate the most recent, complete data related to the specific effects of the Balanced Budget Act of 1997 relative to the overall financial health of teaching hospitals. We also define cost report variables and calculations necessary for continued impact monitoring.
METHODS We undertook a descriptive analysis of hospital cost report variables for 1996, 1998, and 1999, using simple calculations of total, Medicare, prospective payment system, graduate medical education (GME), and bad debt margins, as well as the proportion with negative total operating margins.
RESULTS Nearly 35% of teaching hospitals had negative operating margins in 1999. Teaching hospital total margins fell by nearly 50% between 1996 and 1999, while Medicare margins remained relatively stable. GME margins have fallen by nearly 24%, however, even as reported education costs have risen by nearly 12%. Medicare+Choice GME payments were less than 10% of those projected.
CONCLUSIONS Teaching hospitals realized deep cuts in profitability between 1996 and 1999; however, these cuts were not entirely attributable to the Balanced Budget Act of 1997. Medicare payments remain an important financial cushion for teaching hospitals, more than one third of which operated in the red. The role of Medicare in supporting GME has been substantially reduced and needs special attention in the overall debate. Medicare+Choice support of the medical education enterprise is 90% less than baseline projections and should be thoroughly investigated. The Medicare Payment Advisory Commission, which has a critical role in evaluating the effects of Medicare policy changes, should be more transparent in its methods.
- Education, medical, graduate
- Medicare
- Medicare Payment Advisory Committee
- economics, hospital
- economics, medical
INTRODUCTION
The Balanced Budget Act of 1997 (BBA97) included the largest cuts in the history of Medicare,1 accounting for almost one half of the total reduction in federal expenditures.2 BBA97 was projected to reduce Medicare payments by $119 billion, including $2.3 billion in graduate medical education (GME) payment reductions. Medicare is the largest single source of financing of GME, which accounted for 7% of Medicare expenditures by 1999.2 Medicare funds GME with 2 distinct payments: direct medical education payments, designed to cover the direct costs associated with training residents; and indirect medical education adjustments for additional patient care costs associated with training. Indirect and direct medical education payments were $4.1 billion and $2.2 billion, respectively, in 1998.2, 3 Of course, teaching hospitals also shared in the $212 billion Medicare paid for health care in 1999.4
Specific provisions included caps on GME-funded resident positions set retrospectively to December 1996, a 29% reduction in indirect medical education adjustments over 4 years,5 voluntary resident reductions with temporary financial incentives, limited payment for outpatient training, reductions in payments to hospitals for uninsured care (disproportionate share payments),6 5-year phased-in GME payments from managed Medicare (Medicare+Choice), a required report from the Health Care Financing Administration (now Centers for Medicare & Medicaid Services) on disparities in direct medical education payments to hospitals,3 and phased reduction in payments for unpaid patient care debts.
