Medicare Accountable Care Organizations: Past Performance and Future Directions

Summary

This Congressional Budget Office report summarizes recent research findings about Medicare accountable care organizations (ACOs) and the factors that have contributed to or limited their ability to achieve net budgetary savings for the Medicare program. ACOs are groups of providers, such as physicians and hospitals, that assume responsibility for the quality and cost of care for an assigned group of patients.1 Providers participate in Medicare ACO programs voluntarily. CBO found the following:

  • Certain types of ACOs are associated with greater savings. They include ACOs led by independent physician groups, ACOs with a larger proportion of primary care providers (PCPs), and ACOs whose initial baseline spending was higher than the regional average. (An ACO’s baseline spending is generally the average spending per person in the Medicare fee-for-service, or FFS, program among beneficiaries that would have been assigned to the ACO over several calendar years before the start of the ACO’s contract period.)
  • Some factors limit the savings from Medicare ACOs. Those factors include weak incentives for ACOs to reduce spending, a lack of the resources necessary for providers to participate in ACO models, and providers’ ability to selectively enter and exit the program on the basis of the financial benefits or losses they anticipate from participating.

Researchers and outside experts have suggested various policy approaches that could increase the savings that ACOs generate for the Medicare program. Those approaches include increasing providers’ incentives to participate in ACO models, increasing their incentives to reduce spending, and increasing beneficiaries’ awareness of and engagement with ACO models. CBO has not assessed the effects of those policy approaches or determined their net budgetary impact.

ACOs and the Medicare Shared Savings Program

In previous analyses, CBO assessed the budgetary effects of Medicare ACO programs, which have been operating for more than a decade. From those assessments, CBO concluded that some ACO models produced small net savings, but those savings were not a major factor in the slowdown in the growth of per-person Medicare spending that occurred over the past decade.2

To further evaluate Medicare ACOs and assess the factors that aided or hindered their performance in achieving savings, CBO took two approaches. First, the agency reviewed the relevant research literature, focusing on studies from peer-reviewed journals, official evaluations of Medicare ACOs, and reports from think tanks and other research organizations. Second, CBO consulted with outside experts from eight organizations, including industry executives, academic professors, former government officials, a leader of an ACO, and representatives of a philanthropic organization.

Those reviews and discussions mostly centered around the Medicare Shared Savings Program (MSSP)—the permanent and largest ACO program in the Medicare FFS program—which had about 11 million beneficiaries in performance year 2023.3 (A performance year refers to a specific period during which the performance of an ACO is evaluated. It typically aligns with the calendar year.) Various other Medicare ACO models have been developed and operated by the Center for Medicare and Medicaid Innovation (CMMI) as part of that agency’s statutory mission to test new ways to deliver and pay for health care, but those models have been limited in duration.4

Comparing Spending for Beneficiaries Assigned to an ACO With a Benchmark

The financial performance of ACOs can be evaluated in more than one way. Under the MSSP, the Centers for Medicare & Medicaid Services (CMS) evaluates an ACO’s financial performance by comparing the spending for beneficiaries assigned to it with a specific target, known as a benchmark.5 Benchmark-setting approaches differ among ACO models. In general, the initial benchmark for an ACO is determined by combining data on the historical spending among beneficiaries who were eligible for assignment to the ACO during the baseline period with the average spending in that ACO’s region; then, for the rest of the contract period, that amount is projected to grow on the basis of trends in FFS spending. After the initial contract period ends, the benchmark is reset for another period, using the historical spending among beneficiaries who would have been assigned to the ACO (including that from the previous contract period) and the average regional spending.

The difference between the benchmark and the annual spending among an ACO’s beneficiaries determines the savings or losses for that ACO. In the MSSP, if annual spending for beneficiaries assigned to an ACO is less than the ACO’s benchmark for that year and the ACO meets certain standards for quality of care, then CMS pays a predetermined percentage of those savings to the ACO. That amount, known as the shared savings rate, ranges from 40 percent to 75 percent for the 2024 performance year.

ACOs whose annual spending among their attributed beneficiaries exceeds the benchmark may or may not have to pay a penalty. Under MSSP agreements with what is known as downside financial risk, an ACO whose spending among its attributed beneficiaries exceeds its benchmark generally has to pay a portion of that additional spending to CMS. The amount of that payment is largely determined by the shared loss rate defined in the ACO’s participation agreement, which in 2024 ranges from 30 percent to 75 percent. (ACOs in the MSSP that have participation agreements with no downside financial risk are not liable for financial penalties if spending among their beneficiaries exceeds the benchmark.)6

Comparing Spending for Beneficiaries Assigned to an ACO With Spending for a Control Group of Beneficiaries

A second way in which the financial performance of ACOs can be evaluated is by comparing changes in spending among beneficiaries assigned to ACOs with changes in spending among a control group of beneficiaries not assigned to ACOs. In that counterfactual approach, the control group generally approximates what spending growth would have been in the absence of an ACO program. (By contrast, ACO benchmarks are not an estimate of counterfactual spending among the beneficiaries assigned to an ACO. Rather, they are a design feature of the ACO model intended to accomplish certain policy goals—such as encouraging providers to participate in the program and motivating them to achieve greater savings.) CBO put considerably more weight on the results of studies and evaluations that used that counterfactual approach to determine the effects of ACOs on Medicare spending. Studies that used a counterfactual approach generally found that ACOs, particularly those in the MSSP, were associated with small net budgetary savings to the Medicare program.7

Factors Associated With Increased Savings

In its review of the literature, CBO found that certain types of ACOs were associated with greater savings for the Medicare program, in terms of both gross savings and net savings. Gross savings represent the reduction in spending on benefits from ACOs’ activities. Net savings equal gross savings minus any shared savings payments to ACOs plus any payments made from ACOs to CMS in cases in which ACOs’ contracts entail downside risk and their spending exceeds the benchmark. Most of the literature CBO reviewed reported both types of savings.

