CMS Bundled Payments for Care Improvement (BPCI) Initiative: A Discussion on Bundled Payments

There is a large disconnect between what occurs after a patient is discharged from a hospital (acute) and what occurs thereafter. The quality of care is entirely unmanaged, uncontrolled, and unmonitored. Moreover, the patient is placed at greater risk whe




CMS Bundled Payments for Care Improvement (BPCI) Initiative: A Discussion on Bundled Payments


The ACA enacted a range of sweeping reform including the authorization to establish the Center for Medicare and Medicaid Innovation which has the powers to pilot initiatives to improve care delivery and reduce costs in accordance with the Triple Aim. Interestingly, the legislation (ACA) was written and passed so as to grant the Center powers to expand any model it finds effective without Congressional proceedings so as long as the Secretary of the Actuary certifies to the model’s efficacy in quality of care improvement and cost reduction. A number of Alternative Payment Methods (APMs) have been piloted—and continue to be piloted—under   these premises; one such program focuses on the post-acute care continuum of services.


Background: Setting the Foundation for Value-Based Care Driven by Fundamental Reform to Payment  

Although an in-depth discussion into the variety of Alternative Payment Methods (APMs) remains beyond the scope of this discussion, it behooves the paper to briefly introduce some of the regulatory and industry efforts currently in motion to support the implementation of APMs.


Secretary of Health and Human Services Burwell’s sweeping announcement, made in January 2015, solidified CMS’s determination to shift the industry’s direction toward a quality-centric, outcomes driven, delivery environment (Burwell, 2015). With a goal of having 90% of Medicare fee-for-service (FFS) payments based on quality metrics by 2018, with intermediate goals to transition to this level by phasing-in different reimbursement modalities, regulations/mandates, and pilot programs. Multiple reimbursement transition mechanisms were proposed by CMS, and continually developed, on an ongoing basis to reach this aim with both Medicare Part A and Part B components. Some examples include the Value-Based Payment Modifier (VBPM) program which is applicable to the Part B physician fee schedule. 

Physician- or physician group-specific# reimbursement modifications are made to reflect the extent to which the provider has delivered quality outcomes during a given performance year. VBPMs are determined based on several measures extracted from CMS’s Physician Quality Reporting System (PQRS)#, from composite auto-reported EMR data, which is also tied to Meaningful Use (MU) standards or manual reporting by provider office(s). Reimbursement to providers will either remain neutral, unchanged, or be subject to adjustments by up to +/- 9% of prevailing Medicare rates (Medicare Access and CHIP Reauthorization Act of 2015 [MACRA], 2015) using algorithms to determine the extent to which providers have aligned to value-based care and complied with reporting requirements#.  


Effective as of 2019, CMS will consolidate these three pathways—VBPM, PQRS, MU—into one umbrella system called the Merit-Based Incentive Payment System (MIPS) in an effort to establish a coherent reformation of payment based on value in Medicare FFS claims to Part B providers (MACRA, 2015). A similar model has been implemented in the Part A arena.  


In the managed care world, among many initiatives to curb costs, a similar model of reimbursement adjustment has been implement. The premise in the managed care sphere is based on aggregate risk scores and contingencies on the capitation paid to panel providers caring for high-risk patients. Some Medicare Advantage plans will adjust provider capitation rates based on a given provider’s assigned panel’s hierarchical condition category (HCC) risk scores. HCCs are designed to risk-stratify and provide a risk score for the provider, hence incentivizing more comprehensive screenings to capture 1) incentive dollars, and 2) obtain a risk-profile.  Studies (Li, Kim, & Doshi, 2010) have provided evidence for the validity of the hierarchical condition category (HCC) model’s ability to risk-adjust at a 95% confidence interval. Nonetheless, it is a less daunting task to control utilization and expenditures in a managed care environment. The greater challenge is when providers themselves become utilization managers without appropriate checks and balances in place. 