BBA97 Backlash
BBA97 generated a political backlash fueled by projections and anecdotes. An early report from the Medicare Payment Advisory Commission (MedPAC), which advises Congress about the Medicare program, projected that hospitals would lose 8.9% from inpatient prospective payment system (PPS) margins by 2002 but would still enjoy higher overall margins than before the BBA97. MedPAC also projected that major teaching hospitals would fare better than other hospitals.6 The Association of American Medical Colleges (AAMC) presented a strikingly different picture, projecting an average $47 million loss for members of the AAMC Council of Teaching Hospitals and Health Systems (COTH) by 2002 compared with $17 million for other teaching hospitals and $9 million for nonteaching hospitals. The overall margins of COTH hospitals were projected to fall from 4% to about 1.3%.7, 8
In 1999, the Congressional Budget Office reported that BBA97 had overshot its Medicare spending reduction mark by $88 billion. This amount appeared to corroborate AAMC projections and other anecdotal reports of worsening financial performance for hospitals.9 For example, after strong revenue reports for several years, the University of Pennsylvania reported a $198 million deficit in 1999. At least 3 other academic medical centers posted losses exceeding $50 million that same year.10, 11 The Commonwealth Fund also indicated that academic health centers shouldered an increasing uncompensated care burden even as they faced declining margins.12
Conjecture vs Reality: Incorporating Real Figures Into the Analysis
Although assessments of the BBA97 impact rely on projections or partial data, actual data about hospitals’ financial health and status of their GME programs through 1999 have recently become available.9, 13, 14 MedPAC first used actual hospital cost reports and determined that the proportion of teaching hospitals with negative overall margins increased from 25% in 1996 to 38% in 2000.15 We sought to verify the MedPAC assessments of the financial health of teaching hospitals and clarify the methods used to examine the effects of BBA97. We also evaluated differences between types of teaching hospitals using 2 distinct groups: those that belong to the AAMC COTH, and hospitals with only family medicine residencies (family practice single-residency hospitals). We sought to (1) evaluate the financial health of teaching hospitals; (2) evaluate effects of the BBA97 on Medicare payment margins; (3) evaluate changes in full-time-equivalent residents since BBA97; and (4) evaluate differences by teaching hospital type, COTH vs non-COTH, and family practice single-residency vs multiple-residency hospitals. We also define cost report variables and calculations necessary for continued impact monitoring.
We chose to study cohorts to reflect opposite ends of the teaching hospital spectrum. AAMC COTH hospitals comprise the largest part of the medical teaching enterprise and have sustained advocacy for BBA97 relief. Family practice single-residency teaching hospitals are most unlike COTH hospitals, tend to be more diversely distributed across rural-urban and community size, and place many graduates in underserved communities. We hypothesized that COTH hospitals would have better financial margins and realize less impact on full-time-equivalent residents than other teaching hospitals. We also hypothesized that family practice single-residency hospitals would fare worse than COTH hospitals.
1998 was the first year of a 5 year phase-in of Medicare+Choice (managed Medicare) direct medical education payments, and it was anticipated to provide $4 billion in funding to teaching hospitals by 2002 (about $730 million in 1998, rising to about $2.6 billion in 2002).16– 19 There is, however, little evidence accounting for Medicare+Choice direct medical education payments, and we hypothesized that the reported payment amounts would be lower than expected.
METHODS
The sample frame consisted of all teaching hospitals in the 1996 Medicare hospital cost report public user file (PPS13). Hospitals were included if they also received Medicare GME payments in 1998 and 1999 (PPS15 and 16) and if reports for each of the 3 years included at least 364 days but not more than 366 days (to permit direct comparisons of annual data). Hospital cost reports are filed annually and capture income and costs from all sources. Closed Medicare public user files of hospital cost reports were used for all analyses. Sole community hospitals were excluded because of their special Medicare payment status. Excluded hospitals were compared with those that included using self-classification variables.
COTH status was determined using a file provided by the AAMC, which included 278 unique hospital provider identification numbers (received February 2001). Family practice single-residency designation relied on a survey of family practice residency programs asking whether their residents trained in hospitals with no other residents.20 Hospitals were also excluded from family practice single-residency status if they sponsored other residency programs according to the American Medical Association Fellowship and Residency Interactive Database or if the Accreditation Council on Graduate Medical Education listed them as a teaching institution for other residents. All other teaching hospitals were classified as multiple residency.