ACOs associated with greater gross and net savings include those led by independent physician groups, those with a larger proportion of primary care providers, and those whose initial baseline spending was higher than the regional average. Those findings are consistent with the views of the many experts interviewed by CBO.

ACOs Led by Independent Physician Groups

Several studies found that ACOs led by independent physician groups generated substantially larger savings than ACOs led by hospitals.8 Two potential explanations underlie that finding.

First, independent physician groups have clear financial incentives to reduce hospital care and thereby generate shared savings, whereas hospital-led ACOs have conflicting incentives.9 Compared with physician groups, hospitals have larger fixed costs, which they incur regardless of the quantity of services they provide. If hospitals reduced admissions, then they would earn less revenue to cover their fixed costs, weakening their incentive to generate shared savings.10

Second, compared with physician groups, hospitals have less direct control over which types of services are provided to their patients. Hospitals have to find ways to align the behavior of the providers who are on the frontlines of care with the hospital’s objectives, whereas physician groups have more direct control, prescribing medications to their patients and handling referrals for ancillary services and specialty care. That arrangement allows physician group ACOs to more easily redirect their patients’ care, steering them away from services that have low value (in other words, services that yield little or no health benefits relative to their costs).

ACOs With a Larger Proportion of Primary Care Providers

CBO’s review of the literature found some evidence that Medicare ACOs with a higher proportion of PCPs, advanced primary care practices (which limit the number of patients assigned to a clinician to ensure high-quality, comprehensive care), or patient-centered medical homes (a care-delivery model that focuses on providing coordinated, comprehensive, and patient-centered primary care) were more likely to generate larger savings. Some experts regard advanced primary care as a key contributor to ACOs’ success.11

Those larger savings mainly stem from PCPs’ role in redirecting patients from higher-cost settings to lower-cost settings. For instance, PCPs can use primary care visits to help manage their patients’ chronic conditions or provide preventive screenings, which may reduce spending on more expensive services, such as specialty care, advanced imaging, inpatient care, or postacute care.12 Under the ACO framework, PCPs are particularly well positioned to facilitate such reductions in spending because Medicare beneficiaries are assigned to an ACO on the basis of their use of primary care services through a process known as attribution.13

ACOs Whose Initial Baseline Spending Was Higher Than the Regional Average

ACOs with initial baseline spending above the average in their local area generated larger savings than ACOs with spending below the local area average.14 That finding comes from an early study of the MSSP and a study of the Pioneer ACO model, a demonstration project operated by CMMI.

One reason that ACOs with higher-than-average initial baseline spending have historically been more likely to generate savings is that their assigned beneficiaries might be receiving more low-value care at the onset of the ACOs’ first performance period and that care is easier to reduce when there is more of it.15 That is the case for postacute care, in particular; less spending on that type of care is a primary factor contributing to such savings.16 That is because postacute care, which includes skilled nursing care and rehabilitation services, is costly and has the potential to be overused or used inefficiently, making it a readily identifiable target for ACOs aiming to decrease spending. (ACO provider groups typically do not include postacute care providers, but they are responsible for the total spending of their assigned beneficiaries. Therefore, they generally have clear financial incentives to limit that type of care.)17

Factors Associated With Limited Savings

CBO’s review of the literature points to several factors that limit the performance of Medicare ACOs and thus the savings they are able to achieve. Those factors include weak incentives for providers to reduce spending, a lack of the resources necessary for providers to participate in Medicare ACOs, and a model design that facilitates favorable selection (meaning that the providers who initially participated in an ACO model and continue to do so may be the ones who were most likely to achieve savings to begin with).

All of those factors stem from the trade-offs and challenges inherent in designing an effective ACO model in which providers’ participation is voluntary. A key consideration under a voluntary model is to strike a balance between having financial incentives that are appealing enough to providers to encourage participation while still yielding substantial savings for the Medicare program.

Weak Incentives for Providers to Reduce Spending

An ACO’s benchmarks are empirically determined—that is, they are set and updated over time to reflect observed health care spending among that ACO’s attributed beneficiaries. Because of that structure, ACOs that lower their spending are effectively penalized with lower subsequent benchmarks. That can occur in two ways: explicitly (when ACOs renew their contract in the MSSP) or implicitly (when they switch to a different ACO model).18 The explicit reduction (known as a ratchet effect) greatly weakens ACOs’ incentives to reduce their spending.

In certain cases, the shared savings rates may not be high enough to incentivize providers to generate savings. That is the case particularly for large health systems (including hospitals), whose incentives to reduce low-value care or shift care to lower-cost settings are weakened by the loss in FFS revenue they would experience by making those changes and by those facilities’ higher fixed costs.19

Lack of the Resources Necessary for Providers to Participate in ACO Models

Providers who want to participate in an ACO must pay the start-up costs involved with program setup. Those costs can be steep and can serve as a barrier to participation, especially for smaller physician practices and providers in rural or underserved areas. Those types of providers may not have the up-front capital and infrastructure needed to implement information technology systems or establish care coordination processes with other providers in an ACO’s network. For instance, providers in rural areas have high fixed costs and relatively few patients (and thus fewer billable services), resulting in smaller operating margins and contributing to the challenges of participating in an ACO model, particularly one with downside risk.20

CMS has taken steps to alleviate some of those burdens and spur participation among providers. Those steps apply to ACOs and, in some cases, other alternative payment models, or APMs. To begin with, CMS has provided up-front payments (lump sum or recurring) to ACOs in rural and underserved areas, which are then recouped from those ACOs’ future shared savings.21 In addition, CMS provides what are known as APM incentive payments to eligible clinicians who participate in certain payment arrangements that reward high-quality and cost-efficient care. Those payments, which are in addition to the professional fees paid under the Medicare physician fee schedule, are calculated as a percentage of the estimated aggregate payments for professional services covered under Medicare Part B among a clinician’s entire patient pool during the base period (the calendar year immediately preceding the year when a clinician receives that incentive payment).22

APM incentive payments have two limitations. One is that those payments vary in size in absolute dollar terms, proportionally increasing with clinicians’ revenue. As a result, the size of the incentive payment is much smaller for lower-revenue providers (such as primary care providers and clinicians in independent physician practices) than it is for higher-revenue providers (such as specialists and clinicians in large health systems).23 Therefore, incentive payments tied to FFS revenue may inadvertently discourage savings because providers may prefer to maintain their greater revenue and the associated incentive payments.

