Analysis of Bundled Payment

Spending

Bundled payment approaches have the potential to reduce spending:

• Bundled payment approaches create incentives for providers to eliminate unnecessary services and reduce costs. Read more below
• From the limited evidence in the literature, we would expect that bundled payment approaches involving multiple providers would lead to decreased spending. Read more below
• Savings will depend on the design of the payment system and the particular services that are bundled. Read more below

Bundled payment approaches create incentives for providers to eliminate unnecessary services and reduce costs.

A bundled payment approach, whereby multiple providers are reimbursed a single sum of money for all services related to an episode of care (in this case, a hospitalization plus a period of post-acute care), rather than being reimbursed for each individual service, should reduce spending by reducing the volume of services provided. Whereas in a fee-for-service system, reimbursement is directly related to the volume of services provided and there is little incentive to reduce unnecessary care, the use of bundled payment mechanisms promotes a more efficient use of services.

Bundled payment is intended to decrease spending via several mechanisms: a reduction in the number of unnecessary physician services during a hospitalization, more judicious use of health care resources during the hospital stay, and a reduction in postdischarge costs, including unnecessary post–acute care services and avoidable readmissions (Medicare Payment Advisory Commission [MedPAC], 2008). If the costs of an episode of care are less than the bundled payment amount, the providers (hospital and physicians) can keep the difference; if the costs of care exceed the bundled payment, the providers bear the financial liability. Bundled payment systems can yield savings for payers if a discounted rate is negotiated at the outset or if payment amounts are adjusted downward to reflect the efficiencies achieved after the system is in place.

One of the unintended consequences of reducing reimbursement surrounding a hospitalization is that it could result in a reduction in both necessary and unnecessary care. To minimize this possibility, we assume that quality of care standards would have to be met as part of the bundled payment system. We also assume that some type of case mix adjustment would be included in the calculation of the bundled payment amount.

From limited evidence in the literature, we would expect that bundled payment approaches involving multiple providers would lead to decreased spending.

The evidence is limited regarding bundled payment for an episode of care surrounding a hospitalization. Most of the literature on bundled payment comes from the Health Care Financing Administration's (HCFA) Medicare Participating Heart Bypass Center Demonstration, which ran from 1991 to 1996. Four hospitals were chosen for the demonstration, and each was paid a single fee for all inpatient and physician services for heart bypass patients (for coronary artery bypass graft [CABG] surgeries). Note that many providers opposed the demonstration. They raised concerns about the government designating some providers as "high quality," with an associated increase in payment. Despite this concern, a large number of hospitals applied to participate (MedPAC, 2008).

As for savings to Medicare, Cromwell, Dayhoff, and Thoumaian (1997) found that Medicare spending declined during the first two years of the demonstration, with a total savings to the Medicare program and its beneficiaries of $17.2 million (a 15.5 percent decrease) between 1991 and 1993. Inpatient savings constituted 85–93 percent of total savings. Wynn's (2001) overview of the project found that, for the duration of the demonstration project, total savings by Medicare were $52.3 million, of which $42.3 million was due to discounts negotiated with the hospitals and $7.9 million was from reduced copayments.

Cromwell, Dayhoff, and Thoumaian (1997) also found that hospital costs declined during the period of the study; in three of the four hospitals, average total costs per case fell by 2–23 percent. Liu, Subramanian, and Cromwell (2001) found that, after controlling for preoperative risk factors and postoperative outcomes, all four hospitals had significant reductions in total direct variable costs (those costs that vary with the number of patients treated) over the entire period of the demonstration. These cost reductions came primarily from the nursing intensive care unit, the routine nursing unit, pharmacy, and catheter lab. Furthermore, this study found that the cost reduction increased over time.