The ensuing analysis will focus on one of the Alternative Payment Models currently being piloted by CMS’s Center for Medicare & Medicaid Innovation (CMMI) to address one of the most challenging and costly segment of the Medicare FFS delivery chain which is the post-acute care (PAC) setting. There is a large disconnect between what occurs after a patient is discharged from a hospital (acute) and what occurs thereafter. The quality of care is entirely unmanaged, uncontrolled, and unmonitored. Moreover, the patient is placed at greater risk when they are subjected to the notoriously siloed and fragmented nature of the American healthcare delivery infrastructure during such a vulnerable time in their lives. Bundled Payment Care Initiative (BPCI) is being tested to address the continuum of care via episodic payment using four models using financial-risk as an incentive to align delivery and performance (Centers for Medicare & Medicaid Services, 2014). Models 1-3 are retrospective bundled payments and Model 4 is a prospective bundle payment. Briefly, Model 1 is a Retrospective Acute Care risk-only. Model 2 places the Acute Care center at-risk for all acute care services provided plus all Post-Acute (PAC) services rendered 30 or 90-days# post-discharge and readmissions within 30-days. Model 3 placed the PAC providers at risk only. Model 4 provides a prospective payment for the Acute Care stay only (which includes all services should readmissions occur within 30-days post-discharge) and the facility cannot collect any additional dollars outside of that prospective bundle. For Models 1-3 billing and reimbursement remain as usual. An explanation into the mechanics of the reconciliation processes are discussed below in greater detail with the remainder of the topic focused on Model 2. The subsequent pages will also present some of the current BPCI Model 2 implementation strategies seen in various participating markets, challenges unique to this program, and suggestions to consider adopting in order to better secure a successful piloting from an acute-care center’s perspective. 


Bundled Payment Care Initiative (BPCI)

The need to move towards value based purchasing in healthcare is all the more salient with the highest costing and the least controlled of the population subtypes and industry segment: Medicare FFS beneficiaries, post-acute care (PAC) discharge, and the series of events which occur after they are discharged from the hospital acute care setting. The fiscal rationale for this focus is based on the staggering finding that 73% of Medicare spending variation is attributed to wide regional disparities in structure, patterns, and payment to PAC settings (Scully & Mellard, 2014). Furthermore, about 20% of patients in the Medicare FFS program are readmitted within 30 days post-discharge from acute. Research supports about 75% of these readmissions as being avoidable or otherwise preventable (Jencks, Williams, & Coleman, 2009). This is an important area of focus since each readmission increases per case cost by an order of about 2.2 times (Dobson DaVanzo & Associates, 2012). Readmissions also speak to a broader quality of care culture in the PAC environment as well as the lack of incentives to reinforce quality care which can prevent unnecessary ED use and hospitalizations.

CMS rolled out the BPCI pilot on various timelines based on each of the four program models. This paper discusses the BPCI Model 2 Retrospective Acute & Post-Acute Care Episode Bundles (Centers for Medicare & Medicaid Services, 2014) which includes episodic bundle payments retrospectively reconciled with the acute care awardee at-risk for the entire episode length and expenditures per case DRG. A total of 48 DRGs have been made eligible for testing under this program; about 70% of Medicare spending is on these 48 BPCI eligible DRGs (American Hospital Association, 2016). 

Post-acute care settings (PAC) include inpatient rehabilitation facilities (IRF), long-term acute care (LTAC), skilled nursing facilities (SNF), and home health agencies (HHA or HH). By placing the hospital at financial risk 90-days post-discharge across the PAC continuum and beyond, the CMS BPCI Model 2 demonstration aims to align outcomes along the continuum of care with the payment. In model 2, reimbursement for all parties is maintained as usual–the providers bill Medicare FFS per applicable guidelines and rates. The mechanism of accountability is a retrospective reconciliation of claims which is conducted by CMS on a quarterly basis. During this process, CMS reviews claims data for the previous quarter for each Awardee/Convener and tracks financial performance on select DRGs for 90-days post-discharge from the hospital for at-risk BPCI cases during the performance period. The total cost for any given episode of care is assessed against a previously agreed upon target-price. Subsequently, case-mix severity, community experience, geographic factors, and risk-tracks (maximum gains/maximum losses) are applied and the Net Payment Reconciliation Amount (NPRA) is determined. Any positive NPRA is shared via gain-sharing dollars; and any negative NPRA is paid to the CMS by Awardee/Convener. The total shared losses/gains possible are risk-tracked at 20%, meaning the maximum possible gains that the Awardee/Convener can collect cannot exceed 20% and inversely cannot plummet below 20% losses. This is a stop-loss measure in its own respect. An added layer of complexity is the process of possible true-ups, where CMS can retrospectively adjust past reconciled NPRAs three quarters (9 months) prior (American Hospital Association, 2016). This makes not only the financial projections a very complex matter, but it also adds another moving component to an already moving element of getting a ‘working DRG’ to a ‘final DRG’ which causes patients to drop in and out of the BPCI program. 