A second limitation of incentive payments is that they are generally restricted to providers in Advanced APMs with downside financial risk. (An Advanced APM is a type of APM that includes specific features, such as incorporating performance on quality measures as a factor when determining payment to participants, using certified electronic health record technology, and assuming financial risk in instances when the spending for an APM’s assigned beneficiaries exceeds the target amount specified under a certain type of payment model. Some, but not all, ACOs are considered Advanced APMs.)24 For that reason, provider groups that operate under ACO contracts without downside risk or that are less likely to transition to such risk arrangements—as has traditionally been the case with rural ACOs—cannot receive such incentive payments regardless of the savings they achieve.25

Favorable Selection Under a Voluntary Model

Because participation in Medicare’s ACO models is voluntary, providers can selectively enter the program and exit the program after their contract period ends. If providers anticipate financial benefits from joining an ACO, then they can choose to enter the program—and if they anticipate losses, then they can decide to exit or not enter the program.

That dynamic is particularly sensitive to the way benchmarks are set. Starting in 2017, CMS began blending historical spending among beneficiaries assigned to an ACO with average spending in that ACO’s region. Regionally blended benchmarks weaken the link between an ACO’s savings and its subsequent benchmark, thereby reducing the ratchet effect and increasing incentives for providers to reduce spending. At the same time, regionally blended benchmarks favor providers with baseline spending levels that are lower than the regional average because those providers would find it easier to stay below the regionally blended benchmark and earn shared savings. One study found evidence that, following the introduction of regionally blended benchmarks, the composition of participants in the MSSP shifted toward entrants with lower baseline spending.26

Such shifts in participation can reduce Medicare’s savings. That is because ACOs whose initial baseline spending is higher than average spending have greater potential for reducing spending, and ACOs whose initial baseline spending is lower than average may earn shared savings without necessarily reducing actual spending.

Other aspects of the ACO model’s design—such as a risk-adjustment system, which is intended to adjust payments on the basis of patients’ diagnoses and other characteristics—may exacerbate those participation dynamics. A risk-adjustment system that does not adequately capture differences in beneficiaries’ health among ACOs, for instance, may create an uneven playing field for providers. Inadequate risk adjustment can put certain ACOs, particularly those serving patients who are more costly than risk adjustment predicts, at a disadvantage.27

Approaches that require providers to accept downside risk under a voluntary program can further weaken incentives to participate, particularly among providers that are less confident about meeting benchmark spending targets and achieving financial benefits.28 Following a change in the MSSP that accelerated the timeline for ACOs to accept downside risk, more ACOs exited the program than entered it for the first time since the MSSP launched.29

The extent of savings generated by ACOs that participate in the MSSP is also a subject of debate because of questions about whether such savings resulted from increased efficiency or from the characteristics of providers who choose to participate in an ACO. CBO reviewed three studies that used different methods to adjust for providers’ selective participation; it found that two studies’ estimated savings were sensitive to those adjustments and one study’s estimate was not.30

Policy Approaches That Could Increase Savings

In its literature review and conversations with experts, CBO identified general policy approaches that could increase the net budgetary savings for the Medicare program generated by ACOs, particularly through the MSSP. Those approaches can be grouped into the following categories: those that would increase providers’ incentives to participate in ACOs, those that would increase providers’ incentives to reduce spending, and those that would increase the awareness and engagement of beneficiaries.

CBO has not assessed the impact of the approaches or their net budgetary effects. Any future assessment from CBO would depend on further analysis and the specific details of the legislative proposal. As part of its nonpartisan mandate, CBO does not make policy recommendations.

Increase Providers’ Incentives to Participate in ACOs

Medicare ACO programs do not require providers to participate.31 Various strategies could be used to strengthen providers’ incentives to participate in voluntary models and potentially achieve savings for Medicare over a longer period.32

  • Make participation more attractive. Examples of strategies in this category include restructuring or extending the APM incentive payments, temporarily limiting downside risk, and improving the accuracy of risk-adjustment systems (for instance, by including measures that better represent patients’ social risk factors, such as poverty, social isolation, and limited community resources).33
  • Make nonparticipation less attractive. Some providers are reluctant to participate in ACOs, so one strategy involves making nonparticipation a less attractive option than participation. Some ways to do that include applying site-neutral payments to hospital outpatient facilities owned by health systems that choose not to participate, excluding health systems that choose not to participate from the 340B drug-pricing program, and allowing payment for telehealth visits only for providers that choose to participate.34 (Site-neutral payments are a reimbursement policy in which payments for the same or similar services are equal even though the care is provided at different sites. The 340B drug-pricing program allows certain hospitals and other covered entities that serve low-income or underserved populations to purchase outpatient prescription drugs at a discount.)
  • Provide capitation payments to PCPs. Two strategies could make more primary care providers want to participate in ACOs. One strategy would be to provide participating PCPs with a fixed payment per patient per unit of time (paid in advance to the clinician for the delivery of health care services) above current levels of primary care spending.35 A second strategy would be to establish hybrid capitation payments—basically, a mix of FFS and capitation payments—for PCPs who participate in ACOs. Physician groups that are oriented toward primary care generally find primary care capitation payments under ACO programs appealing because that structure can support the goals of many primary care practices to decrease overall health care spending by modifying the delivery of primary care and by reducing low-value care without decreasing their FFS revenue.36

Increase Providers’ Incentives to Reduce Spending

This category comprises two main strategies.

