• Center on Health Equity and Access
  • Clinical
  • Health Care Cost
  • Health Care Delivery
  • Insurance
  • Policy
  • Technology
  • Value-Based Care

The Cost of Learning: Participating in APMs Despite Projected Financial Losses

The American Journal of Accountable Care®September 2019
Volume 7
Issue 3

This article explores Northwestern Medicine’s decision to participate in a Medicare alternative payment model (APM) despite projected losses.


When determining whether to participate in a voluntary alternative payment model (APM), many healthcare organizations consider financial performance to be the leading decision factor. That is, is the provider estimated to make money or lose money if it participates in the program? In this article, Northwestern Medicine (NM) discusses the alternate criteria that influenced its decision to enter a Medicare APM, Bundled Payments for Care Improvement (BPCI) Advanced, despite projected losses. Ultimately, NM determined that opportunity existed to further streamline care pathways and, in doing so, eventually break even financially. In addition, NM contends that participation in voluntary Medicare APMs provides a unique opportunity to build competencies such as strategic data analysis, high-risk patient identification, and strong inter—care setting partnerships. Finally, NM argues that although APMs hold promise, policy changes must be made to ensure the long-term sustainability of models such as BPCI Advanced.

The American Journal of Accountable Care. 2019;7(3):30-33

Over the past decade, the United States has experienced a large increase in the number of providers participating in alternative payment models (APMs), which are designed to incentivize the provision of cost-efficient and high-quality care. In 2018, the Health Care Payment and Learning Action Network reported that the percentage of healthcare payments reimbursed under shared savings, bundled payment, and/or population-based payment models had increased from 23% in 2015 to 34% in 2017. This climb was facilitated by the large number of new APMs, such as the recently launched Bundled Payments for Care Improvement (BPCI) Advanced model. With the increasing availability of voluntary models, providers have a number of factors to consider when contemplating participation. Using BPCI Advanced as a case study, this article explores these decision factors, as well as how much stock should be placed in projected financial performance. Finally, we present thoughts regarding the long-term sustainability of APMs such as bundled payment programs.

BPCI Advanced Overview

Under BPCI Advanced, participants may select between 1 and 37 clinical episodes for which to accept financial risk, meaning that they are accountable for the care that beneficiaries receive and the associated healthcare expenditures across a 90-day postdischarge period. If Medicare payments for services provided to beneficiaries during this time exceed a target price, the participant must pay back the difference to Medicare.

Meanwhile, if payments fall below the target price, participants are eligible to receive the difference as savings. Northwestern Medicine (NM) is participating in BPCI Advanced with the major joint replacement of the lower extremity (MJRLE) episode (eg, knee replacement, hip replacement). It is doing so despite the fact that it is projected to sustain a loss in this episode based on the current data made available to it by the Center for Medicare and Medicaid Innovation (CMMI).


Like many healthcare organizations, NM chose to enter its first Medicare bundled payment program with the MJRLE episode. Relative to other conditions, such as simple pneumonia, MJRLE is often viewed as comparatively easier to operationalize in a bundled payment model due to the relatively low variation in patient acuity; the mostly elective nature of the procedure, which lends to easier patient identification; easily identifiable efficiency opportunities; and ease of mapping out optimal care pathways from before to well after the procedure.

NM entered the original BPCI in 2015 and successfully executed a multidisciplinary, pre- and postoperative clinical pathway across the clinical care continuum with high provider adherence rates for patients undergoing MJRLE. Early in the program, performance was uneven, and NM even experienced a few quarters of losses, but its financial performance improved and became more consistent as it executed and refined interventions. To date, NM has averaged a 4% quarterly savings rate in BPCI MJRLE episodes.

Despite this fact, when NM received its preliminary BPCI Advanced target prices, it found itself in an unfortunate position, as did many other health systems with previous success in BPCI MJRLE or the mandatory Comprehensive Care for Joint Replacement program. When NM compared average episode payments from the most recent full year of BPCI Advanced baseline data (fourth quarter of 2015 to third quarter of 2016) with the 2018 preliminary target prices, NM was projected to lose approximately 9% over the course of a year.

Why Is NM Projected to Lose Money?

How is it that NM could generate average savings of 4% in one program but be projected to lose 9% in another program for the same 90-day MJRLE episode? The answer comes down to the target price methodology. Under the original BPCI program, participants’ target prices were based on their own historical average episode payments. The methodology utilized for BPCI Advanced is much more complex. Instead of focusing on a provider’s own historical experience alone, BPCI Advanced uses national data with regional and provider-specific adjustments. Rather than a simple historical average, BPCI Advanced uses 2 compound log normal models to estimate the impact of various patient- and provider-level factors on episode spending. These models generate parameter estimates, which are ultimately used to calculate the 3 major components of the target price calculation (Table). Together, the components generate a target price that is adjusted for patient case mix and the hospital’s historical efficiency and trended forward based on the episode utilization of like providers in a regional peer comparison group. Although this method is complex, CMMI is to be applauded for developing the most sophisticated and comprehensive risk adjustment methodology featured in a Medicare bundled payment model to date. The modification also appropriately acknowledges the need to move beyond purely historical averages, as over time efficient providers would be disincentivized to participate.

Back to the original question: What about this methodology makes it more difficult for NM to achieve savings in the MJRLE episode? One of the main drivers is the peer adjusted trend (PAT) factor, which is based on the utilization of like providers in a Census region. NM’s MJRLE episode has a PAT factor of less than 0.90, which essentially means that the utilization and corresponding expenditures for the MJRLE episodes initiated by providers in NM’s peer comparison group were lower than what CMMI predicted. In other words, providers in this peer group, like NM, have already become more efficient at delivering care to patients undergoing MJRLE. Additionally, the PAT factor effect is likely to compound over time as providers continue to pursue efficiencies in the MJRLE care pathway, and the resultant target price will continue to decline. Similar arguments can be made for the acute care hospital (ACH) efficiency measure, which is used to calculate the standardized baseline spending (SBS) figure. The SBS will decline as a provider becomes more efficient.

Why NM Is Moving Forward

With all this in mind, readers are probably, rightfully, wondering why NM decided to participate in BPCI Advanced with the MJRLE episode. The answer boils down to a few key points:

NM believes that further opportunity exists to better coordinate care for patients undergoing MJRLE. During BPCI, it made great strides toward optimizing the immediate pre- and postoperative periods for patients undergoing MJRLE. NM engaged physicians through data sharing and education, identified and conducted risk assessments to ascertain the appropriate discharge setting for patients, proactively educated patients on their care pathways to help set informed expectations, and partnered with skilled nursing facilities to coordinate care for patients and set quality goals. Process iteration has revealed what works and where roadblocks lie. NM has yet to meet all of its quality and utilization goals, but it believes that lessons learned during BPCI have enabled it to develop the plans and infrastructure necessary to attain these goals during BPCI Advanced.

NM believes the pursuit of these opportunities will enable us, at a minimum, to break even. Generating buy-in to engage in a new model is difficult if stakeholders believe that the economics of the model are so stacked against them that no matter what they do, losses are a certainty. If NM is able to meet its quality and utilization goals for the MJRLE population, it is projected to eventually break even in the MJRLE episode.

NM has momentum. NM participated in the BPCI original MJRLE episode for 2.5 years, and it took considerable time to build the infrastructure and clinical engagement necessary to generate success. After spending more than 2 years sharing data and results with physicians, more and more NM physicians understand the role of APMs and how they can translate to care improvements for patients. NM does not want to lose this crucial engagement.

New competencies are necessary to succeed in a rapidly shifting reimbursement landscape. Finally, the BPCI Advanced MJRLE episode is not the end game. NM strives to develop the competencies necessary to effectively manage patient care within and beyond its walls for all patient populations. As time goes on, the economic incentives established by payers will increasingly align with this goal. HHS Secretary Alex Azar recently signaled his intent to release additional mandatory bundles1 while CMS Administrator Seema Verma has touted the benefits of Medicare Advantage.2 The lessons that NM learns and the capabilities that it develops from managing the Medicare MJRLE population will be applied to the broader orthopedics population, as well as to patients with other disease groups, in the future.

Recommendations for Long-term Sustainability

This article was written in the context of NM’s current preliminary MJRLE target price. Although the final target price will be updated to reflect Medicare rate changes and its patient case mix, NM does not expect its projection to change significantly as a result. However, it is likely that the target price could decline in future performance years when Medicare sets a new baseline period and reruns its methodology. It is possible that following this change, NM (and many other BPCI Advanced participants) may be forced to withdraw from the MJRLE episode due to significant unavoidable financial losses.

CMMI can prevent this with refinements to the target price methodology. Specifically, we recommend that CMS consider one or all of the following possible changes:

Do not rebase; utilize the 2013-2016 period as a fixed baseline for the entirety of BPCI Advanced. CMS is set to rebase targets for performance year 2020, meaning it will alter the baseline to include more recent years—years that reflect a higher portion of providers pursuing efficiencies in the MJRLE care pathway. This action will inevitably cause target prices to decline. In order to provide more predictability and stability to target prices, CMS should elongate the periods between rebasing (minimum of 3 years). If and when rebasing is required, CMS should simply elongate the baseline by 1 year (add on 2017) instead of shifting the entire period forward (eliminating 2013 and potentially 2014).

Apply a floor to the ACH efficiency measure and/or PAT factor. As previously stated, ACH efficiency measures and PAT factors below 1.0 signal that a provider is historically efficient and that providers in peer comparison groups are historically efficient, respectively. These factors are directly applied to a dollar amount to calculate the target price. To prevent target prices from reaching unsustainably low figures, CMS could place a floor on these adjustment factors by prohibiting them from falling below 0.85, for example. Identifying an appropriate floor would require extensive consideration, but the principle of applying a floor is a worthy and sensible policy. CMS has previously applied caps when setting target prices in other models, as it capped quarterly changes in trend factors under the original BPCI.

All of these proposals seek to address the ever-present challenge of any APM’s benchmark methodology: how to prevent the race to the bottom. When a model adjusts a target price to reflect a single provider’s or provider group’s historical efficiency, the target will continue to decline as providers become more efficient. However, there is a floor to this efficiency beyond which the decreases in utilization necessary to match a declining target price would likely threaten the quality of patient care. Thus, a floor must exist in target price methodologies.


APMs such as BPCI Advanced play an important role in healthcare. It is likely that innovation will occur at NM as a result of participation and that performance will continue to improve. Its efforts and the lessons learned via participation will position NM as best possible to manage future challenges in joint replacement and other episodes of care. However, the threat and realization of significant financial losses will stifle innovation and performance improvement if NM and others withdraw from participation. There is a floor to possible efficiencies in care, and CMS must acknowledge this fact as it refines target price methodologies.Author Affiliations: Northwestern Memorial HealthCare (JW, HAJ, BW, DM), Chicago, IL.

Source of Funding: None.

Author Disclosures: The authors report no relationship or financial interest with any entity that would pose a conflict of interest with the subject matter of this article.

Authorship Information: Concept and design (JW, HAJ, BW, DM); analysis and interpretation of data (JW, DM); drafting of the manuscript (JW, DM); critical revision of the manuscript for important intellectual content (JW, DM); administrative, technical, or logistic support (HAJ, BW); and supervision (HAJ, BW).

Send Correspondence to: Jessica Walradt, MS, Northwestern Memorial HealthCare, 251 E Huron St, Chicago, IL 60611. Email: Jessica.walradt@nm.org.REFERENCES

1. Azar AM II. Remarks on primary care and value-based transformation. HHS website. hhs.gov/about/leadership/secretary/speeches/2018-speeches/remarks-on-primary-care-and-value-based-transformation.html. Published November 8, 2018. Accessed March 15, 2019.

2. Speech: remarks by CMS Administrator Seema Verma at the Medicare Advantage and Prescription Drug Plan Spring Conference. CMS website. cms.gov/newsroom/fact-sheets/speech-remarks-cms-administrator-seema-verma-medicare-advantage-and-prescription-drug-plan-spring. Published May 9, 2018. Accessed March 15, 2019.

Related Videos
Carrie Kozlowski
Carrie Kozlowski, OT, MBA
Carrie Kozlowski, OT, MBA
kimberly westrich
Bruce Sherman, MD
Dr Jeffrey Sippel
Lalan Wilfongd, MD, US Oncology Network
Related Content
© 2024 MJH Life Sciences
All rights reserved.