Characteristics of Providers

There is considerable overlap in the types of services provided in the various PAC sites.⁴ In general, LTCHs are the most expensive setting and provide care to the most clinically complex patients. IRFs generally treat patients who need more rehabilitation than patients who are admitted to a SNF, and the IRF patients must be healthy enough to participate in intensive therapy. In aggregate, SNFs provide fewer hours of therapy than IRFs and have a less skilled staffing configuration, with less physician involvement and lower nurse-to-patient ratios.⁵ HHAs provide rehabilitation or skilled nursing care in the homes of patients who are unable to leave without assistance. Over half of home health episodes are not initiated after a hospital stay. HHA payments tend to be the lowest among the providers, reflecting their less extensive services.

Although the Medicare program has different coverage criteria, facility requirements, and payment methods for each post-acute care setting (Figure 3), policymakers remain concerned about the lack of clear distinctions in services and expertise across providers, particularly given the significant differences in payment amounts.

FIGURE 3

Medicare Post-Acute Care Coverage Requirements: Patients, Facilities, and Services.

According to the Medicare Payment Advisory Commission (Med-PAC) “PAC [post-acute care] settings lack clear boundaries around the services furnished and the types of patients treated…. Yet, the setting where a patient is treated has very different cost implications for the program (and for the beneficiary, through the copayments).”⁶

Use of Post-Acute Care

Post-acute care use varies widely across individuals with similar conditions and can follow numerous patterns. The choice of post-acute setting may be based on ownership or contracting relationships of the discharging hospital. Further, LTCHs and IRFs are not located in all geographic areas, so they are not options for a large number of patients. Decisions may also hinge on the patient’s living situation. A patient who does not have support in the home, for example, may not be able to be discharged from the hospital with home health care.

For some patients, care may proceed from a more intensive to a less intensive site, for example from an IRF to home health care, as treatment progresses. Other patients may be discharged from a hospital to a SNF for recuperative care and then proceed to an IRF for more rehabilitation. Many post-acute care episodes are punctuated by a return to the acute care hospital, which may be one of the most expensive components of an episode.

Based on an examination of post-acute care use conducted for the Assistant Secretary for Planning and Evaluation (ASPE), in 2006, over 14 percent of Medicare fee-for-service (FFS) beneficiaries had an index hospitalization, which was defined as one following 60 days without an inpatient stay or post-acute care services.⁷ Of these beneficiaries, just over 35 percent were discharged to post-acute care for further treatment. Most beneficiaries received their post-acute care in SNFs (41.4 percent) or through home health (37.4 percent); the rest were treated in an IRF (10.3 percent) or with hospital outpatient therapy (9.1 percent), and a small share received care in a LTCH (2 percent). The proportions of patients who used post-acute care varied based on the cause of the index hospitalization. The majority (87.3 percent) of patients hospitalized for a major joint procedure,⁸ for example, used post-acute services, with considerable variation in the site of care (Figure 4). Moreover, Medicare’s payments for the care of these patients differed as well. Of the patients hospitalized for stroke,⁹ 58.1 percent received post-acute care, with about one-third admitted to an IRF, another third to a SNF, and about 20 percent to home health.

Generally, the higher the severity and cost of the index hospitalization, the more expensive the post-acute service use. The mean total payment (the initial admission plus post-acute use until there was a 60-day period without service use) was $23,985 for post-acute users following a major joint procedure (DRG 544).¹⁰ Payments for beneficiaries who had no comorbid or complicating conditions (one measure of severity) averaged $17,724, compared with $35,319 for those with five or more conditions. For beneficiaries with an index hospitalization for pneumonia (DRG 089), one-third used post-acute care and the average episode payment was $20,476, but ranged from $16,096 to $23,241, depending on level of severity.

FIGURE 4

Selected Patterns of Care for Medicare Beneficiaries.

Few conclusions can be drawn from the literature about the most effective site or pattern of care, or factors that affect the type and amount of service use.¹¹ Research has been limited by the lack of consistent patient assessment measures across settings, including outcome measures upon discharge. According to MedPAC, “Ideally, a common assessment tool would gather uniform information to help providers make appropriate placement decisions and enable CMS [the Centers for Medicare & Medicaid Services] to evaluate patient outcomes within and across settings.”¹² CMS was directed by law to develop such a tool for use in all post-acute care sites. The Continuity Assessment Record and Evaluation (CARE) instrument was developed to measure clinical and functional status of patients upon hospital discharge and post-acute care admission and discharge.¹³ So far, the CARE instrument has been tested by almost 200 providers; an evaluation of the data collected is scheduled for release in 2011.