  • Revise ACO benchmarks. Several policies could be implemented to change the benchmarks, which influence spending among beneficiaries attributed to an ACO because benchmarks are the basis for shared savings or losses. One such policy would eliminate the rebasing of benchmarks using historical spending among an ACO’s beneficiaries and instead use administratively set benchmarks.37 An administratively set benchmark would start with a base rate (such as historical spending among beneficiaries attributed to the ACO) and then increase that rate over time by some administratively determined factor that was not tied to historical spending among that ACO’s beneficiaries. One example of such an administrative factor’s being incorporated into a benchmark is the Accountable Care Prospective Trend (ACPT) within the MSSP. The growth rate of the ACPT is a modified version of the projected spending growth of the entire Medicare FFS population that was forecast at the start of an agreement period and remains unchanged throughout the agreement period.38 (The MSSP currently uses a three-way blended growth factor for the benchmark, integrating the ACPT into empirically determined growth rates.) Some alternative methods to administratively determine an ACO’s benchmark include aligning it with growth in gross domestic product, the consumer price index for all urban consumers, or another external index.39
  • Increase shared savings rates. All else being equal, higher shared savings rates under the MSSP could motivate providers to further reduce spending for their services because they would receive a larger share of the savings they generated. Although higher shared savings rates would lessen the amount of savings captured by Medicare in the short run, they could boost participation. That pattern of greater participation among providers and a larger share of spending under the MSSP could provide Medicare with a mechanism to control long-term spending growth.40 Increased participation in ACOs could spur additional savings for the broader Medicare program or for other payers as well. For example, savings would accrue if ACO providers delivered care more efficiently to patients not in an ACO.41

Increase Beneficiaries’ Awareness and Engagement

People enrolled in the Medicare fee-for-service program, including those who participate in an ACO, can see any provider that accepts Medicare—they are not locked in to seeing only ACO providers. Strategies that increased beneficiaries’ awareness of the flexibilities available to ACOs (which beneficiaries could take advantage of) and strategies that rewarded patients for obtaining services from ACO providers could enhance communication and engagement between providers and patients and reduce Medicare spending. Such enhancements could transform care patterns—for instance, by encouraging the use of primary care to manage chronic conditions—which could reduce spending through decreased reliance on specialty services, emergency care, and advanced imaging.

  • Increase beneficiaries’ awareness of the flexibilities available to ACOs. Beneficiaries tend not to know about the flexibilities available to many ACOs compared with the rules under the FFS program. For example, certain ACO providers can admit beneficiaries to skilled nursing facilities for covered services without a three-day hospital stay preceding the admission, and some ACOs receive payment for telehealth services provided to beneficiaries at their homes.42 Such flexibilities might be attractive to some beneficiaries and, if those flexibilities were promoted more, could increase beneficiaries’ engagement in choices about their care. One way to increase that awareness would be to allow ACOs to draft information notification letters (subject to CMS’s standards and approval) tailored to their own beneficiaries; that approach would allow ACOs to better explain to beneficiaries the potential benefits of getting care from their providers.43
  • Expand the rewards and strengthen the incentives for beneficiaries to obtain services from ACOs’ providers. Without a clear financial incentive, beneficiaries tend not to obtain care exclusively from ACO providers. One way to strengthen that incentive would be by waiving Part B cost-sharing requirements for beneficiaries who obtain care from their primary care physician or another provider in the ACO.44 Another approach would be to give ACOs more flexibility in determining how to share the savings they generate with their beneficiaries—perhaps by providing an incentive payment to each assigned beneficiary for services beyond those related to primary care.45

1. Medicare Payment Advisory Commission, Accountable Care Organization Payment Systems (revised October 2023), http://tinyurl.com/vsbhtytw (PDF).

2. Congressional Budget Office, Federal Budgetary Effects of the Activities of the Center for Medicare & Medicaid Innovation (September 2023), Box 1, www.cbo.gov/publication/59274; and testimony of Chapin White, Director of Health Analysis, Congressional Budget Office, before the Senate Budget Committee, Alternative Payment Models and the Slowdown in Federal Health Care Spending (October 18, 2023), www.cbo.gov/publication/59660.

3. Centers for Medicare & Medicaid Services, “Shared Savings Program Fast Facts—as of January 1, 2023,” http://tinyurl.com/55aeatfm (PDF).

4. Congressional Budget Office, Federal Budgetary Effects of the Activities of the Center for Medicare & Medicaid Innovation (September 2023), www.cbo.gov/publication/59274.

5. Centers for Medicare & Medicaid Services, “Shared Savings Program Fast Facts—as of January 1, 2023,” http://tinyurl.com/55aeatfm (PDF). Under the MSSP, the ACO is responsible for all Medicare Part A and Part B spending. Also see Centers for Medicare & Medicaid Services, Medicare Shared Savings Program: Application Reference Manual (March 2023), http://tinyurl.com/yc6vwcpe (PDF).

6. ACOs in the MSSP are eligible for shared savings or liable for repayments to CMS if spending for their attributed beneficiaries as a percentage of their benchmark exceeds a minimum savings rate or minimum loss rate. For additional details about how the MSSP calculates shared savings or the amount ACOs must pay to CMS, see Centers for Medicare & Medicaid Services, Medicare Shared Savings Program: Shared Savings and Losses, Assignment and Quality Performance Standard Methodology Specifications, version 11 (January 2023), https://tinyurl.com/5fkebs9h (PDF).

7. Studies that have evaluated the MSSP using a counterfactual approach vary in terms of the time periods and comparison groups used as well as in whether they include the potential effects on the delivery of services to beneficiaries not assigned to an ACO. Those differences make it difficult to compare estimates among studies or to arrive at a single estimate of the effect of the program. Additionally, more recent studies using a counterfactual approach are limited. Some researchers attribute that to the increasingly difficult task of finding a reasonable comparison group. For further discussion, see J. Michael McWilliams and others, “Medicare Spending After 3 Years of the Medicare Shared Savings Program,” New England Journal of Medicine, vol. 379, no. 12 (September 2018), pp. 1139–1149, https://doi.org/10.1056/NEJMsa1803388; J. Michael McWilliams, “Changes in Medicare Shared Savings Program Savings From 2013 to 2014,” JAMA, vol. 316, no. 16 (October 2016), pp. 1711–1713, https://doi.org/10.1001/jama.2016.12049; Carrie H. Colla and others, “Association Between Medicare Accountable Care Organization Implementation and Spending Among Clinically Vulnerable Beneficiaries,” JAMA Internal Medicine, vol. 176, no. 8 (August 2016), pp. 1167–1175, https://tinyurl.com/5ppdh2ra; J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and Medicare Program; Medicare Shared Savings Program; Accountable Care Organizations—Pathways to Success and Extreme and Uncontrollable Circumstances Policies for Performance Year 2017, 83 Fed. Reg. 68049 (December 31, 2018), https://tinyurl.com/yek6ndtx.

