NOVEMBER 23, 2015
Bundled Payments for Care Improvement Initiative
The Centers for Medicare and Medicaid Services is testing how to pay providers for episodes of care instead of for individual services.
|What's the issue?|
The Affordable Care Act (ACA) gave the Centers for Medicare and Medicaid Services (CMS) broad authority to test out new payment models that have the potential to reduce Medicare spending, as long as those models preserve or enhance the quality of care provided to beneficiaries. Sylvia Mathews Burwell, secretary for health and human services, has committed to tying 50 percent of Medicare payments to these new alternative payment models by the end of 2018.
The Center for Medicare and Medicaid Innovation (CMMI) within CMS is testing a variety of new approaches, including paying providers for episodes of care instead of for each service provided. The Bundled Payments for Care Improvement (BPCI) initiative tests four different models based on episodes of care that involve an inpatient hospital stay. One model focuses on care provided during the hospital stay, while the other three models include postacute care provided once the patient is released from the hospital. CMS hopes that by paying for related care as part of a broad payment bundle, different providers that treat a patient during a single episode will have incentives to better coordinate care, avoid unnecessary services, and improve patient health. Most of the savings are expected to come from reducing spending on physician services during the hospital stay and efficiencies in providing postacute care.
This brief describes the different models being tested and CMS's experience with the project to date.
|What's the background?|
Medicare pays for services under original or fee-for-service Medicare, using distinct and separate payment systems for each type of service (physician, hospital inpatient, hospital outpatient, skilled nursing facility, and so forth). Some systems make a single payment for a bundle of services furnished by a provider to a patient, such as the lump sum payment made under the inpatient prospective payment system (IPPS) for hospital services provided to a Medicare beneficiary during an inpatient hospital stay. Other systems, such as the physician fee schedule, pay individually for each service.
These distinct payment systems often lead to wide variation in reimbursement. Because the payment systems use different methodologies and data to set payment rates, Medicare may pay one type of provider more than another for furnishing the same service.
In addition, none of these systems give providers a strong incentive to pay attention to related care the patient receives in other settings. For example, a beneficiary who breaks her hip might be treated in a hospital that is paid under the IPPS by a physician who is paid under the physician fee schedule. She may then be transferred to a skilled nursing facility for postacute care and rehabilitation services that are paid under the skilled nursing facility prospective payment system. None of the individual providers caring for the patient have an incentive to coordinate care across these different settings or seek the most efficient care setting based on the patient's needs.
CMS believes that waste, overtreatment, and lack of care coordination account for . A contributing factor to this inefficiency is payment structures that pay for services individually. Since a provider will receive additional revenue if additional services are furnished regardless of the marginal impact those services may have on the health of the patient or whether they could be furnished by a lower-cost provider, the provider has little financial incentive to consider the added value to the patient and the payer of those services.
To encourage providers to furnish care in a more efficient and coordinated manner, CMS is expanding the number and scope of services bundled together for payment. For example, under the end-stage renal disease prospective payment system, CMS expanded the bundle of items and services that are considered part of the base dialysis payment rate and stopped paying separately for certain lab tests and drugs from that bundled rate. Under the outpatient prospective payment system, CMS packaged the cost of items considered integral to performing a procedure, such as drugs used in diagnostic procedures, into the payment for that procedure.
CMS is also experimenting with payment bundles that include care furnished by different providers during a single episode of care. Previous demonstration projects have tested episode-based payments for specific conditions such as coronary artery bypass graft (the Medicare Participating Heart Bypass Center Demonstration) and cardiac and orthopedic procedures (the Medicare Acute Care Episode Demonstration). Prior projects have also focused on the changing relationship between providers. Gainsharing demonstration projects looked at the impact on Medicare spending and quality of care if a hospital is able to share with physicians some of the savings the hospital achieves from improvements in efficient care delivery. Gainsharing helps offset the forgone revenue physicians would otherwise receive from providing additional services.
Section 3023 of the ACA authorizes CMS to conduct a national pilot program on payment bundling following a specific structure. CMS has not implemented that provision and has instead used the broad authority given to the CMMI to test a variety of bundled payment models under the BPCI initiative.
|What's the debate?|
The ultimate aim of the BPCI initiative is to not just change how Medicare pays for services but to prompt changes in how health care is delivered in the United States. Medicare's participation is critical to this goal: Other payers are also experimenting with bundled payments, but as a national program with more than forty-five million beneficiaries, is a primary driving force behind change in the health care arena. Therefore, the design and structure of the BPCI initiative are particularly important as they will likely influence how other payers approach episode-based payments.