In 2000 MedPAC began using a new formula for Medicare margins that included a PPS, GME, bad debt, skilled nursing facility, home health, and non-PPS Medicare payments and costs. MedPAC published the formula for calculating PPS inpatient margins but declined to reveal the actual variables used to calculate the overall Medicare margin.21 Because public user files do not contain information on skilled nursing facility, home health, or non-PPS payments, we constructed a Medicare margin formula using PPS (both inpatient and outpatient), GME, and bad debt variables. We believe our Medicare margin formula closely approximates that of MedPAC, although skilled nursing facility and home health payments were 13.2% of total Medicare payments in 1999.22
Descriptive statistical analyses of GME payments and hospital margins (PPS, GME, bad debt, and total Medicare margins) were done for each group. Margins are the difference between payments and costs, divided by payments. PPS payments include part A payments for inpatient services, indirect medical education adjustments, and disproportionate share and capital payments, as well as outpatient coinsurance deductibles and Part B payments. PPS costs included inpatient operating and capital costs, as well as outpatient Part B operating costs. Because PPS cost variables exclude education costs, we subtracted indirect medical education adjustments from the PPS margin calculations to avoid inappropriately inflating the margin. GME payments included indirect medical education adjustments, direct medical education payments (parts A and B). Costs included direct salary, benefits, and program costs, and indirect paramedical education. Bad debt margins included part A and B payment and cost variables. Calculation of overall Medicare margins included payments and costs from PPS, GME, and bad debt, and avoided double-count of indirect medical education payments.
We were concerned that averaging margins might normalize effects across hospital groups, when, in fact, there is considerable variation in hospital payments. For example, a hospital with 1 resident might have the same Medicare margin as a hospital with 500 residents but have very different payments and costs. To account for this difference, we also calculated margins for each hospital cohort using total payments and costs to dollar-adjust the effect for volume of GME dollars and costs. Finally, total margins were calculated and included payments and costs from all sources including Medicare.
Financial reserves were calculated as the difference between total assets less fixed assets and liabilities. We subtracted fixed assets because these cannot be readily liquidated. We included accounts receivable even though these might overestimate actual reserves. Full-time-equivalent resident costs and Medicare+Choice direct medical education payments came directly from cost reports. All formulas and variables are displayed in the supplemental Appendix, which is available online at http//www.annfammed.org/cgi/content/full/2/1/71/DC1.
RESULTS
Sample
Selection criteria reduced the sample to 713 of a potential 1,241 hospitals that received any GME payments in 1999 (57.5%). Those hospitals included were significantly more likely to be general short-stay hospitals, to be for profit, and had a significantly higher mean number of beds (Table 1⇓). Although COTH and family practice single-residency hospital cohorts are not mutually exclusive, only 4 of 108 family practice single-residency hospitals were members of COTH (Table 2⇓).
Total Margin, Financial Reserves, and Medicare Margin
By 1999 the average teaching hospital margin was 2.5%, and nearly 35% were operating with negative total margins. Between 1996 and 1999, mean total margins fell more than 50% for all teaching hospitals except family practice single-residency hospitals, for which total margins fell only 21% (Table 3⇓). Despite this difference, the proportion of family practice single-residency hospitals operating with negative total margins nearly tripled.
Medicare margins remained relatively stable, averaging nearly 13% for all teaching hospitals (Table 4⇓). The major component of the Medicare margin is the PPS margin, which on average rose from 17.5% to 23.5% between 1996 and 1999. The dollar-adjusted PPS margin was not substantially different. Bad debt margins fell 33% in 1998 and 67% in 1999, reflecting scheduled reductions.
GME margins have the widest variation, even within hospital cohorts. COTH hospitals had the best median GME margins, and family practice single-residency hospitals had the worst, but both experienced substantial declines in GME margins. Using dollar-adjusted GME margins made a considerable difference and brought down both COTH and non-COTH margins. Average 1999 financial reserves were $17 to $30 million (Table 3⇑).
GME Payments
Direct medical education payments actually increased for all groups except family practice single-residency hospitals (Table 5⇓). Indirect medical education payments declined, however, by an average of 12.1% even as reported teaching costs increased on average by 11.9%. Total Medicare+Choice GME payments to teaching hospitals were only $31 million for 1998, the first year of the 5-year phase-in of Medicare+Choice payments (Table 6⇓), an amount significantly and substantially lower than any projections.23 In 1999, Medicare+Choice GME payments rose to $69 million, still only a fraction of that projected.
Resident Full-Time-Equivalent Changes
In aggregate, approved full-time-equivalent resident positions did not change much between 1996 and 1999. On average, almost 1 primary care full-time-equivalent position was added by hospitals (Table 7⇓).