8. CBO identified the following studies that examine the performance of physician-led ACOs compared with hospital-led ones; all of these analyses are based on a counterfactual approach. See J. Michael McWilliams and others, “Savings or Selection? Initial Spending Reductions in the Medicare Shared Savings Program and Considerations for Reform,” Milbank Quarterly (July 22, 2020), https://doi.org/10.1111/1468-0009.12468; J. Michael McWilliams and others, “Medicare Spending After 3 Years of the Medicare Shared Savings Program,” New England Journal of Medicine, vol. 379, no. 12 (September 2018), pp. 1139–1149, https://doi.org/10.1056/NEJMsa1803388; J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and Medicare Payment Advisory Commission, “Assessing the Medicare Shared Savings Program’s Effect on Medicare Spending,” Chapter 6 in Report to the Congress: Medicare and the Health Care Delivery System (June 2019), pp. 177–201, https://tinyurl.com/mr37rrjc (PDF).

9. J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and Carrie H. Colla and others, “Hospitals Participating in ACOs Tend to Be Large and Urban, Allowing Access to Capital and Data,” Health Affairs, vol. 35, no. 3 (March 2016), https://doi.org/10.1377/hlthaff.2015.0919.

10. J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv.

11. For more information about advanced primary care and the patient-centered medical home (PCMH) model, see Primary Care Collaborative, “Advanced Primary Care/PCMH,” https://thepcc.org/topic/advanced-primary-carepcmh; and Agency for Healthcare Research and Quality, “Defining the PCMH,” https://tinyurl.com/4pdtnhk9. Many of the studies that examined the role of advanced primary care and PCMH in ACOs assessed the ACOs’ savings compared with benchmark spending and conducted simple correlational analyses (not relative to a counterfactual). For an extensive literature review, see Yalda Jabbarpour and others, Advanced Primary Care: A Key Contributor to Successful ACOs, Milbank Memorial Fund (August 2018), http://tinyurl.com/23p7v9r4. For the correlation between ACOs’ savings relative to a benchmark (and not relative to a counterfactual estimate of ACOs’ spending) and the proportion of PCPs in ACOs, see Centers for Medicare & Medicaid Services, “Medicare Shared Savings Program Saves Medicare More Than $1.6 Billion in 2021 and Continues to Deliver High-Quality Care” (press release, August 30, 2022), http://tinyurl.com/38u9nv3v; and Zach Davis and Elliot McLeRoy, Wakely Risk Insight for the Medicare Shared Savings Program: Performance Year 2022 Results (November 2023), http://tinyurl.com/mue9w4ej (PDF).

12. J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and R. Shawn Martin and Farzad Mostashari, “CMMI and Value-Based Care: Advancing and Safeguarding Primary Care,” Health Affairs Forefront (January 22, 2024), https://doi.org/10.1377/forefront.20240118.316942.

13. CMS released the Pathways to Success final rule for the MSSP in 2018, which allows ACOs renewing after July 1, 2019, to choose between prospective or retrospective patient assignment. Prospective assignment involves assigning beneficiaries on the basis of primary care services that beneficiaries receive in the year before the performance year, whereas retrospective assignment is based on the services that beneficiaries receive during the performance year. For further discussion, see Sam Shellabarger, Charlie Mills, and Lance D. Anderson, Prospective and Retrospective Assignment in MSSP and Beyond, Milliman issue brief (January 2020), http://tinyurl.com/bde8774x.

14. J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and J. Michael McWilliams and others, “Performance Differences in Year 1 of Pioneer Accountable Care Organizations,” New England Journal of Medicine, vol. 372, no. 20 (May 2015), pp. 1927–1936, https://doi.org/10.1056/NEJMsa1414929.

15. Aaron L. Schwartz and others, “Changes in Low-Value Services in Year 1 of the Medicare Pioneer Accountable Care Organization Program,” JAMA Internal Medicine, vol. 175, no. 11 (November 2015), pp. 1815‒1825, https://doi.org/10.1001/jamainternmed.2015.4525.

16. J. Michael McWilliams and others, “Early Performance of Accountable Care Organizations in Medicare,” New England Journal of Medicine, vol. 374, no. 24 (June 2016), pp. 2357–2366, https://doi.org/10.1056/NEJMsa1600142; and J. Michael McWilliams and others, “Performance Differences in Year 1 of Pioneer Accountable Care Organizations,” New England Journal of Medicine, vol. 372 (May 14, 2015), pp. 1927–1936, https://doi.org/10.1056/NEJMsa1414929.

17. One key indication that postacute care is used inefficiently is that spending for it varies widely among geographic areas, but evidence suggests that more spending on that type of care is not consistently related to better health outcomes. Those geographic differences in spending for postacute care were found to be the largest contributor to geographic variation in overall Medicare FFS spending. See National Academies of Sciences, Engineering, and Medicine, Variation in Health Care Spending: Target Decision Making, Not Geography (2013), https://tinyurl.com/272uk333; Rachel M. Werner and others, “The Value of an Additional Day of Post-Acute Care in a Skilled Nursing Facility,” American Journal of Health Economics, vol. 9, no. 1 (Winter 2023), pp. 1–21, https://doi.org/10.1086/721706; and Peter J. Huckfeldt and others, “Postacute Care for Medicare Advantage Enrollees Who Switched to Traditional Medicare Compared With Those Who Remained in Medicare Advantage,” JAMA Health Forum, vol. 5, no. 2 (February 2024), e235325, https://tinyurl.com/yw9jruky. For empirical evidence on the relationship between ACOs and postacute care, see Carrie H. Colla and others, “Changes in Use of Postacute Care Associated With Accountable Care Organizations in Hip Fracture, Stroke, and Pneumonia Hospitalized Cohorts,” Medical Care, vol. 57, no. 6 (June 2019), pp. 444‒452, https://doi.org/10.1097/MLR.0000000000001121. For further discussion, see J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv.