Appropriate Site

There are no widely accepted standards for choice of post-acute care, and there is little systematic evidence to indicate which site of post-acute care or which services are clinically indicated for a patient.¹⁴ Often, the choice of setting is based on availability, proximity, and patient preferences. Although these factors need to be considered, the dearth of clinical and rehabilitation outcome information may result in suboptimal decisions by patients and providers. This information gap can leave the Medicare program vulnerable to paying too much and the beneficiary vulnerable to inappropriate or inadequate care. Because the cost of care across sites varies considerably and the services provided in each setting are the same or similar, when the situation is ambiguous, the entity receiving the bundled payment will favor the site with the lowest net expense.

Ownership or other organizational connections between providers could strongly influence the choice of post-acute care setting or particular provider. Many hospitals own post-acute care providers and that number could grow under the incentives of bundling. In 2006, close to 13 percent of patients who were discharged to a SNF went to a hospital-based provider, about half of hospital discharges to an IRF were to a hospital-based provider, and almost 21 percent of patients discharged to home health received their care from a hospital-based HHA.¹⁵ If the payment entity, for example, was a hospital that had a SNF unit, it would strongly prefer to send patients needing post-acute care to that SNF. In this way, it would keep more of the payment, have better control of costs and occupancy in both settings, minimize disruptions for the patient, and have more opportunities to ensure continuity of care. Although these incentives exist under the current payment methods, they would be strengthened under a bundled approach. Whether this is always consistent with providing quality care is not known.

The lack of widely accepted standards for post-acute care site and service needs will complicate efforts to ensure appropriate care during an episode.

Another factor that affects choice and use of post-acute care is that some settings are not widely available. The two most expensive, IRFs and LTCHs, are not located in all states or some larger regions. In 2007, IRF beds per 1,000 Medicare beneficiaries ranged from 1.82 in Arkansas and 2.09 in Louisiana, to 0.19 in Maryland and 0.30 in Oregon.¹⁶ The disparities in the availability of LTCH beds is even greater, ranging from 3.08 beds per 1,000 Medicare beneficiaries in Louisiana and 3.92 in Massachusetts to none in Alaska, Iowa, Maine, Montana, New Hampshire, Oregon, and Vermont. This wide disparity in availability of providers raises questions about the use of post-acute care for beneficiaries in areas without access to IRFs or LTCHs. It may complicate efforts to develop an appropriate bundled payment amount and evaluate the impact of bundling on spending.

Freedom of Choice

Whether by ownership or through contractual arrangements, entities accepting bundled payments will have to develop affiliations across providers to deliver all required services, manage the episode of care, and determine the distribution of the payment. These arrangements, in effect, will limit patients’ freedom of choice of providers, which has been a basic tenet of Medicare fee-for-service. Under Medicare’s managed care option (Medicare Advantage) freedom of choice is restricted, but beneficiaries voluntarily enroll in that program. How beneficiary preference or choice of provider would be considered under a bundled payment option is not clear, nor is it clear whether or how a patient could choose to participate in the pilot.

Just as Medicare beneficiaries have freedom to choose their providers, providers that do not have a contractual arrangement with the entity that receives the bundled payment may argue that they should be able to participate in the bundling pilot. They may say that the entity should take “any willing provider” that accepts the terms and conditions offered. Providers may be especially concerned about exclusive relationships developed to accept a bundled payment if they are excluded and their market share is threatened. This might particularly be the case for smaller or independent providers that are in markets where a dominant system that can provide the range of acute and post-acute services required for the episode receives the bundled payment.

Payment Entity

At a minimum, the entity that receives the bundled payment would need to be large enough to manage the risk that the payment may not cover the costs of care for some patients, to be able to arrange for all of the services required by every patient, and to apportion the payment across providers. Other requirements likely include the ability to manage care, report patient-level data, and ensure the quality of care across the entire episode. In many instances, the entity is likely to be a hospital or provider system, which may own post-acute care facilities or have other established contractual relationships with them. Acute care hospitals would be likelier than a post-acute care provider or physician practice to have the resources and infrastructure to manage this kind of endeavor. An integrated delivery system may also have advantages in accepting a bundled payment. Stricter requirements for entities with respect to these factors may provide better assurances of program success, but would also restrict the number and type of bundling options that could be evaluated.