BPCI Implementation Strategies: Acute & Post-Acute Collaboration   

Current implementation strategies seen in some markets include narrowing of PAC networks. Some acute care facilities have taken steps to implement “soft steerage” methods of discharging patients to top performing PAC providers. The patient choice letters provided to patients and caregivers during the discharge planning process include all of the surrounding SNF and HHAs available to the patient, however providers who have been vetted by hospital administration based on a number of objective criteria are bolded and listed at the top of the list. These criteria take into account factors such as each facility’s CMS’s Five-Star Quality Ratings, staffing ratios, service capabilities, readmission rates, responsiveness, and the extent to which they tend to cherry-pick/lemon-drop. There should be collaborative agreements in place to reinforce accountability by the risk-bearing party. Although the program is by definition fee-for-service, managed care-like mechanisms can be implemented to support the program objectives by building networks, offering objective data, and education to stakeholder in the marker. Soft-steerage and participating provider networks, within this context, will work only if there is a substantial volume shift that the hospital can leverage to incentivize the SNF and HHA network and bring them in alignment. The hospital should also be more than willing to hear feedback and put structures in place to support bilateral information sharing. Furthermore, the hospital billing practices need to be revisited since the DRGs dictate program eligibility and expenditures during reconciliation. For example, a urinary tract infection (UTI) and a sepsis case can look clinically similar (R. Villobos, personal communication, May 2, 2016), yet the reimbursement for these two cases are drastically different—with sepsis paying at a substantially higher rate than UTI. The billing office is more inclined to scrub UTI claims and up-code to sepsis which affects the target price negotiated with CMS and Net Payment Reconciliation Amount (NPRA) figures at the end of the quarter. 

Data and information technology also enhances the visibility of the at-risk entity beyond its walls. In addition, bundling localizes delivery. In other words, when risk is situated closer to the provision of care, the hospital can use its influence more directly with the surrounding community to move the needle on some of these metrics (i.e.: ALOS, readmissions, discharge to appropriate setting of care/transitions, etc.). 

Challenges of the BPCI bundled payment method, as briefly mentioned above, include random variation and outliner cases with small sample size. Working DRGs are moving targets. Some providers have alliances with motives and incentives that are in conflict with programmatic objectives and they might refer patients according to their personal aims. Retroactive reconciliation of payment with a large lag time in claims data tremendously delays feedback as to how the acute center is performing. This is particularly challenging for large systems, such as Dignity or HCA, with multiple facilities enrolled in the program. Therefore, other ‘informal’ or non-CMS tracking should be established to measure real-time performance.



Although BPCI is called a pilot program, it is not a fad which will disappear. In fact, effective as of April 1st, 2016 bundled payments are mandated for Comprehensive Joint Replacement (CJR) procedures in 67 markets throughout the nation (American Hospital Association, 2016). Bundled payment models are poised for wide implementation in the coming years and those providers participating voluntarily now, are strategically better aligned with the changing landscape.  Program will be successful if acute care providers figure how to manage episodes of care 90-days post-discharge. 

The first PAC setting post discharge from an anchor hospital (or acute setting) will greatly dictate total episode cost. If the patient is discharged to an IRF, then one should expect that patient to be discharged to SNF before transitioning to another setting. It is rarely the case that a patient is sent directly from an IRF to home. As discussed about, readmissions are major programmatic concerns as they drive the cost of a given episode by approximately 2.21 times per case per readmission (Dobson DaVanzo & Associates, 2012).