18. For discussion about empirical benchmarks and how they compare with administrative benchmarks (which are not influenced by ACOs’ spending), see Michael E. Chernew, J. Michael McWilliams, and Shivani A. Shah, “The Case for Administrative Benchmarks (and Some Challenges),” NEJM Catalyst, vol. 4, no. 10 (September 2023), https://doi.org/10.1056/CAT.23.0194; Michael E. Chernew, Jermaine Heath, and J. Michael McWilliams, “The Merits of Administrative Benchmarks for Population-Based Payment Programs,” American Journal of Managed Care, vol. 28, no. 7 (July 2022), pp. e239–e243, https://doi.org/10.37765/ajmc.2022.88799; and J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv.

19. J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 1,” Health Affairs Forefront (November 12, 2020), http://tinyurl.com/returpuk.

20. Government Accountability Office, Medicare: Information on the Transition to Alternative Payment Models by Providers in Rural, Health Professional Shortage, or Underserved Areas, GAO-22-104618 (November 17, 2021), www.gao.gov/products/gao-22-104618.

21. Such up-front payments were available to providers in the Advance Payment ACO Model and, beginning on January 1, 2024, as part of a payment option under the MSSP known as Advance Investment Payments. See Centers for Medicare & Medicaid Services, “Medicare Shared Savings Program Advance Investment Payments (AIP),” www.cms.gov/files/document/mssp-aip-glance.pdf, and “Advanced Payment ACO Model,” https://tinyurl.com/2ufytp49.

22. The incentive payments were calculated as 5 percent of the estimated aggregate payments up to the end of performance year 2022. Those payments were set to expire at that time but were extended for one year (through performance year 2023) at a reduced rate of 3.5 percent. Lawmakers have enacted another extension for performance year 2024, but the rate is 1.88 percent. To become a qualifying APM participant and receive incentive payments, during the performance period providers must receive at least 50 percent of their payments under Medicare Part B or see at least 35 percent of their Medicare patients through an advanced APM entity, within which 75 percent of practices must use certified electronic health record technology. For more information, see Centers for Medicare & Medicaid Services, Quality Payment Program, “Advanced Alternative Payment Models (APMs),” https://qpp.cms.gov/apms/advanced-apms, and “Calendar Year (CY) 2024 Medicare Physician Fee Schedule (PFS) Final Rule Fact Sheet and Policy Comparison Table,” https://tinyurl.com/mr3nb3wa (PDF); and Consolidated Appropriations Act, 2024, Public Law 118-42, www.congress.gov/bill/118th-congress/house-bill/4366.

23. Testimony of J. Michael McWilliams, Professor, Harvard Medical School, before the Subcommittee on Oversight and Investigations of the House Committee on Energy and Commerce, MACRA Checkup: Assessing Implementation and Challenges That Remain for Patients and Doctors (June 22, 2023), http://tinyurl.com/5n8zasdc (PDF).

24. Advanced APM Criteria, C.F.R. §414.1415, https://tinyurl.com/yhfavy9y.

25. Xi Zhu and others, “Medicare Accountable Care Organization Characteristics Associated With Participation in 2‐Sided Risk,” Journal of Rural Health, vol. 39, no. 1 (Winter 2023), pp. 302–308, https://doi.org/10.1111/jrh.12672.

26. Peter F. Lyu, Michael E. Chernew, and J. Michael McWilliams, “Benchmarking Changes and Selective Participation in the Medicare Shared Savings Program,” Health Affairs, vol. 42, no. 5 (May 2023), pp. 622–631, https://doi.org/10.1377/hlthaff.2022.01061.

27. Carrie H. Colla, Toyin Ajayi, and Asaf Bitton, “Potential Adverse Financial Implications of the Merit-Based Incentive Payment System for Independent and Safety Net Practices,” JAMA, vol. 324, no. 10 (September 2020), pp. 948–950, https://doi.org/10.1001/jama.2020.13093; and Sunny C. Lin and others, “Exit Rates of Accountable Care Organizations That Serve High Proportions of Beneficiaries of Racial and Ethnic Minority Groups,” JAMA Health Forum, vol. 3, no. 9 (September 2022), e223398, https://doi.org/10.1001/jamahealthforum.2022.3398.

28. Downside risk may, however, strengthen incentives for providers in ACOs to reduce health care spending. The empirical evidence about whether an ACO’s downside risk accelerates savings in the MSSP is lacking, and one reason could be providers’ selective participation. For example, ACO providers that entered into MSSP contracts with downside risk had lower baseline spending than their benchmarks even before entering into such contracts. That finding suggests that those ACOs entered into downside risk agreements knowing that they were likely to achieve shared savings. For further discussion, see J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 1,” Health Affairs Forefront (November 12, 2020), http://tinyurl.com/returpuk; and J. Michael McWilliams and others, “Getting More Savings From ACOs—Can the Pace Be Pushed?” New England Journal of Medicine, vol. 380, no. 23 (June 2019), pp. 2190–2192, https://doi.org/10.1056/NEJMp1900537.

29. CMS released the Pathways to Success final rule for the MSSP in 2018, which accelerated the transition to contracts with downside risk for participating ACOs. For further discussion, see Centers for Medicare & Medicaid Services, “Final Rule Creates Pathways to Success for the Medicare Shared Savings Program” (December 21, 2018), https://tinyurl.com/mpnzpmk5; and Meiling Ying and others, “Changes in Shared Savings Program Participation After Launch of Pathways to Success,” Journal of General Internal Medicine, vol. 38, no. 7 (May 2023), pp. 1780–1782, https://doi.org/10.1007/s11606-022-07947-x.

