The policy debates concerning heart hospitals boil down to two issues: the extent to which these hospitals compete unfairly with traditional hospitals, which must cross-subsidize non-cardiac care, and the extent to which physician ownership affects the patient mix.1 Largely ignored in these debates is the extent to which traditional hospitals receive better reimbursement from both the government and private health insurers to cross-subsidize the losses incurred by providing certain non-cardiac services.2
While cross-subsidization is an unfamiliar concept to many, it is not a hard concept to grasp.3 By law, hospitals are required to provide a multitude of money-losing services (e.g. emergency care). To cover such losses, hospitals tend to maximize their income from profitable services (e.g. invasive cardiac services). Indeed, the need to provide profitable cardiac services in order to offset the losses arising from providing emergency room services helps to explain why urban regions have a surplus of hospitals that provide invasive cardiac services (ICS) such as coronary artery bypass grafting (CABG) and percutaneous coronary intervention (PCI).4 Conversely, to the degree that government and private insurers provide premium reimbursement for ICS, these payers tacitly place their seal of approval on such fiscal sleight of hand. To a degree, the discussion on cross-subsidization has been muted by a lack of objective data. However, now that the Department of Health and Human Services publishes hospital Diagnosis Related Group (DRG)-specific reimbursement on its Hospital Compare website,5 it is possible to gain some insight into a hospital’s revenue stream.6 Accordingly, by using this publicly available data, we examined Medicare reimbursement for ICS to heart and traditional hospitals. From these data we concluded that Medicare intentionally overpays traditional hospitals for ICS in order to facilitate cross-subsidization.
Methods
Nine hospital markets for ICS were selected for review. Markets were defined as all hospitals providing ICS within a 25-mile radius of a heart hospital. A hospital was considered to provide ICS if it offered either CABG or PCI services to the public.
The key criterion for a market’s inclusion in this study was its geographical location. Markets were selected from diverse geographical regions to avoid the impact of a single state’s law on the market’s behavior. Secondarily, markets were selected to provide diversity in the population density per metropolitan statistical areas and to provide diversity with respect to heart hospital market penetration. Estimates of market size were based on the total amount of Medicare reimbursement for ICS for fiscal year 2006. (As a corollary, because data from 2006 are used in this paper, ‘classic’ DRG numbering is used rather than the MS-DRG numbering system in current use.)
In the selected markets, heart and traditional hospitals were compared based on DRG-specific volume and reimbursement data for ICS, other ancillary invasive cardiac services (implantation of heart values, defibrillators, and pacemakers), management of cardiac conditions (chest pain, heart failure, and myocardial infarction [MI]), major vascular procedures, and medically managed conditions (diabetes, pneumonia, and chronic obstructive pulmonary disease [COPD]). The data are presented as average ± standard deviation. Statistical comparisons were performed by Student’s T-test.
Results
Table 1 summarizes the demographics for the nine study markets. Because the Indianapolis and Wichita markets each contained two heart hospitals, the total number of heart hospitals under review is 11 (or 10% of all heart hospitals operating in the US). On average, each market had 8.1±5.8 traditional hospitals (range: three to 20 hospitals) competing with the heart hospital(s) to provide ICS. The average market size for ICS was $13.1±8.4 million (range: $2–25 million), with heart hospitals having an average market share of 36.1% (range: 0.3% heart hospital market share for the Naperville market to a 70% heart hospital market share for the Bakersfield market).
The two markets with the most traditional hospitals in competition (Naperville and Indianapolis) were examined to determine the within-market variation in ICS reimbursement (see Table 2). Substantial variation in DRG-specific reimbursement was observed between traditional hospitals in the Naperville market for DRG-549 complex CABG ($19,772–45,691; a difference of 251%), DRG-550 CABG ($19,772–39,997; a difference of 202%), and DRG-588 PCI ($11,533–17,992; a difference of 156%). The findings in the Indianapolis market were similar: DRG-549 complex CABG ($24,898–38,057; a difference of 153%), DRG-550 CABG ($20,595–26,929; a difference of 131%), and DRG-588 PCI ($10,911–17,820; a difference of 163%).
Table 3 summarizes the average DRG-specific Medicare reimbursement to the 11 heart hospitals and 65 competing traditional hospitals. No significant differences in Medicare reimbursement were observed between heart and traditional hospitals for DRG-122 myocardial infarction ($5,510±1,686/ patient versus $6,171±1,703/ patient; p=NS) and for DRG-143 chest pain ($3,365±1,252 /patient versus $3,709±961/patient; p=NS). However, when heart and traditional hospitals are compared by surgical procedures, traditional hospitals were significantly better reimbursed for the following procedures: DRG-105 heart valve implantation ($33,710±4,056/case versus $39,819±6,356/case; p=0.001), DRG-110 major vascular procedures ($21,432±2,688/case versus $24,202± 7,264/case; p=0.027), DRG-515 defibrilator implantation ($29,186±2,824/case versus $34,381±6,358/case; p<0.001), DRG-549 complex CABG ($27,929±4,658/case versus $26,695±5,760/case; p=0.004), DRG-550 CABG ($20,281± 3,047/case versus $23,958±4,562/case; p=0.004), DRG-552 pacemaker implantation ($11,245±706/case versus $13,212± 2,043/case; p<0.001), and DRG-588 PCI ($11,230±742 versus $13,3472,662/case; p<0.001). With a few exceptions, in all nine markets, heart hospitals were reimbursed the least by Medicare for providing ICS.
Interestingly, this pattern of reimbursement carried over to non-surgical services. When heart and traditional hospitals are compared by reimbursement for medical conditions, Medicare reimbursed traditional hospitals significantly better for the following procedures: DRG-88 COPD ($4,893±802/patient versus $5,641±841/patient; p=0.013), DRG-89 pneumonia ($5,708±763/patient versus $6,456±1,136/patient; p=0.012), DRG-127 management of heart failure ($5,622±489/patient versus $6,482±1,010/ patient; p<0.001), and DRG-294 diabetes ($4,115± 355/patient versus $4,963±812/patient; p<0.001).
Table 4 summarizes the hospitals’ ICS volumes and reimbursement. While 66% of heart hospitals performed more than 10 DRG-549 complex CABG procedures, only 25% of traditional hospitals performed more than 10 DRG-549 procedures. However, on average, heart and traditional hospitals performed the same number of DRG-549 complex CABGs (33±20 DRG-549 CABGs/heart hospital versus 22±8 DRG-549 CABGs/traditional hospital; p=NS). Therfore, on average, heart and traditional hospitals received the same total revenue in the form of Medicare reimbursement for providing DRG-549 complex CABG procedures ($0.7±0.4 million for DRG-549 CABGs/heart hospital versus 0.7± 0.3 million for DRG-549 CABGs/traditional hospital; p=NS).
Similarly, while 100% of heart hospitals performed more than 10 DRG-550 CABGs, only 66% of traditional hospitals performed more than 10-DRG 550 procedures. Again, on average, heart and traditional hospitals performed the same number of DRG-550 CABGs (49±44 DRG-550 CABGs/heart hospital versus 32±35 DRG-550 CABGs/traditional hospital; p=NS). Not surprisingly, on average, heart and traditional hospitals received the same total revenue in the form of Medicare reimbursement for providing DRG-550 services ($1.0±0.8 million for DRG-550 CABGs/heart hospital versus 0.8±0.6 million for DRG-550 CABGs/traditional hospital; p=NS).
By contrast, all hospitals reviewed performed more than 10 DRG-588 PCIs. However, the volume of DRG-588 PCIs performed in heart hospitals was significantly greater than the volume of PCIs performed in traditional hospitals (153±70 DRG-588 PCIs/heart hospital versus 72±78 DRG-588 PCIs/traditional hospital; p=0.005). This increased volume translated to a significantly better Medicare reimbursement revenue stream for heart hospitals for providing DRG-588 PCIs ($1.7±0.7 million for DRG-588 PCIs/heart hospital versus $0.9±1.0 million for DRG-588 PCIs/traditional hospital; p=0.016).
Discussion
Medicare DRG reimbursement turns principally on two factors.7 First, there is a weight factor, which reflects the average level of resources for an average Medicare patient in the DRG. More generally, Medicare’s DRG weight factor represents the methodology by which the government provides greater reimbursement for complex healthcare services that consume more healthcare resources. Second, the hospital factor reflects a hospital’s geographical location and intensity of market competition. In theory, by adjusting one factor or the other, the government can tailor a hospital’s reimbursement to reflect a hospital’s actual expenses.
In a small geographical hospital market, such as those reviewed in this study, it is likely that all hospitals are purchasing their supplies from a common pool of vendors, face similar medical malpractice liabilities, and have only minor variations in the general level of patient health. These observations suggest that, with the possible exception of academic medical centers (AMCs) that attract a sicker class of patients, the actual expenses incurred by traditional hospitals for providing specific healthcare services should be similar. Accordingly, in narrowly defined geographical markets, it would be expected that Medicare’s actual reimbursement to hospitals for DRG-specific services would be similar.
However, this is not what is observed. For example, in the Naperville and Indianapolis markets, Medicare reimbursement to traditional hospitals varied from 31 to 151% depending on the DRG. Nor can the variability in hospital-specific reimbursement be explained by the presence of an AMC because in both the Naperville and Indianapolis markets the AMCs did not receive the best reimbursement for DRG-549, 550, or 588. Moreover, if we compare Medicare’s DRG-specific reimbursement to traditional and heart hospitals, the difference in reimbursement becomes significant. Of the 13 categories of DRG reimbursement reviewed, we found that in 11 categories traditional hospitals were reimbursed significantly better than heart hospitals.