In 2006, Geisinger Health System (GHS), a large integrated health system in Pennsylvania, implemented ProvenCareSM, which included a bundled payment system for all non–emergency CABG procedures. The price of the bundle included the estimated cost of a typical hospitalization plus half of the average cost of post–acute care for the 90 day period following surgery. The bundle included the preoperative evaluation, all hospital and professional fees, all routine postdischarge care, and management of any complications occurring within 90 days of the procedure (Paulus, Davis, and Steele, 2008). In conjunction with the bundled payment, Geisinger adopted of set of best practices to reduce complications among CABG patients and a "patient compact" that was designed to engage patients as participants in ensuring favorable outcomes. A small study (Casale et al., 2007) that compared the 117 patients in the intervention with 137 patients from a year prior to the implementation found that hospital costs dropped by 5 percent.

Geisinger has expanded the bundled payment system for hip replacement surgery and is considering using it for cataract surgery, knee replacement, low back pain, prenatal care, and bariatric surgery (Miller, 2008). We do not know how a bundled payment system would influence spending for other types of procedures or for conditions with less tangible beginning and end points for services.

Savings will depend on the design of the payment system, the particular services that are bundled, and the performance of the participating system before implementation.

Bundled payments for an episode of care are typically based on average costs for that episode with a planned reduction in payments for increased efficiency and reduced complications. To ensure that the payment amounts are fair, some adjustment for patient severity must be included in the calculations. If bundled payments do not take severity of disease into account, providers may not want to care for sicker patients because of the risk of financial losses. How well this aspect of the bundled payment system is designed will affect potential savings.

A bundled payment system may be easier to implement for some conditions than for others. For example, it can be easier to define an episode of care for a CABG or a hip replacement than for an exacerbation of a chronic condition such as diabetes. Conditions or procedures with clear begin and end dates may be more feasible for a bundled payment system. Thus, the number of procedures or conditions that can be included in bundled payment will affect the potential to reduce spending.

The limited prior experience with bundled payment was conducted with hospitals and an integrated delivery system that were at baseline performing as better than average institutions. If the bundled payment system is initially voluntary, it is likely that high performing systems will be the first to sign up and they may have less room for improvement (and thus less potential for reducing spending) than systems that do not volunteer.

This particular type of payment system is so new that the possible unintended consequences of implementing the policy are based more on generalizing from other experiences than from direct observations of bundled payment experiences. For example, providers may try to shift care beyond the post-acute period to increase reimbursement (similar to discharging patients more quickly from hospitals after the implementation of Diagnosis Related Groups [DRGs]). Adjustments for case mix severity could lead to the type of upcoding (i.e., patients coded as having more severe conditions in order to increase the reimbursement amount) seen in the DRG system, which could reduce expected savings. Providers may also try to increase the numbers of discrete bundles to maintain their income. Past bundled payment systems have not included post–acute care facilities in the bundle; including such facilities in the bundle would clearly affect decisions about when the patient was ready for transfer from the acute care hospital to another facility and how payments are distributed among providers.

Back to top

Consumer Financial Risk

Bundled payment has been shown to reduce consumer financial risk, but the evidence is limited to a single evaluation of a demonstration project:

• If bundled payment results in reduced service utilization and costs, the savings are likely to be shared by consumers. Read more below

If bundled payment results in reduced service utilization and costs, the savings are likely to be shared by consumers.

If a bundled payment approach for the reimbursement of multiple providers over an episode of care reduces the costs of care, we would expect that a portion of these cost savings might be passed on to the consumers. The savings will depend on the cost sharing features of a given health plan. For example, if patients were paying a fixed proportion of the medical bill, which in total is less, the cost to patients would also be less. If bundled payment is applied to a subset of procedures and conditions, only patients requiring those services would be likely to experience savings.

The effect of this mechanism will vary with the cost sharing features of the health plan. Patients without any cost sharing, as with Medicaid, will experience no changes in their financial risk with this policy. Patients who pay a percentage of the contracted amount, on the other hand, will pay less under a bundled payment system.

In the Medicare Participating Heart Bypass Center demonstration, the program explicitly determined what portion of the negotiated savings relative to fee for service would have been paid by the patient through cost sharing and reduced the bundled cost sharing amount accordingly. In their study of the demonstration, Cromwell, Dayhoff, and Thoumaian (1997) found that Medicare beneficiaries saved $1.8 million dollars between 1991 and 1993. Ninety–five percent of the savings were the result of the lower negotiated payment for the bypass hospitalization, with the remainder from reductions in postdischarge utilization. For the entire period of the demonstration (1991–1996), Wynn (2001) found that consumers saved a total of $7.9 million from reduced copayments.