30. Providers’ ability to avoid or selectively refer patients who have higher spending than what the risk-adjustment model would predict also might explain some observed savings. For more information, see Mariétou H. Ouayogodé and others, “Estimates of ACO Savings in the Presence of Provider and Beneficiary Selection,” Healthcare, vol. 9, no. 1 (March 2021), https://tinyurl.com/ycydknp6; Adam A. Markovitz and others, “Performance in the Medicare Shared Savings Program After Accounting for Nonrandom Exit: An Instrumental Variable Analysis,” Annals of Internal Medicine, vol. 171, no. 1 (July 2019), pp. 27–36, https://doi.org/10.7326/M18-2539; and J. Michael McWilliams and others, “Savings or Selection? Initial Spending Reductions in the Medicare Shared Savings Program and Considerations for Reform,” Milbank Quarterly, vol. 98, no. 3 (September 2020), pp. 847–907, https://doi.org/10.1111/1468-0009.12468.

31. ACO programs could be designed in ways that made mandatory participation more likely to achieve net savings for Medicare, but any actual savings would depend on the specific details of the legislation and other factors. For information about a mandatory APM, see Centers for Medicare & Medicaid Services, CMS Comprehensive Care for Joint Replacement Model: Performance Year 5 Evaluation Report (report prepared by the Lewin Group and others, April 2023), http://tinyurl.com/4s5enys3; and Congressional Budget Office, Federal Budgetary Effects of the Activities of the Center for Medicare & Medicaid Innovation (September 2023), www.cbo.gov/publication/59274.

32. J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 1,” Health Affairs Forefront (November 12, 2020), http://tinyurl.com/returpuk, and “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 2,” Health Affairs Forefront (November 13, 2020), http://tinyurl.com/ykrr4wcr.

33. The incentive payment could be changed by restructuring it as a flat per-beneficiary sum and expanding eligibility to include ACO contracts without downside risk. For further discussion, see the testimony of J. Michael McWilliams, Professor, Harvard Medical School, before the Subcommittee on Oversight and Investigations of the House Committee on Energy and Commerce, MACRA Checkup: Assessing Implementation and Challenges That Remain for Patients and Doctors (June 22, 2023), http://tinyurl.com/5n8zasdc; J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 2,” Health Affairs Forefront (November 13, 2020), http://tinyurl.com/ykrr4wcr; J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv; and Value in Health Care Act of 2023, H.R. 5013, 118th Cong., www.congress.gov/bill/118th-congress/house-bill/5013.

34. Given the choice between joining an ACO and an alternative of either receiving site-neutral payments or being excluded from the 340B drug-pricing program (or both), hospitals would be more likely to join an ACO as long as the anticipated losses from doing so were smaller than the alternative reduction in profits that would result from the site-neutral payments or 340B exclusion. (Site-neutral payments would be applied to hospital outpatient facilities for those hospitals that decided not to participate in an ACO.) Under the Bipartisan Budget Act of 2018, certain MSSP ACOs already have greater flexibility to use telehealth services than what is permitted in the FFS program. (The FFS program’s flexibility to use telehealth has been expanded since the start of the coronavirus pandemic, but under current law, many of those flexibilities will expire at the end of calendar year 2024.) For further discussion, see J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv; and Centers for Medicare & Medicaid Services, “Telehealth Factsheet,” http://tinyurl.com/2r2nsu78 (PDF).

35. On March 19, 2024, CMS announced a new ACO model, the ACO Primary Care Flex Model, aimed at improving person-centered primary care, particularly for low-revenue ACOs. The voluntary model, which will operate within the MSSP, is slated to start on January 1, 2025, and to run for five years. It will include monthly prospective primary care payments, thereby shifting payment away from a fee-for-service basis for primary care and enhancing the predictability of revenue for low-revenue ACOs. Additionally, the model offers an Advanced Shared Savings Payment to assist with ACO formation and administrative costs. For more information, see Centers for Medicare & Medicaid Services, “ACO Primary Care Flex Model,” https://tinyurl.com/bdzh6xra.

36. J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv; NAACOS, “A New Approach to Paying for Primary Care in the Medicare Shared Savings Program” (April 13, 2023), http://tinyurl.com/3sby8tjy; and National Academies of Sciences, Engineering, and Medicine, “Payment to Support High-Quality Primary Care,” in Implementing High-Quality Primary Care: Rebuilding the Foundation of Health Care (2021), https://nap.nationalacademies.org/read/25983/chapter/12.

37. Michael E. Chernew, J. Michael McWilliams, and Shivani A. Shah, “The Case for Administrative Benchmarks (and Some Challenges),” NEJM Catalyst, vol. 4, no. 10 (September 2023), https://doi.org/10.1056/CAT.23.0194; and Centers for Medicare & Medicaid Services, “Calendar Year (CY) 2023 Medicare Physician Fee Schedule Final Rule—Medicare Shared Savings Program” (fact sheet, November 1, 2022), http://tinyurl.com/57ytzbff.

38. Centers for Medicare & Medicaid Services, “Calendar Year (CY) 2023 Medicare Physician Fee Schedule Final Rule—Medicare Shared Savings Program” (fact sheet, November 1, 2022), http://tinyurl.com/57ytzbff.