CMS has identified multiple goals for the BPCI initiative including the following: aligning Medicare payment with how patients experience care, supporting and encouraging providers interested in redesigning care, and providing as much flexibility as possible for providers to redesign care to meet the needs of their specific community.
Providers are participating in the BPCI initiative in a variety of ways, starting with their role. Providers had the choice of participating in the BPCI initiative individually or as part of a larger group organized by a "convener," such as a hospital association or health system. Individual providers or conveners can be "awardees," which means they take on financial risk for participating in the BPCI initiative. Providers that do not want to bear financial risk participate as "episode initiators" for an awardee with which they are associated. Care provided by an episode initiator identifies which Medicare patients are part of the BPCI initiative; for example, if a hospital is an episode initiator, the cost of caring for any patient admitted to that hospital for one of the selected conditions is included in the BPCI initiative.
The BPCI initiative tests four different approaches (see Exhibit 1). Each of the four models involves a different type of care that may be provided to Medicare beneficiaries who have been hospitalized. Models 1 and 4 focus on care provided during an inpatient hospital stay, while models 2 and 3 include a period of postacute care, the length of which is determined by the participant. In model 1, all Medicare patients admitted to a participating hospital are included. In models 2, 3, and 4, the participant can choose to include patients with only certain diagnoses from a list of forty-eight episodes developed by CMS. The eligible episodes include heart attacks, diabetes, and major joint replacement surgeries such as hip and knee replacements.
CMS sought participants in the BPCI initiative through multiple open periods for each model. Enrollment in model 1 began in April 2013 and in models 2, 3, and 4 in October 2014. All four models are ongoing.
The payment methodology varies by model, but each model assumes a level of savings for the Medicare program. Participants keep any additional savings. Participants are allowed to enter into gainsharing arrangements with physicians to improve care efficiency and lower costs.
Under model 1, hospitals are paid a discounted rate based on the IPPS payment amount. Others who care for the patient during the inpatient stay, such as physicians, are paid the standard Medicare rates under the physician fee schedule. Model 1 hospitals are allowed to enter into gainsharing arrangements with certain physicians (called enrolled practitioners) who care for their patients.
Under models 2 and 3, providers including hospitals, skilled nursing facilities, home health agencies, and physicians are paid the standard Medicare rates during the episode. Spending over the course of the episode is compared to a target amount that is based on historical claims data for similar episodes for episode initiators associated with the awardee. Depending on the risk tracks (these tracks protect participants from unusually high-cost episodes) selected by the awardee, the cases with extreme costs (both high and low) compared to the national average are excluded from the target calculation. Each awardee chooses a risk track for each episode it decides to include. A discount is applied to the target amount, which can vary depending on the length of the episode.
Under model 4, providers are not paid standard Medicare rates. Instead, the participating hospital is paid a prospectively set bundled rate for both the hospital services that would be paid under the IPPS and the physician services provided during the hospital stay that would be paid under the physician fee schedule. The bundled rate is developed based on historic claims data for each hospital participating as an episode initiator. In addition to paying for the care provided during a hospital stay under the BPCI initiative, model 4 participants are also responsible for any readmissions that occur within thirty days of discharge from an episode included in the BPCI initiative.
CMS waived certain statutory and regulatory provisions under the BPCI initiative to give participants the ability to provide services or establish relationships that would otherwise be prohibited by the Medicare program. For example, CMS waived certain fraud and abuse provisions to allow BPCI initiative participants to engage in gainsharing, as long as the gainsharing supports care redesign, is voluntary, and is tied to preserving or enhancing the quality of care provided.
CMS also waived additional provisions to allow greater flexibility to participants in models 2 and 3 to provide postacute care. At the request of awardees in model 2, CMS waived the statutory requirement that a Medicare beneficiary must have been in the hospital three days or more for Medicare to cover a skilled nursing facility stay following discharge. However, CMS allowed this waiver only if the majority of skilled nursing facilities with which the awardee is partnered have a quality rating of three stars or higher on the CMS 5-Star Quality Rating System.