In the Geisinger Health System experience with bundled payment for coronary artery bypass graft (CABG) surgery, hospital costs decreased by about 5 percent and 30 day readmission rates fell from 15.5 percent to 7 percent (Casale et al., 2007). Although the report on this study does not explicitly identify patient savings, both of these results are likely to have reduced patients' cost sharing amounts.

Evidence of the effect of bundled payment on consumer financial risk is still scarce. It is not clear whether consumers will experience the savings resulting from bundled payment if the bundled payment approach in the insurance plan is not explicitly designed to pass cost savings on to the consumer.

Back to top

Waste

Based on limited evidence, bundled payment is expected to reduce waste:

• Bundled payment approaches are expected to create incentives for physicians and hospitals to improve efficiency in patient care. Read more below
• A few studies have shown that bundled payment approaches have improved efficiency. Read more below

Bundled payment approaches are expected to create incentives for physicians and hospitals to improve efficiency in patient care.

With a bundled payment approach, all services related to an episode of care, including physician services, are reimbursed with a single payment to the hospital. This creates incentives for the physicians and hospitals to work together to improve efficiency in the care of the patient. We assume that reduced costs and utilization reflect waste reduction.

Efficiency improvements might involve changes in clinical care, such as the use of generics instead of more costly name brand drugs, reductions in the frequency of laboratory or radiological studies, or changes in the length of a hospital stay. Bundled payment might also affect staffing decisions or such processes in hospital care as operating room flow. Finally, whereas current hospital care involves separate claims from multiple providers, a bundled payment approach involves only a single payment for an entire episode of care and could potentially lead to a more efficient billing process.

However, some hospitals might have an easier time cutting costs and improving efficiency in patient care than others. For example, the relationship between the hospital and its affiliated staff or the culture of collaboration within the organization could facilitate or impede the ability to make substantive changes in care patterns. In addition, bundled payment approaches may have differential effects in the context of different types of hospitalizations; complex hospitalizations could offer more opportunities for waste reduction than more straightforward ones.

A few studies show that bundled payment approaches have improved efficiency.

The Medicare Participating Heart Bypass demonstration led to a number of efficiencies and cost savings among the participating hospitals, although the efficiencies and savings were not distributed uniformly among the participating hospitals. Cromwell, Dayhoff, and Thoumaian (1997) found significant cost savings in three of the four participating hospitals, with costs declining between 2 and 23.4 percent per case, unadjusted for inflation. Based on case study interviews, most of these cost savings appeared to result from changes in physician practice and patient protocols. For example, the hospitals introduced a new 24 hour protocol for post-surgery intensive care unit (ICU) stays and introduced shorter acting anesthetic agents to allow more rapid post-operative recovery. Physicians became more interested in discharge planning and encouraging earlier patient discharge. Physicians promoted the use of generic anesthetics and contrast materials. Average length of stay also declined in all four hospitals, by 14 to 32 percent.

Liu, Subramanian, and Cromwell (2001) also found reductions in direct variable costs (those costs directly associated with the number of patients) among the participating hospitals, after controlling for patient risk factors and complications. Most of the cost savings were attributable to changes in nursing, in the catheter lab, and in the pharmacy. For example, participating hospitals developed critical care pathways to improve patient flow. One source of much of the pharmacy savings in two of the hospitals came from using a less expensive contrast material in the catheter lab.

In 2006, Geisinger Health System (GHS), a large integrated health system in Pennsylvania, implemented ProvenCare, which included a bundled payment system for all non-emergency coronary artery bypass graft (CABG) procedures. The price of the bundle included the estimated cost of a typical hospitalization, plus half of the average cost of post-acute care for the 90 day period following surgery. The bundle included the preoperative evaluation, all hospital and professional fees, all routine postdischarge care, and management of any complications occurring within 90 days of the procedure. Casale et al. (2007) found that the average length of stay decreased by 16 percent.