39. Setting the administrative growth factor at the rate of growth for gross domestic product is analogous to how global hospital budget growth was set in Maryland under the Maryland All-Payer Model, in which the annual growth of per capita hospital spending for all payers was limited to the historical growth rate of the gross state product. Per capita hospital spending growth in Maryland continues to be limited to that rate under the Total Cost of Care Model, which started in 2019 following the conclusion of the All-Payer Model. For further discussion, see Centers for Medicare & Medicaid Services, “Maryland All-Payer Model to Deliver Better Care and Lower Costs” (fact sheet, January 10, 2014), https://tinyurl.com/ytwcsh3d, and “Maryland Total Cost of Care Model,” https://tinyurl.com/bdhs68h4; Michael E. Chernew, J. Michael McWilliams, and Shivani A. Shah, “The Case for Administrative Benchmarks (and Some Challenges),” NEJM Catalyst, vol. 4, no. 10 (September 2023), https://doi.org/10.1056/CAT.23.0194; J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 2,” Health Affairs Forefront (November 13, 2020), http://tinyurl.com/ykrr4wcr; and Michael E. Chernew, Jermaine Heath, and J. Michael McWilliams, “The Merits of Administrative Benchmarks for Population-Based Payment Programs,” American Journal of Managed Care, vol. 28, no. 7 (July 2022), pp. e239–e243, https://doi.org/10.37765/ajmc.2022.88799.

40. J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 1,” Health Affairs Forefront (November 12, 2020), http://tinyurl.com/returpuk, and “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 2,” Health Affairs Forefront (November 13, 2020), http://tinyurl.com/ykrr4wcr.

41. Spillovers also could occur because ACO-generated savings would lower Medicare FFS spending, and lower FFS spending would reduce Medicare Advantage benchmarks and the average payments made to those plans. For more information, see J. Michael McWilliams, Bruce E. Landon, and Michael E. Chernew, “Changes in Health Care Spending and Quality for Medicare Beneficiaries Associated With a Commercial ACO Contract,” JAMA, vol. 310, no. 8 (August 2013), pp. 829–836, http://tinyurl.com/4ncpuv8e; J. Michael McWilliams and Alice J. Chen, “Understanding the Latest ACO “Savings”: Curb Your Enthusiasm and Sharpen Your Pencils—Part 2,” Health Affairs Forefront (November 13, 2020), http://tinyurl.com/ykrr4wcr; and J. Michael McWilliams, Alice Chen, and Michael E. Chernew, From Vision to Design in Advancing Medicare Payment Reform: A Blueprint for Population-Based Payments, USC‒Brookings Schaeffer Initiative for Health Policy (October 2021), http://tinyurl.com/2s3mhwpv.

42. Centers for Medicare & Medicaid Services, Medicare Shared Savings Program: Skilled Nursing Facility 3-Day Rule Waiver: Guidance, version 10 (May 2022), https://tinyurl.com/c6s7w5t7, and “Telehealth Factsheet,” https://tinyurl.com/3rk34n4c.

43. Under the MSSP, beneficiaries must be notified that they are part of an ACO before or at their first PCP visit during the ACO’s contract period. In addition, ACOs are required to have a follow-up communication with beneficiaries within 180 days of the initial notification. However, some entities have raised concerns that the form letter provided by CMS may not be clear enough for beneficiaries to understand how an ACO operates and to ascertain the potential benefits they may gain from receiving services from ACO providers. For more information about the current rule, see Centers for Medicare & Medicaid Services, “Program Guidance & Specifications,” https://tinyurl.com/2p2r46wu. For further discussion about proposed changes to notifications, see Clif Gaus, President and CEO, NAACOS, letter to Meena Seshamani, Deputy Administrator and Director, Center for Medicare, Centers for Medicare & Medicaid Services (July 7, 2022), www.naacos.com/improving-the-beneficiary-notification-letter; and Rosha Champion McCoy, Acting Chief Health Care Officer, Association of American Medical Colleges, to the Honorable Chiquita Brooks-LaSure, Administrator, Centers for Medicare & Medicaid Services (September 6, 2022), www.aamc.org/media/62416/download?attachment (PDF).

44. For example, some providers that participated in the Next Generation ACO or Direct Contracting Model were allowed to reduce or eliminate cost-sharing requirements for certain Medicare Part B services for their assigned beneficiaries. By reducing those financial barriers, providers might have enabled some beneficiaries to more closely adhere to their treatment plans. See Centers for Medicare & Medicaid Services, “Cost Sharing Support for Part B Services Benefit Enhancement,” http://tinyurl.com/mw3bt963, and “Global & Professional Direct Contracting Model,” http://tinyurl.com/yck5z2ez.

45 Currently, CMS allows ACOs in the MSSP with downside-risk contracts to provide a payment of up to $20 to each assigned beneficiary for each qualifying primary care service that the beneficiary receives as an incentive to maintain good health. Some researchers have suggested expanding such incentive payments beyond those related to primary care. See Centers for Medicare & Medicaid Services, “Program Guidance & Specifications,” https://tinyurl.com/2p2r46wu; and Lindsey A. Herrel and others, “Association of Medicare Beneficiary and Hospital Accountable Care Organization Alignment With Surgical Cost Savings,” JAMA Health Forum, vol. 3, no. 12 (December 2022), e224817, https://doi.org/10.1001/jamahealthforum.2022.4817.

At the request of the Chairman of the Senate Budget Committee, the Congressional Budget Office reviewed the research literature on Medicare accountable care organizations (ACOs), conducted interviews with experts to enhance its understanding of the research findings, identified proposed policy approaches that could increase savings from Medicare ACOs, and prepared this report. In keeping with CBO’s mandate to provide objective, impartial analysis, the report makes no recommendations.

Michael Cohen, Grace Hwang, and Jared Maeda prepared the report with contributions from Joyce Shin and with guidance from Berna Demiralp, Tamara Hayford, and Chapin White. Carrie H. Colla of the Dartmouth Institute for Health Policy and Clinical Practice offered comments. The assistance of an outside reviewer implies no responsibility for the final product; that responsibility rests solely with CBO.

Jeffrey Kling and Robert Sunshine reviewed the report, Christine Bogusz edited it, and R. L. Rebach prepared it for publication. Joyce Shin and Joshua Varcie fact-checked the report. It is available at www.cbo.gov/publication/59879.

CBO seeks feedback to make its work as useful as possible. Please send comments to communications@cbo.gov.

Phillip L. Swagel

Director