DISCUSSION
Teaching hospitals of all sorts are in dire straits. Contrary to our hypothesis, COTH hospital Medicare margins are 50% less than non-COTH hospital Medicare margins, and COTH hospital overall margins are nearly 25% lower. Also contrary to our hypothesis, family practice single-residency hospitals have better Medicare margins, and their total margins are more than double those of multiple-residency hospitals. Despite having better average margins, however, the number of family practice single-residency hospitals with negative margins tripled in just 3 years, and family practice single-residency hospitals had the lowest median GME margins. Average hospital financial reserves provide little protection given narrow total margins and the overall costs of teaching and service missions. Despite funding caps, the number of full-time-equivalent residents remained stable with slight growth in primary care positions between 1996 and 1999.
We found support for our hypothesis that Medicare+Choice GME payments would be low, but were surprised that they were only 5% to10% of expected. The difference may contribute to the dramatic decline in GME margins. Inquiries of Centers for Medicare & Medicaid Services regarding this finding remain unanswered, and the 2002 MedPAC report to Congress did not provide an explanation. Payments of only 5% to 10% of expected may reflect an avoidance of teaching hospitals by Medicare+Choice plans or a failure to collect GME payments. Medicare+Choice enrollment actually peaked in 1999 within the anticipated range at nearly 6.5 million enrollees, so reduced enrollment is not a likely explanation.24 A fuller accounting of Medicare+Choice GME payments is needed.
Total margins, GME, and bad debt margins were in steep decline between 1996 and 1999; however, overall Medicare margins for teaching hospitals appear to be relatively stable. Although this finding supports our hypothesis that the financial health of teaching hospitals has been affected by BBA97, it suggests that other forces contributed as much or more.
Contrast with MedPAC 2002 Report
In March 2002, MedPAC reported on the financial standing of 852 of 1,107 teaching hospitals (no sampling method described). It is not clear whether MedPAC calculated and averaged margins for each hospital, or whether total payment and cost dollars for each hospital type were used. MedPAC reported total margins, overall 1999 Medicare margins, and the percentage with negative margins as 2.4%, 13.1%, and 43.0% for major teaching hospitals, and 4.0%, 5.1%, and 31.1% for other teaching hospitals. It is unclear whether our COTH cohort corresponds with the MedPAC major teaching category, but assuming it does, the major difference in our findings is the Medicare margin, which we found to be higher for non-COTH hospitals (8.2% for COTH vs 17.7% for non-COTH hospitals in 1999). MedPAC reported that between 1996 and 1999 the overall Medicare margin fell 23.5% for major teaching hospitals and 46.9% for all other hospitals. We found Medicare margins to be more stable than did MedPAC. It will be essential to know the MedPAC formula constructs to explain these differences.
Relief Before Research
In response to aggressive lobbying, Congress passed legislation designed to ameliorate the effects of the BBA97 in late 1999. The Medicare, Medicaid, and SCHIP Balanced Budget Refinement Act (BBRA) delayed indirect medical education adjustment cuts by 2 years and partially restored disproportionate share payments. BBRA also established a new average payment method for direct medical education payments, with minimum payments set at 70% of the locally adjusted national average. The BBRA increased resident caps for primary care programs by up to 3 full-time-equivalent positions and allowed rural hospitals to increase their cap by up to 30%. The new legislation did not allow adjustment of the 1996 cap to account for full-time-equivalent residents in ambulatory settings before 1997.13, 25
Whereas the BBRA took important steps to reverse the unintended negative impact of the BBA97 on primary care training programs and rural hospitals, its main effect was to postpone it by 1 to 2 years. One study projected that the BBRA would restore just 8.6% of Medicare payment reductions.14 The Medicare, Medicaid and SCHIP Benefits Improvement and Protection Act of 2000 (BIPA) provided another $11.5 billion in relief for hospitals distributed over 5 years. Indirect medical education payments were frozen for 1 additional year, and disproportionate share payment reductions slowed. Minimum direct medical education payments were further raised to 85% of the adjusted national average to correct well-documented variations.26, 27 As later cost reports become available, it will be possible to measure BBRA and BIPA effects.