The Medicare statute also puts limits on coverage of telehealth services, including that the beneficiary receiving those services be located at a site in a rural area. CMS waived the geographic location requirement for telehealth services provided under models 2 and 3, allowing participants to use telehealth services to treat more patients. CMS also waived supervision requirements that limit coverage for home visits to those provided while the physician is physically present (called direct supervision). CMS will allow licensed clinical staff under the general supervision of a physician (but not physically present) to furnish one home visit during a thirty-day episode, up to two visits during a sixty-day episode, and up to three visits during a ninety-day episode.
Quality Measurement and Cost Shifting
One significant concern with bundled payment arrangements is that providers may achieve savings not just by forgoing wasteful or redundant care but by not furnishing needed care or choosing the cheapest alternative, even when other care options are more appropriate for the patient. CMS is monitoring the quality of care provided by participants through an array of quality measures focusing on case-mix, use and efficiency, clinical quality processes and outcomes, and care experience.
will try to identify unintended consequences of the initiative such as increased admissions of relatively healthy patients who are likely to need less complex care (cherry picking) or avoiding treatment of high-cost patients (lemon dropping). CMS will continue to monitor care and Medicare spending for thirty days after each episode included in the BPCI initiative has ended to determine whether providers are shifting care to the post-episode period; if spending in that period exceeds a threshold amount, awardees will have to return 100 percent of the excess amount to CMS.
Concerns with the BPCI Initiative
Overall, CMS has given potential participants considerable flexibility to design a bundled payment model that would work best for their community and providers. However, stakeholders have concerns with some elements of the design.
CMS has implemented models 2, 3, and 4 in phases. (CMS did not phase in participation in model 1.) Phase 1 is a preparation phase in which participants receive data and determine whether to proceed to phase 2, the risk-bearing phase. Once a participant enters phase 2, the participant becomes liable to CMS for costs that exceed the target. Participants have criticized CMS's approach to setting spending targets, which has allowed target prices to fluctuate dramatically over the course of the project. Analysis conducted during phase 1 has not necessarily been predictive of experience under phase 2, in part because of a complicated and, , flawed methodology for establishing cost trends and setting prices under the BPCI initiative.
In addition, the BPCI initiative models do not include an adjustment for differences in patient health status from patients treated in historic data periods to those being paid under the BPCI initiative episodes. Costs of care provided during the historic periods are used to set the targets to which the costs of care provided under the BPCI initiative are compared. Lack of risk adjustment is most problematic for episodes with only a small number of patients in which one or two high- or low-cost cases may dramatically affect the average cost.
Under models 2 and 3, postacute care providers are concerned that patients are selected based on their assignment to a particular Medicare severity diagnosis-related group (MS-DRG) under the IPPS. MS-DRGs are groups of patients who have similar clinical condition and cost of hospital inpatient care. The postacute needs of patients assigned to the same inpatient MS-DRG can vary significantly, however. Therefore, the targets set under model 3, which includes only postacute care, may not accurately reflect variation in patient care costs.
Observers worry about unintended consequences of bundled payments, such as avoiding or delaying needed care to keep down costs. also suggest that instead of reducing Medicare spending overall, the project could increase spending by freeing up health care resources that will be used to provide care that is paid outside of the bundle.
The ultimate goal of the BPCI initiative is to identify payment models that could be expanded to the larger Medicare program to encourage changes in care delivery. However, the extreme flexibility given participants in the BPCI initiative to choose the cases and care for which they are accountable has raised questions about whether findings from the project can be generalized to other groups.
Critics have also questioned how much the BPCI initiative is actually incentivizing new delivery patterns versus allowing large provider networks that are providing the full spectrum of care to benefit from their already integrated delivery systems. Restrictions on use of waivers, such as limiting waiver of the three-day hospital stay for covered skilled nursing facility care to transfers to skilled nursing facilities that meet minimum quality standards, may limit participation in the BPCI initiative in areas where the provider community does not meet those standards.
Despite these concerns, CMS has seen a healthy interest in participating in the BPCI initiative. More than 1,500 providers are participating as episode initiators, and more than 300 organizations have agreed to take on the financial risk of participation. Exhibit 2 identifies the number and type of participants in each model as of October 1, 2015.
Participation in the BPCI initiative has not been constant. The number of participants in model 2 as of October 1 was 9 percent less than the number in model 2 as of July 15, 2015, and as of the beginning of October, there were 30 percent fewer participants in model 3. Most of the model 3 departures were physician group practices. Thirteen of the original twenty-four awardees in model 1 had terminated their agreements by July 15, 2015.
