Introduction and summary
Covering more than 60 million seniors and disabled people, Medicare is the second-largest health insurance program in the United States, after Medicaid.1 It pays for millions of patients’ care throughout the country and influences millions of providers’ business and clinical decisions. Given Medicare’s size and role as a public program, reforms to improve its value have tremendous potential to improve care quality and health outcomes throughout the country as well as to generate savings for beneficiaries and taxpayers.
These changes offer promising avenues for improving health care quality, reducing expenditures, or both—all while preserving access to care and maintaining value for Medicare beneficiaries.
Moreover, improvements to Medicare can have positive spillovers into other parts of the health care system. Medicare’s reimbursement rates have been shown to influence the rates that private insurers pay.2 Actions by the federal Centers for Medicare and Medicaid Services (CMS) serve as a powerful catalyst for change throughout the health care system. Policy changes to Medicare payment and participation rules often spur adoption by private payers and nonparticipating providers as well.3 For example, private insurers adopted several value-based payment programs that CMS pioneered and followed CMS in setting goals for the level of care provided under value-based payment.4
This report examines previous reforms to tie Medicare payment more closely to value, including the Medicare Shared Savings Program and models implemented through the Center for Medicare and Medicaid Innovation (CMMI). It then recommends several policies aimed at reshaping the health care delivery system and leveraging competition in Medicare, including:
- Sharpen the focus of CMMI demonstrations.
- Make participation mandatory in CMMI models.
- Expand competitive bidding for medical equipment and other supplies.
- Halt the rise of coding intensity in Medicare Advantage.
- Introduce competitive bidding for Medicare Advantage plans.
The report also outlines policy options to address the misvaluing of several kinds of care that Medicare funds, including reforming value determinations for Medicare services, reducing bad debt payments, improving the accuracy of post-acute care provider rates, and implementing site-neutral payments.
These changes offer promising avenues for improving health care quality, reducing expenditures, or both—all while preserving access to care and maintaining value for Medicare beneficiaries.
Previous Medicare delivery system reform efforts
Medicare’s traditional fee-for-service (FFS) reimbursement model pays health care providers for specific items and services furnished to patients. This structure rewards higher volume, regardless of the quality of the services provided. In the past 15 years, Medicare has been moving toward paying for a greater amount of beneficiary care using alternative payment models (APMs), shifting from fee-for-service toward value-based care payment. Under APM arrangements, payments reflect the quality of care provided to patients, rather than the pure volume of services provided,5 thereby incentivizing providers to strive for high-quality and cost-efficient care.6 These APMs can apply to specific health conditions, episodes of care, provider types, geographic communities, or programs within Medicare.7
Currently, the main institution driving development of new APMs in Medicare is the Center for Medicare and Medicaid Innovation (CMMI).8 CMMI was established in 2010 by the Affordable Care Act (ACA). While the Medicare statute had previously included some authority to test new payment approaches through demonstration programs, the ACA gave CMS more tools and funding to do so, as well as the ability to expand successful demonstrations without congressional action.9 In addition, CMMI has the authority to develop and test additional payment models, which gives the agency the flexibility to create new ways to pay for care and to refine models in response to the demonstrations’ results.10
Some of CMMI’s model demonstrations are conducted on a voluntary basis, meaning that providers decide for themselves whether to participate in a project.11 Other demonstrations are mandatory, with CMMI randomly assigning providers to participate.12 All demonstrations are evaluated for their impact on quality and cost. If CMS determines that a model either reduces spending without reducing the quality of care or improves the quality of care without increasing spending, the secretary of health and human services (HHS) is authorized to expand the project’s scope or duration, including expanding it nationwide.13
Medicare has tested dozens of models designed to improve the value of care provided to beneficiaries, both through CMMI and its pre-ACA authority. There is great variety in these models’ approaches; some, such as accountable care organizations and patient-centered medical homes, seek to coordinate care, and others, including bundled payment models, focus on disincentivizing unnecessary care. However, all of these models seek to revolutionize the way care is both delivered and paid for, with the ultimate goal of providing high-quality care at a lower cost to Medicare. Some APMs, such as the Pioneer Accountable Care Organization (ACO) model, have proved successful in increasing value, while others, such as the Bundled Payments for Care Improvement (BPCI) initiative, have failed to generate savings.