All the evidence of improvement in efficiency and waste reduction with bundled payments comes from studies involving CABG surgeries; we do not know how this payment approach would affect the care of other types of clinical conditions. In addition, only a limited number of hospitals were involved in the studies, so we do not know whether the results can be generalized to other institutions.

Back to top

Reliability

Bundled payments to multiple providers are designed to provide incentives to improve the reliability of care, but we know of only one study that demonstrates this relationship:

• Bundled payment provides incentives that could lead providers to examine care processes to improve reliability. Read more below
• Bundled payment approaches are often proposed in tandem with other mechanisms, such as pay for performance, that aim to improve reliability. Read more below

Bundled payment provides incentives for physicians and other health care professionals to improve reliability.

In theory, bundled payment approaches among multiple providers would be expected to increase reliability through a focus on key processes that improve outcomes, reductions in complications, and improved coordination among those providers. Such improvements in reliability would stem from the fact that all the providers involved in the bundle are working under the same financial incentive to treat the patient in the most efficient and effective manner, so as to maximize profits.

Alternatively, there is some concern that bundled payments might lead providers to provide inappropriately low levels of service. This behavior may be minimized by implementing quality of care standards along with the bundled payment approach. In addition, because complications or readmissions would be covered under the original bundled payment, there would be further disincentives to avoid under-provision of needed care.

Bundled payment approaches are often proposed in tandem with other mechanisms, such as pay for performance, that aim to improve reliability.

In Geisinger Health System's ProvenCareSM program, a bundled payment approach for reimbursement of coronary artery bypass graft (CABG) procedures was combined with the implementation of a set of 40 best practices. To improve clinical outcomes and reduce costs, doctors used national guidelines in the care of CABG patients to create checklists of best practices associated with the procedure. Staff had to use these checklists before, during, and after care to ensure that all of the best practice guidelines were being followed. At the beginning of the program, adherence to best practices was 59 percent; after the program began, performance rose steadily, reaching 100 percent at three months but then fell to 86 percent. Performance once again reached 100 percent at six months and stayed there for the remaining six months of the evaluation period (Casale et al., 2007).

Studies of the Medicare Participating Heart Bypass Demonstration do not specifically examine the reliability of care. However, the hospitals included in the demonstration were chosen, in part, because of their high quality care (Wynn, 2001).

Back to top

Patient Experience

We know of no studies that have assessed how bundled payment approaches affect patient experience:

• Bundled payment to multiple providers could potentially improve patient experience through improved outcomes, reduced complications, improved care coordination and simplification of billing procedures for patients. On the other hand, patient experience could decline if doctors have less time to spend with patients or if patients have fewer choices of physicians under a bundled payment approach. Read more below

Bundled payment to multiple providers could potentially improve patient experience through improved care coordination and simplification of billing procedures for patients. On the other hand, patient experience could decline if doctors have less time to spend with patients or if patients have fewer choices of physicians under a bundled payment approach.

There are only a few examples of bundled payment approaches, and the effect of such approaches on patient experience is unknown. Neither the Medicare Cardiac Bypass Center demonstration nor the Geisinger Health System ProvenCare program specifically reported information on patient experience. Geisinger's program did include a "patient compact," an agreement between Geisinger and the patient that stressed the importance of communication between patients and providers and encouraged patients to be active participants in their own care (Casale et al., 2007).

Because a bundled payment approach is primarily an issue of how care is reimbursed, we expect that the effect on patient experience would be secondary. However, the effects on patient experience may be mixed because of competing incentives.

On the one hand, in an attempt to contain costs, providers might be motivated to coordinate the services rendered to patients across providers and across the course of care in order to avoid unnecessary duplication. Better coordination on the part of the providers would most likely lead to better patient experiences. In addition, bundled payment might improve patient experiences because the costs of care would be established upfront, making copayments predictable and simplified.

On the other hand, depending on the level of provider participation, patients might not have as much choice of physicians under a bundled system of care, and there may be pressure on physicians to spend less time with patients or on hospitals to decrease amenities in an effort to cut costs. Both possibilities could result in negative patient experiences.