Recognizing the difficulty of assigning an entity the responsibility of coordinating and monitoring the range of services across multiple providers and the potential of unintended consequences, MedPAC recommended that Medicare pursue bundling in an incremental manner.²² It recommended beginning with reduced payments to hospitals with high avoidable readmissions, which Medicare is currently implementing. This essentially makes the hospital accountable for the quality of the transition out of the hospital, extending its responsibility through payment incentives beyond the hospital stay. This will allow the Medicare program to assess the use of payment incentives to encourage a more collaborative approach to care transitions, and to evaluate how hospitals respond to these incentives.

Bundle Definition

Critical aspects of the bundle definition have been specified in the law, but many decisions remain that will affect the payment incentives. Although almost all Medicare-covered services are included in the definition of the bundle, some services could be covered under the bundled payment and others paid separately, depending on the need for the particular service. There could be distinctions, for example, between services for the condition that caused the hospitalization and other services that may be required by the beneficiary during the episode period. The entity could be responsible for all services necessary to treat the designated condition; other necessary services would be paid for separately. A further distinction could be made between services for chronic conditions, such as arthritis or diabetes, and acute conditions like a urinary tract infection or an injury from a fall. Among the acute conditions, some could reasonably be expected to be avoided with appropriate medical care, and others may be unavoidable.

Distinguishing between related and unrelated services may not be straightforward. When a hospitalization is for a planned event, such as a knee replacement, for example, it may be easier to make this distinction than for an unplanned admission, like for treatment of a stroke. Similarly, services associated with an acute event, such as a fall or accident, may be easier to distinguish from services associated with the episode than services related to chronic conditions such as arthritis or diabetes. With any definition, however, the entity that receives the payment has strong incentives to exclude services from the bundle and the payer, in this case Medicare, has incentives to be inclusive with services. The distinctions between the services and treatments related to the acute/post-acute episode and other conditions may have a large effect on the costs under the episode payment and the assignment of the responsibility for managing health care conditions.

The more inclusive the bundle definition, the more opportunities there would be to substitute across services to improve care and to tailor treatment to meet specific clinical or personal needs. If the entity was responsible for treating a patient’s chronic conditions as well as the condition requiring the hospitalization, it might be more likely to coordinate all of the patient’s medications or integrate therapies that would be beneficial for coincident conditions, for example. With the greater opportunities come greater risks, however.

The entity would be at greater financial risk for unexpected or high costs associated with the need for particular services. And the patient would be at greater risk of inadequate or substandard care if an entity tried to avoid providing particular services.

One option for defining the bundle for a single payment is a “building block” approach, in which episode definitions expand as more information becomes available.²³ Initially, for example, the bundle may be the hospital care and the physician services provided during the hospital stay. Medicare is currently testing this narrower definition of an episode. With more information, the bundle could be expanded to include one or more post-acute care services, particularly for conditions that have a more limited and established pattern of care.

Payment Amounts

The payment amount for the bundle of services needs to be set appropriately to reward entities that deliver an episode of care efficiently and provide incentives for improvement. If the payment is too low, participation may be limited and any entity that does participate will have stronger incentives to stint on services or avoid potentially high-cost patients. If the payment is too high, Medicare and the beneficiaries will be paying too much and the entity will not need to implement the desired efficiencies. The bundle definition and other features of the pilot will affect the appropriate level of the payment amount.

Medicare typically has used national average provider costs as a proxy for the costs of an efficient provider when it sets prospective payment rates. Given the wide variation in costs of care for patients with a similar diagnosis, this proxy, even when adjusted for patient characteristics and geographic cost differences, will significantly overpay for many episodes and severely underpay for the most expensive ones. The average Medicare payment for an episode of care for beneficiaries hospitalized with pneumonia who received post-acute care, for example, was just over $20,000 in 2006. A payment based on this average, however, would be too high for the two-thirds of those patients who did not receive post-acute care. For the most severely ill pneumonia patients, it could be 20 to 30 percent too low.²⁴

Risk-adjustment of payments to account for patient characteristics, such as complications or comorbidities, that reflect their severity and may predict their resource use is the primary method to help ensure payment amounts are appropriate. In the absence of well-developed risk-adjustment methods, other ways to guard against payments that are too high or too low include adjustments that limit the entity’s profits or losses for an individual episode or all episodes, such as risk corridors or risk sharing. Outlier payments could protect entities from unexpectedly high costs of certain individuals. These methods could mitigate the potential negative consequences of a bundled approach until more refined payment adjustments and oversight mechanisms are fully developed.