Most participants in models 2 and 3 chose ninety-day postacute periods and included all forty-eight episodes. Participants in model 4 were more selective in choosing which episodes to include, with only two hospitals including all forty-eight. Most model 4 hospitals included between ten and twelve episodes, and the most commonly selected episodes were coronary artery bypass graft and major joint replacement of the lower extremity.
CMS has released its describing the experience of the earliest participants in the BPCI initiative. Because of the small number of participants and limited data, the authors primarily identify issues for further evaluation.
Both CMS and providers are continuing to gain experience with and learn from the BPCI initiative. All four models have projects underway that will continue into 2016.
Congress gave CMS the authority to expand use of a tested model without requiring a change in law if CMS finds that the model reduces spending without reducing the quality of care or improves the quality of care without increasing spending. CMS has not proposed to expand the use of any BPCI initiative model at this time but could do so in the future.
However, CMS is proposing to expand testing of concepts used in the BPCI initiative on a non-voluntary basis. Under its comprehensive care for joint replacement (CCJR) model, hospitals in the sixty-seven metropolitan areas selected for the model will receive a bundled payment for the inpatient stay and ninety days of postacute care.
While the structure of the CCJR model is similar to model 2, there are significant differences. Hospitals in selected areas will be required to participate in the CCJR model (unless the hospital is already participating in certain tracks of the BPCI initiative), and hospitals in other areas will not be able to participate. All lower extremity joint replacement cases at hospitals within those geographic areas will receive the bundled payment; under the BPCI initiative, cases could be limited to those from specific episode initiators, such as only those physician group practices with which the hospital has a relationship. CMS will implement this new payment model on April 1, 2016.
CMS is also testing bundled payments for care that does not include an inpatient hospital stay. Under its oncology care model, CMS is creating bundled payments for six-month episodes that start with the first claim for chemotherapy. This model is expected to start in spring 2016.
|About Health Policy Briefs||
Editorial review by
Health Policy Briefs are produced under a partnership of Health Affairs and the Robert Wood Johnson Foundation.
Cite as: "Health Policy Brief: Bundled Payments for Care Improvement Initiative," Health Affairs, November 23, 2015.
Sign up for free policy briefs at: www.healthaffairs.org/healthpolicybriefs
- From The Editor-In-Chief
- 2014 Behavioral Health Spending
- Veterans' Mental Health Service Use
- Mental Health Care Disparities
- Quality Measurement For Behavioral Health
- Mental Health Parity Emerges
- Drug Monitoring and Opioid Prescribing
- Gun Violence, Mental Illness, Gun Laws
- Suicide Prevention As A New Priority
- The Recovery Movement
- Behavioral Health Information Technology
- Media Coverage of US Mental Illness
- Narrative Matters: Refugee Children
- View Table of Contents »
- Are The FDA’s New Definitions And Labeling Requirements Good For Us, Or Just Empty Calories? 24 Jun 2016
- What Is The Private Sector Doing To Help Fight Zika? 24 Jun 2016
- Health Spending Growth: Still Facing A Triangle Of Painful Choices 23 Jun 2016
- Four Ways To Address The Ethical Tensions Around Expedited Approval Of New Prescription Drugs 23 Jun 2016
- Prescription Drug Monitoring Programs Help Reduce Opioid-Related Death Rates 22 Jun 2016
Now Playing Global Health: Patient Safety, Cancer Care, Universal Health Coverage and InnovationBriefing June 14, 2016
- The Future Of The Individual And Small Group Markets May 18, 2016
- Asthma And Injustice On Chicago's Southeast Side May 02, 2016
- Patients' And Consumers' Use Of Evidence April 07, 2016
- Variation In Health Outcomes: The Role Of Spending On Social Services, Public Health, And Health Care, 2000-09
- A More Detailed Understanding Of Factors Associated With Hospital Profitability
- Steady Increase In Prices For Oral Anticancer Drugs After Market Launch Suggests A Lack Of Competitive Pressure
- The Triple Aim: Care, Health, And Cost
- Cancer Drugs Provide Positive Value In Nine Countries, But The United States Lags In Health Gains Per Dollar Spent
- Patients' And Consumers' Use Of Evidence
- Understanding An Informed Public's Views On The Role Of Evidence In Making Health Care Decisions
- Incorporating Patient-Reported Outcomes Into Health Care To Engage Patients And Enhance Care
- Yelp Reviews Of Hospital Care Can Supplement And Inform Traditional Surveys Of The Patient Experience Of Care
- The Patient Engagement Imperative