Our study has several limitations. We are able to report the impact of BBA97 only through 1999 because of the time lag in hospital cost report closure. We were not able to include many teaching hospitals. Excluded hospitals were less likely to be short-stay hospitals, but this limitation should not affect generalizability, because most of this difference is explained by the exclusion of children’s, psychiatric, and rehabilitation hospitals, none of which received GME payments for the period studied. Included hospitals were more likely to be for-profit and to have more beds, both of which could inflate margins for the included hospitals. MedPAC experienced similar difficulty in evaluating all teaching hospitals.
The skewness of GME margins may be a result of the variability of direct medical education, but questionable validity of GME cost reporting could also be a factor. Since the uncoupling of actual GME costs from payments in 1984, there has been little incentive to account for these costs correctly and perhaps a real incentive to assign them to other cost centers where payments are linked to costs. This possibility suggests that GME margins might actually underestimate the real costs of educating resident physicians. Given available information, we were able only to approximate MedPAC impact analyses, and we did not include Medicare skilled nursing facility, home health, and non-PPS payments or costs, which might explain the differences in our findings for overall Medicare margins. Even so, these 3 items make up a small proportion of overall Medicare costs and payments.
CONCLUSION
Our findings indicate that the period of 1996 through 1999 was one of deep cuts in hospital profitability. These cuts are not entirely attributable to the BBA97. Overall, Medicare payments remain an important financial cushion for teaching hospitals, which is poor consolation when more than one third of teaching hospitals already operate in the red, and most others have slim margins and financial reserves. How will teaching hospitals cope financially with patient safety mandates, increasing pressure to improve resident work environments and hours, rising malpractice premiums, and other rising health care costs? If changes in the non-Medicare health care marketplace have had a greater impact on the financial health of teaching hospitals, the near future does not offer much optimism.
What role should Medicare have in rescuing these hospitals? Family practice single-residency hospitals, on average, appear financially healthier; however, their larger shifts into negative operating margins and low GME margins suggest they might be particularly vulnerable to market and Medicare payment changes. Both classes of teaching hospital are valuable resources that warrant specific monitoring and, perhaps, advocacy. The substantial declines in GME payments relative to steep GME cost increases mean that the role of Medicare in supporting this important function needs special attention in the overall debate. To their credit, teaching hospitals have absorbed this real loss in support of their teaching mission.28 BBRA and BIPA helped delay GME reductions, but their recent expiration caused additional indirect medical education reductions of $600 to $800 million.29, 30
MedPAC recently proposed another 50% reduction in indirect medical education adjustments but decided against it.31 Medicare+Choice support of the medical education enterprise is 90% less than any baseline projections and should be thoroughly investigated to be sure these programs are not short-changing teaching hospitals purposefully or otherwise. Finally, MedPAC has a critical role in evaluating the effects of rather blunt Medicare policy changes and should be more transparent in its methods.
Acknowledgments
We would like to thank Michael S. Hadad, Lisa Nijm, David Meyers, MD, and Kenneth Fink, MD, for their thoughtful reviews and comments, and Lisa Smith for her considerable help with manuscript preparation.
Footnotes
Conflicts of interest: none reported
Funding support: This project was supported by the American Academy of Family Physicians through its funding of The Robert Graham Center.
Disclaimer: The views expressed are those of the authors. No official endorsement by the Agency for Healthcare Research and Quality is intended or should be inferred. The information and opinions contained in research from The Graham Center do not necessarily reflect the views or policy of the AAFP.
Limited, preliminary results of this study were presented as posters at the 2001 Society of Teachers of Family Medicine and North American Primary Care Research Group meetings.
- Received for publication December 3, 2002.
- Revision received March 10, 2003.
- Accepted for publication March 26, 2003.
- © 2004 Annals of Family Medicine, Inc.