Risk and Other Adjustments—Medicare uses different patient classification methods with different underlying data requirements in hospitals and each post-acute care provider type to risk-adjust payments (Figure 5). Typically, patients are grouped on the basis of characteristics that are associated with the services they need or their expected resource use, and the payment for all patients in the same group are adjusted by the same amount. Medicare’s hospital payment system uses the Medicare severity-diagnosis related group (MS-DRG) system to group patients based on diagnosis, age, and complicating conditions.

FIGURE 5

Post-Acute Care Providers: Medicare Payment and Classification Systems.

The MS-DRG system will likely be used to identify the conditions that will be included in the pilot and to risk-adjust the bundled payment. The ASPE studies and analyses by MedPAC indicate that higher severity patients (as indicated by higher severity MS-DRGs during the inpatient stay) are likelier to use post-acute care and have higher average post-acute care spending.²⁵ Whether these distinctions are robust enough to use in adjusting a bundled payment is not clear. There is little reason to believe, however, that any of the existing risk adjustment methods used in post-acute care settings would improve the adjustment for an episode payment.²⁶ In fact, the risk adjustment systems for SNFs and HHAs in particular have been criticized because they incorporate service use as a measure of severity. This means that the payment is increased for patients who receive more therapy services, which provides incentives for these providers to deliver more services. The SNF and home health risk adjusters have driven up the provision of therapy services and the resulting payments, even though there is no evidence that the patients need or are benefiting from the extra therapy.²⁷

CMS is testing the CARE instrument in various post-acute settings to determine if it is adequate for collecting consistent patient assessment information across post-acute settings. It includes information on medical, functional, cognitive impairments, and social/environmental factors that either measure severity differences within medical conditions or predict outcomes. Consistent data across all of the settings could go a long way toward the development of adequate risk adjustment methods. Although these are the types of factors that might be needed to supplement the MS-DRGs, understanding the relationship between these factors, episode costs, and episode outcomes will require considerable analysis and lead time.

In addition to helping to ensure that payments are appropriate, categorizing patients according to their expected costs is critical for accurately comparing the performance of the entities that receive the bundled payment.²⁸ Without this type of risk adjustment, one entity may appear to have lower costs and better outcomes merely because of the mix of patients it treated. Performance comparisons are important for the entities as they set up their network of providers, to payers who want to assess whether they are receiving value for their payments, and for patients and providers as they make treatment choices. The factors used in risk adjustment for payment and for measuring performance are likely to differ and be based on different data.²⁹

Other Payment Adjustments—Other adjustments may be incorporated into the bundled payment system. Medicare makes geographic adjustments in all of its payment systems to account for area differences in wages. Medicare also adjusts payments in some situations to account for extra costs associated with operating in a rural area, for example, or an inner city area, when it wants to ensure appropriate patient access. Analyses would be needed to determine if, for example, there are systematic differences in costs between post-acute care episodes provided in rural versus urban areas, or the contribution of wage differences to episode costs. Medicare may also need to consider the maldistribution of certain post-acute care providers in making geographic adjustments to episode payments.

Quality Assessment and Oversight

Medicare collects a significant number of provider-specific performance measures at each site of care. Many of these measures, however, are of discrete services rather than the full spectrum of services across an episode, and they provide little opportunity to assess joint accountability for care delivered during an episode.³⁰ Much like the separate classification and risk adjustment systems across providers, how and whether the performance measures can be combined to adequately assess an episode is not known.³¹ One analysis of episode payment options concluded, “The separate performance measurement and payment systems for each provider type and setting are not aligned around or reflective of the continuum of care that a beneficiary receives within a given course of treatment or episode of care.”³²

Incorporating ongoing oversight mechanisms into the design of a bundled payment approach will also be critical. The vulnerability of patients to underservice or poor care increases with the size and extent of an inclusive payment approach. The larger the bundle, the greater the ability of providers to make large sums of money by stinting on care, particularly if the patient classification system does not adequately distinguish among patients based on their level of need. Without a policy design that ensures that care for certain types of patients is not more profitable than care for other types of patients, the less profitable patients may not have adequate access. Protecting from these potential problems requires continued, and possibly expanded, reporting and evaluation of service-level data across the entire episode. The total number of minutes of therapy, for example, or patient functional status may need to be collected consistently across sites to monitor the level of care patients are receiving and to correlate care with outcomes. Lack of adequate planning for data requirements and oversight would hamper efforts to refine and improve the bundled payment approach, while leaving beneficiaries vulnerable to high costs and inadequate care.