Solid discharge planning, closer SNF and HHA integration, visibility into the trajectory of care beyond the acute, and data integration across the continuum are key success factors.  Acute care centers should also develop and implement community-based care transitions programs (Logue & Drago, 2013) to better facilitate a smooth transition through the care pathways identified for the patient. The collaborative effort—led by the acute center—should include the local SNFs and HHA identified as being in the narrowed network, as discussed above. It is important to instill continuity of information: 1) the patient’s history should follow him/her across the continuum, and 2) the educational material provided at the acute center should be consistent with the material presented at the SNF and by the HHA.  Too often the patient is provided with volumes of educational material in various forms, lacking uniformity of information, and confusing an already frail person. 

Acute care centers should focus on messaging based on outcomes and not utilization. They should emphasize that they are interested in the measure that indicates patients are sent to the appropriate setting of care, transitioned according to evidence-based predictive tools, and safely discharged from the current setting to the next. Physician and care coordination department buy-in will also be a factor that hospital administration must factor into their implementation and evaluation strategy.

Given that the PAC setting has been identified as having the greatest variation in structure and spending, nationally, it might follow that each region must have its own strategy in addressing its outliers. There is a 25% spending discrepancy between Medicare Advantage and Medicare FFS programs on Part A and Part B benefits (Medicare Payment Advisory Commission [MedPAC], 2014), with the latter spending a quarter more than its managed counterpart. This pattern of spending is simply not sustainable and is not equitable. When adjusted for severity and demographics, in theory there should be little to no variation between the managed care patient who transitions through the PAC continuum within 14 days (per UM) and the Medicare FFS beneficiary who can stay in a skilled bed for 100 days.

The CMS BPCI Model 2 initiative poses perhaps one of the most complex VBP models in the market today by attempting to superimpose a managed care model over an existing Medicare FFS structure without any utilization control authority, changes in billing/reimbursement patterns, high dollar risks, and large time-lags in ‘official feedback’ in terms of payment reconciliation. Nonetheless, industry experts, and CMS alike, expect more bundles in the not-so-distant future. 




American Hospital Association. (2016). Medicare’s Bundled Payment Initiatives: Considerations for Providers [Issue brief]. Retrieved from

Burwell, S. M. (2015). Setting value-based payment goals — HHS efforts to improve U.S. health care. The New England Journal of Medicine, 372:897-899.

Centers for Medicare & Medicaid Services. (2014). Bundled Payments for Care Improvement Initiative Fact Sheet [Fact sheet]. Retrieved from

Dobson DaVanzo & Associates. (2012). Clinically Appropriate and Cost‐Effective Placement (CACEP): Improving Health Care Quality and Efficiency (12-106). Retrieved from

Jencks, S. F., Williams, M. V., & Coleman, E. A. (2009). Rehospitalizations among patients in the Medicare Fee-for-Service program. The New England Journal of Medicine, 360: 1418-28.

Li, P., Kim, M. M., & Doshi, J. A. (2010). Comparison of the performance of the CMSHierarchical Condition Category (CMS-HCC) risk adjuster with the charlson and elixhauser comorbidity measures in predicting mortality. BMC Health Services Research, 10(245), 1-10.

Logue, M. D., & Drago, J. (2013). Evaluation of a modified community based care transitions model to reduce costs and improve outcomes. BioMed Central Geriatrics, 13(94), 1-11.

Medicare Access and CHIP Reauthorization Act of 2015 (MACRA), H.R. 2—114th Congress (2015).

Medicare Payment Advisory Commission. (2014). MedPAC Data Book: Health Care Spending and the Medicare Program —June 2014. Washington, DC: Government Printing Office.

Scully, T., & Mellard, K. (2014, March 19). The best of reform: Postacute care bundling. The American Journal of Accountable Care, 3(14), 46-49.


[1] Modifier is attached to billing provider’s TIN


[2] PQRS reporting criteria and base metrics can change based on performance year—2016 reporting details can be found at the following link:


[3] Refer to the Medicare Access and CHIP Reauthorization Act of 2015 for exemptions; 


[4] Awardees choose either a 30- or 90-day episode risk period; for a 30-day period 3% discount is guaranteed to CMS, for a 90-day period 2% discount is guaranteed to CMS.



%d bloggers like this: