The 2010 Patient Protection and Affordable Care Act (ACA) accelerated the transition from traditional fee-for-service (FFS), volume-based reimbursement to value-based reimbursement (VBR), by introducing a variety of new initiatives and payment models.¹ Although the volume-to-value transition is now several years old, data regarding the effectiveness of these programs is still minimal, and the analyses of the data that is available often contradict each other. Two recent examples of VBR models include accountable care organizations (ACOs) and bundled payment models, such as the Bundled Payment Care Improvement (BPCI) Initiative and the Comprehensive Care for Joint Replacement (CJR) model, both of which models were recently examined as to their effectiveness in reducing healthcare spending.

Federal Accountable Care Organizations (ACOs)

ACOs are organizations which physicians, hospitals, and other providers voluntarily join, that seek to offer quality coordinated care and reduce spending.² In most ACO models (federal and commercial), when these entities succeed in both lowering cost growth and meeting quality performance standards, they are able to obtain some amount of shared savings from the payor, e.g., the Centers for Medicare & Medicaid Services (CMS).³ Currently, Medicare Shared Savings Program (MSSP) ACOs are the largest type of this value-based model, with 561 organizations to date, serving 10.5 million Medicare beneficiaries.⁴

On August 9, 2018, CMS released a proposed overhaul of the current risk structure of MSSP ACOs, entitled Pathways to Success Initiative.⁵ This comprehensive initiative would impose more accountability on ACOs, promote patient engagement, and incorporate new technology, among others.⁶ Increased ACO accountability would be accomplished by decreasing the amount of time during which an ACO could participate in upside-only risk, from six to two years, and introducing down-side risk (i.e., shared losses) after those two years.⁷ Shared savings incentives would also decrease from 50% to a maximum of 25%.⁸

CMS Administrator, Seema Verma, presented an analysis of ACO performance data as a basis for why the ACO risk structure should be overhauled, CMS’s snapshot analysis of Track 1 ACOs in 2016 suggests that Medicare costs for these entities increased relative to their target costs,⁹ indicating that upside-only ACOs (both physician-led and hospital-based) had a positive net impact, or increased costs for Medicare. In response, Seema Verma stated, “Medicare cannot afford to support programs with weak incentives that do not deliver value.”¹⁰ Supplementing the poor MSSP ACO outcomes data, CMS conducted projections of the Pathways to Success’s financial impact, estimating savings to Medicare of $2.2 billion over 10 years.¹¹ In contrast to MSSP ACOs, CMS’s evaluation of Next Generation ACOs, which share 80-100% of financial risk, showed a net reduction in Medicare spending, totaling $62.12 million in 2016.¹² This study demonstrates that ACOs can succeed in a downside risk model, providing the foundation for CMS’s assertion that MSSP ACOs should increase risk after 2 years.

Reacting to the proposed structural changes to MSSP ACOs and calculations of federal spending by CMS, the National Association of ACOs (NAACO) released a study suggesting that there were considerably larger savings to Medicare federal spending than CMS analyses suggested. The NAACO study, conducted by Dobson Davanzo & Associates, found that MSSP ACOs saved Medicare $1.84 billion between 2013 and 2015, rather than the $954 million in savings reported by CMS.¹³ After accounting for ACO bonuses, the NAACO study found that MSSP ACOs decreased federal spending by $542 million between 2013 and 2015 – this study stands in direct contrast to the estimated $344.2 million decrease in savings based on CMS’s benchmarks.¹⁴ Another peer-reviewed study by Harvard University researchers, similar to the NAACO study, indicated decreases to Medicare spending, wherein ACOs saved more the longer they participated in the MSSP.¹⁵ This Harvard study also reported that the reduction in FFS spending was 39% greater than what was reported by CMS and net savings to Medicare was 2.8 times greater.¹⁶

While the CMS and NAACO studies both utilized the same set of data, the vast difference in their results is due to the analysis methodology. CMS utilized an administrative formula building off of the benchmarking used to set financial targets of the program,¹⁷ while the NAACO study used the difference-in-differences regression,¹⁸ which compared (a) changes in Medicare spending for ACOs before and after entry into the MSSP to (b) changes in spending by those not participating.¹⁹ It is important to note that CMS used the difference-in-differences methodology in comparing both the Next Generation and Pioneer ACOs, but not the MSSP ACOs, commenting that the reason for using divergent methodologies in evaluating these ACOs and MSSP ACOs was established by the ACA, which contains different evaluation requirements than ACOs established by the CMS Innovation Center.²⁰

In response to the CMS proposal, nine stakeholder groups, including the Medical Group Management Association (MGMA), America’s Health Insurance Plans (AHIP), American Hospital Association (AHA), and American Medical Association (AMA), support the improvements made to the program, but urged CMS to acknowledge the potential unintended consequences. Most notably, CMS does not recognize the millions of dollars of an organization’s own capital that is required to implement an ACO or acknowledge savings presented in other peer-reviewed studies using different methodologies, including the NAACO study. Furthermore, these stakeholders assert that the CMS proposal should be modified to: (a) allow more time for ACOs to be in the shared savings only model; and, (b) keep at least the current shared savings rate of 50%.²¹ A separate survey conducted by NAACO found that over 71% of ACOs were more than likely to leave the program if faced with down-side financial risk in 2019.²²

At this time, CMS has not responded to the results of the conflicting studies or stakeholder comments; however, there is an anticipated response after the 60 day comment period, which closed on October 16, 2018.²³

Bundled Payments

In addition to the recent scrutiny related to the effectiveness of shared savings models, bundled payments have also been analyzed as to their success in achieving the aims of VBR. Bundled payment models take a different approach from ACOs in lowering costs and increasing value. The voluntary BPCI Initiative is intended to cut costs for an episode of care, by paying organizations a single “bundled” payment for that entire episode, encouraging care coordination and unnecessary utilization, because the provider would otherwise effectively lose money on the episode.²⁴ There are four (4) bundled payment models under the BPCI, each of which include different types of services in the associated bundled payment.²⁵ Model 1 of the BPCI (currently inactive) included only Medicare Part A inpatient hospital services, rendered during the episode of care, as part of those services to be reimbursed through the model’s bundled payment.²⁶ Model 2 is the most heavily utilized, bundling payment for acute hospitals and up to 90 days of post-acute care.²⁷  Model 3 bundles payments for post-acute care, excluding acute inpatient hospital stays, and Model 4 is the only prospective payment, bundling acute inpatient hospital stay only.²⁸

Early analysis on the BPCI Initiative suggested that bundled payments generate savings, with a 2016 study (which analyzed the first 21 months of the BPCI program) finding that payments declined approximately $1,166 more per lower extremity joint replacement (LEJR) episode when compared to non-participating hospitals.²⁹ The most recent CMS evaluation of BPCI Models 2 through 4 indicates that Medicare payments were reduced relative to the comparison group in BPCI using the difference-in-differences methodology.³⁰ However, after taking into account the average net payment reconciliation amount (NPRA) paid to participants, the Medicare program likely did not achieve savings for a vast majority of the clinical models.³¹

The CJR, another CMS bundled payment model that was originally mandatory in selected markets, was designed in order to determine whether LEJR bundled payments would succeed when implemented in different hospitals with diverse infrastructures and market composition.³² An early study of the CJR program revealed that joint replacement surgery decreased total spending per episode by as much as 20% between July 2008 and June 2015 for 3,738 episodes of joint replacement without complications.³³ Additionally, a Journal of the American Medical Association (JAMA) study on the CJR model found that, in the first year, there were no significant differences in the admission of patients with lower risk; however, they also found that there were no significant changes in Medicare spending after bonus payments.³⁴ This lack of Medicare savings could be due to the fact that CJR was originally mandatory, incorporating organizations that were not prepared to handle the program, among other reasons.³⁵

A major concern of these studies on the effectiveness of bundled payment models (CJR and BPCI) is that the savings are due to organizations by increasing the volume of episodes paid for by Medicare with lower risk patients and deterring higher risk patients³⁶ to attempt to increase their reimbursement, consequently “padding the numbers” of the study.  A September 2018 JAMA study addressed this concern by measuring the market-level LEJR volume before and after the BPCI periods for hospitals.³⁷ Out of the over 1.7 million beneficiaries observed, it was determined that participation in the BPCI did not affect the case mix or case volume when using the adjusted difference-in-differences estimate.³⁸

Despite the indeterminate and conflicting results of bundled payment savings, on January 9, 2018, CMS announced the new BCPI Advanced model, which builds upon the apparent successes the original BPCI Models.³⁹ BPCI Advanced will: (a) have a bundled period of only 90 days, rather than the choice of 30, 60, or 90 days provided in the original BPCI; (b) have a risk adjustment accounting for patient case mix of the benchmark price at which costs are measured; (c) increase risk from the start of the program; and, (d) link payment to quality measures, incorporating a value aspect.⁴⁰

VBR methods have achieved increasing popularity among public and private payors in the healthcare industry, but their effectiveness is still indeterminate, despite both CMS and external studies on the topic. The data on these VBR models vary in relation to methods used and timeframe, rendering difficult any comparisons between the studies and their reliability. Regardless of their effectiveness, both ACOs and bundled payments remain active, new models are being introduced, and current models are being further modified, in an effort to hold healthcare providers accountable for both their spending and their quality of care.