Back to top

Health

Evidence is mixed regarding the effect of bundled payment approaches on health:

• In theory, bundled payments to multiple providers surrounding a hospital episode of care should encourage care coordination and lead to improved health outcomes. Read more below
• The examples of this type of bundled payment approach are limited, and the health outcomes are mixed. Read more below

In theory, bundled payments to multiple providers surrounding a hospital episode of care should encourage care coordination and lead to improved health outcomes.

The effect of bundled payment approaches on health would likely be related to any changes in the reliability of care that result from this policy option. Under a bundled payment system, hospitals and providers are reimbursed with a single payment for the entire episode of care. They thus share in any financial gains or losses associated with the care of the patient. With this payment system, they would have an incentive to reduce complications from the surgery or hospitalization. As a result, providers might be more likely to adhere to best practices, be more judicious in their prescribing practices to decrease use of drugs or procedures that have potentially adverse effects, and take steps to reduce avoidable readmissions, all of which would affect the profit potential from the bundled payment.

However, there is concern that, under a bundled payment approach, providers would cut back on appropriate care as well as on unnecessary care. Quality of care measures could be added to the bundled payment mechanism to minimize this possibility; however, some areas of clinical care do not have well established national guidelines and developing appropriate quality metrics in these areas may be more difficult.

However, there is concern that, under a bundled payment approach, providers would cut back on appropriate care as well as on unnecessary care. Quality of care measures could be added to the bundled payment mechanism to minimize this possibility; however, some areas of clinical care do not have well established national guidelines, and developing appropriate quality metrics in these areas may be more difficult.

The examples of this type of bundled payment approach are limited, and the health outcomes are mixed.

Most of the literature on bundled payment comes from the Health Care Financing Administration's (HCFA) Medicare Participating Heart Bypass Center Demonstration, which ran from 1991 to 1996. Four hospitals were chosen for the demonstration, and each hospital was paid a single fee for all inpatient and physician services for heart bypass patients (for CABG surgeries). In a study evaluating the program, Cromwell, Dayhoff, and Thoumaian (1997) found that, after controlling for patient characteristics, heart bypass patients in the program had a one–half percentage point lower mortality than the national average. This statistic did not change over the duration of the program; in fact, it improved in one of the hospitals. One caveat for this result is that the hospitals were chosen partly on the basis of their initial high quality care, so the results may not generalize to all hospitals.

In 2006, Geisinger Health System (GHS), a large integrated health system in Pennsylvania, implemented ProvenCareSM, which included a bundled payment system for all non-emergency CABG procedures. In conjunction with the bundled payment, Geisinger adopted of set of best practices to reduce complications among CABG patients and a "patient compact" to ensure patient compliance with care recommendations. A small study (Casale et al., 2007) that compared the 117 patients in the intervention with 137 patients from a year prior to the implementation found that adverse events declined in the intervention group more than in the baseline group, but that the difference was generally not statistically significant, likely because of small sample sizes. However, the intervention group was significantly more likely to be discharged to home. Of note, the patients in the study were admitted for elective CABG surgery and so tended to be healthier than the general population of patients undergoing the procedure. In addition, the rates of adverse events at Geisinger were already lower than the national average. These two factors limit the ability to generalize the results.

Back to top

Coverage

Not applicable

Back to top

Capacity

Not applicable

Back to top

Operational Feasibility

Implementing bundled payment approaches would require fundamental changes in the way that health care providers bill and are paid for services:

• The implementation of bundled payment approaches would require a broad set of operational decisions. Read more below
• Significant challenges would be related to the determination and distribution of payments. Read more below
• The Medicare Payment Advisory Commission (MedPAC) has proposed a strategy for implementation of a bundled payment approach by Medicare. Read more below

The implementation of bundled payment approaches would require a broad set of operational decisions.

Bundled payment approaches pose significant operational challenges. First, the policymaker must decide what types of episodes of care are most appropriate for bundling. For example, the duration of an episode of care for a coronary artery bypass graft (CABG) or other surgical procedure is fairly straightforward to determine: A CABG has a fairly clear beginning and end, whereas the length of an episode for a chronic disease such as diabetes is arbitrary. Thus, this type of payment system may be easier to implement for some conditions and procedures than others.

Although bundled payment provides an incentive to reduce complications, the payment system must account for differences in the illness severity of different patient populations. Thus, the adequacy of coding systems and case mix adjustment methodologies will pose operational challenges for many conditions and procedures. In the absence of adequate case mix adjustment, providers may not want to care for the sickest patients for fear of being financially liable for their inherently more expensive care. On the other hand, if the bundled payment amount is significantly higher for patients who are sicker or more complex, providers may try to code patients as being sicker. Mechanisms to monitor such practices will likely have to be made part of a new payment system.

Once the particular conditions to be covered by bundled payment are determined, the specific services to be included in the bundle must be evaluated. For the Medicare Participating Heart Bypass Center demonstration, the physician services of the thoracic surgeon, cardiologist, anesthesiologist, and radiologist were included in the bundles of all the hospitals, whereas the inclusion of other specialists was determined by the individual hospitals (Cromwell, Dayhoff, and Thoumaian, 1997). Some physicians may be unwilling to enter into a bundled payment system, especially those who provide a specialized, one time service.

Further, decisions would have to be made about which services to include in the bundle. For example, pharmaceuticals can be very costly for certain conditions. Also, past bundled payment programs have not included care in post-acute care facilities, such as nursing homes, in the bundle. The decision to include such care would complicate decisions about when the patient was ready for transfer to another type of facility.

Significant challenges would be related to the determination and distribution of payments.

One challenge involves the stability of health insurance coverage. Whereas the Medicare population is stable in its source of insurance coverage, private insurance can have high turnover rates. If a patient is treated under a bundled payment system and changes insurance during the episode of care, it can be difficult to apportion the payment correctly. If a patient moves to a different geographic area during the period covered by the bundle, there would have to be a mechanism to ensure the fair distribution of payments.

Bundled payment approaches may be significantly easier to implement in integrated delivery systems where contractual and collaborative relationships already exist between providers and facilities. In the absence of such preexisting arrangements, significant administrative structural changes would have to occur within participating entities. For example, hospitals are not typically in the position of having to reimburse physicians for services. If hospitals received the bundled payment and had to distribute it, they would have to set up contracting, billing, and reimbursement systems, all of which could require an expansion of current technologic capabilities. Further, mechanisms would have to be in place to guarantee that physicians were not being double paid (via the bundle and separate billing) for their services and that physicians seeing a patient for a problem unrelated to the bundled diagnosis were still paid appropriately. Determining rates of reimbursement would also be difficult. It can be technically challenging to match the appropriate rate with the current resource use at a given hospital. The bundle amount would have to reflect the costs of treatment per episode per patient and evolve to match the changing costs of providing treatment over time (MedPAC, 2008).

The Medicare Payment Advisory Commission (MedPAC) has proposed a strategy for implementation of a bundled payment approach by Medicare.

MedPAC recommends a general strategy for implementing a bundled payment approach. First, providers must be informed of their current resource use and rates of readmission surrounding a hospitalization and shown how these statistics compare with those of their peers. Second, payment changes should be implemented to encourage providers to collaborate on and coordinate care. Such changes might involve reduced reimbursement for hospitals with high readmission rates or mechanisms to reduce reimbursement to physicians who have high rates of referrals or who stint on care (i.e., provide inappropriately low levels of service). There could also be exploration of "virtual bundling," in which the payer (e.g., Centers for Medicare & Medicaid Services [CMS]) would calculate and distribute any gains or losses without actual agreements among providers. Third, a pilot program could be used to test a bundled payment system. In the pilot, providers could choose to receive a single payment for all Medicare Part A and Part B services surrounding a hospitalization (including a post-acute period of 30 days) (MedPAC, 2008).

Thus, although the bundled payment approach would be difficult to implement, the experiences of both Medicare and Geisinger Health System show that this type of approach is feasible.

Back to top