So, why does the rate of Medicare DRG-specific reimbursement vary between hospitals in close proximity to one another? One explanation is that traditional hospitals receive significantly better reimbursement for ICS because Medicare has assigned these hospitals a greater DRG weight to compensate the traditional hospitals for having a more complex case mix. At first glance, this explanation may seem reasonable because prior government studies have observed that heart hospitals tend to provide ICS to low-surgical-risk/high-profit patients.
While this ‘cherry-picking’ reimbursement theory for ICS is interesting, it cannot completely explain the observed significant variations in Medicare reimbursement. First, heart hospital cherry-picking alone cannot explain the tremendous variation in Medicare reimbursement for ICS observed among traditional hospitals. In addition, many of the traditional hospitals reviewed in this study performed fewer than 10 CABG or 10 PCI procedures per year. Thus, any attempt to explain away the variability in Medicare reimbursement to hospitals because of variations in small operating volumes seems neither logical nor fair. Second, even if heart hospitals are entitled to less reimbursement because they cherry-pick patients for ICS, there does not appear to be an explanation for why heart hospitals are reimbursed less for a number of DRG-specific medical conditions (e.g. diabetes, COPD, and pneumonia) or major vascular surgeries.
Rather, we concluded that the explanation for the significant variation in DRG-specific reimbursement observed between heart and traditional hospitals is evidence of a tacit acknowledgment by the government that without the ability to cross-subsidize, traditional hospitals will fail. By incurring greater losses by providing non-ICS, traditional hospitals require greater reimbursement for ICS (and other medical conditions) to remain solvent. Conversely, Medicare appears to offer heart hospitals less reimbursement for healthcare services because of a perception that heart hospitals, which provide fewer ancillary services, are in need of less cross-subsidization.
Some readers might wonder why traditional hospitals have for so long tolerated a system of reimbursement that requires cross-subsidization. There are at least two reasons. First, there is the reality of politics: politicians like to enact unfunded mandates (e.g. the Emergency Medical Treatment and Active Labor Act [EMTALA]8) that provide benefits but do not require government funding. Given the stiff penalties for violating EMTALA, hospitals are compelled to provide emergency services, even if doing so results in a financial loss. Second, the most profitable markets are those markets with the most knowledge asymmetry, i.e. those markets with the least transparency.9 The lack of transparency in Medicare reimbursement regulations means that some hospitals are likely to have profited handsomely when Medicare’s reimbursement has been in excess of the expenses actually incurred by the hospital. Any hospital that has profited because of a lack of transparency in Medicare’s regulation is unlikely to complain about the system.
Study Limitations. The limitations of this study arise from the hospital selection process and the nature of the data found on the Hospital Compare website. Because the approximately 100 heart hospitals tend to be clustered in southern and western states (where the certificate of need requirements are minimal), had the study markets been selected randomly, the hospital markets chosen would likely have been clustered in heart hospital friendly states. So, to avoid a cluster bias, we selected our study markets by hand. However, because the nine hospitals reviewed contained 10% of all heart hospitals, our results should be representative of all heart hospital markets.
Second, to ensure patient confidentiality, Hospital Compare does not disclose DRG-specific volumes unless a hospital has treated 11 or more patients for that condition. To handle this lack of precision, we treated all hospitals that provided 10 or fewer patients with a DRG-specific service as if they had not provided that service at all. Handling the data in this way may have had a disproportionately negative impact on traditional hospitals, which frequently provided fewer than 10 ICS to Medicare patients per year. However, if this error were corrected by the inclusion of more accurate hospital volume data, it would only increase the reimbursement provided by Medicare to traditional hospitals; thereby reinforcing the conclusion that Medicare provided traditional hospitals with more favorable reimbursement.
Implications. After President Obama proposed funding his version of healthcare reform by cutting reimbursement to hospitals, the American Hospital Association (AHA) cried foul.10 According to the AHA, President Obama’s proposal would result in hospitals losing $220 billion over the next 10 years. In a market that already operates on razor-thin profit margins, many hospitals are likely to fail if $220 billion is removed from the market.
The significant variations in Medicare reimbursement observed between traditional hospitals for ICS in narrowly defined geographical markets suggest that many of these hospitals are being over-compensated for ICS. So, the President is probably correct that Medicare reimbursement to many traditional hospitals for ICS could be reduced and that those hospitals would still be able to make a profit by providing ICS. However, reducing Medicare reimbursement to traditional hospitals will impair the ability of these hospitals to cross-subsidize. So, if the government does cut Medicare reimbursement for ICS, it should consider improving the compensation for the services that are currently being under-reimbursed. In particular, if the government cuts Medicare reimbursement of ICS, it should start to reimburse traditional hospitals for providing emergency room, and other, services.
In short, if the government attempts to control healthcare costs with a blunt knife, such as an across-the-board reduction in DRG reimbursement, all hospitals will be negatively affected. The magnitude of this impact will be determined not only by the volume of Medicare patients a hospital serves, but also by the degree to which a hospital must cross-subsidize to survive. Accordingly, if the government fails to consider the need of hospitals to cross-subsidize healthcare services, any healthcare reform will likely result in decreased access to care.