Accountable care organizations
Many of the delivery system reforms that Medicare has adopted have been through accountable care organizations (ACOs), groups of health care providers who share responsibility for providing low-cost, high-quality care to their patients.14 As an incentive for this shared responsibility, the physicians, hospitals, and other providers involved also share the savings accrued with the Medicare program. By coordinating care across providers and settings, patients ideally receive more appropriate care while avoiding unnecessary or redundant services. This coordination also helps reduce hospital readmissions and connect patients with nonmedical services such as housing, transportation, and food security.15
By coordinating care across providers and settings, patients ideally receive more appropriate care while avoiding unnecessary or redundant services. This coordination also helps reduce hospital readmissions and connect patients with nonmedical services such as housing, transportation, and food security.
Medicare has operated multiple ACO programs in the past decade: the Medicare Shared Savings Program (MSSP), the Pioneer ACO model, the Next Generation ACO model, and the ACO Realizing Equity, Access, and Community Health (REACH) model. Other ACO models have been tested among specific disease groups and health systems.16
The MSSP is a voluntary demonstration project established by the Affordable Care Act. As of January 1, 2022, 483 ACOs participate, serving 11 million Medicare beneficiaries.17 The program has been successful: It saved Medicare nearly $2 billion in 2020 and $1.6 billion in 2021, marking the fifth year in a row the program generated savings.18 The Pioneer ACO model—another ACO demonstration—was operated through CMMI from 2012 to 2016.19 The program was designed to test alternative ACO designs to help improve the MSSP, and participants showed significant improvement in quality scores between the beginning and end of the program.20
Another ACO model that CMMI operated, the Next Generation ACO model, advanced the Pioneer model by allowing ACOs to take on higher levels of risk and reward than the MSSP allowed. It operated from 2016 through the end of 2021 after being extended a year due to the COVID-19 pandemic.21 Thirty-five ACOs participated in the fourth year of the program, although the number of participating ACOs dropped to nine in the final year.22 Exiting ACOs over the years offered a number of explanations, with many pointing to unachievable savings metrics.23 The model reduced Medicare Parts A and B spending by about $670 million across the first four years of operation.24 After accounting for incentive payments, however, the program resulted in about $240 million more in spending.25 This reduction in spending was concentrated among patients with several chronic illnesses and patients who were hospitalized within the year prior, but importantly, it was not associated with a change in the quality of care that patients received.26
The ACO REACH model, set to begin January 1, 2023, and run through 2026, redesigns the Global and Professional Direct Contracting (GPDC) model to better prioritize underserved populations, advance health equity, encourage provider governance, and protect beneficiaries through greater transparency.27 CMMI has provisionally accepted 110 participants into the program.28
Bundled payment models
Medicare has also recently pursued another set of APMs called bundled payment models. Bundled payments are fixed amounts paid to health care providers for the services a patient is expected to need during an episode of care.29 These models are also intended to disincentivize unnecessary care and encourage providers to ensure patients receive proper follow-up care. Bundled payments can be prospective, in which payment is made for the expected cost of care delivery, or retrospective, in which costs are measured against a predetermined rate and payments are adjusted based on adherence to this rate.30 CMMI has tested and planned several bundled payment models, including for hospital-based acute care, joint replacement, and cancer treatment.
The Bundled Payments for Care Improvement (BPCI) initiative was a combination of four bundled payment models that operated from 2013 to 2018.31 Two of these models were prospective, focusing on acute care in hospitals.32 The other two were retrospective bundles that worked to address post-acute care. Both prospective models achieved net reductions in Medicare spending.33 While the retrospective models did not achieve overall savings due to higher-than-expected spending on provider incentive payments spending on health care services still declined and none of the models had a significant impact on quality.34 BCPI Advanced is an ongoing retrospective bundled payment demonstration.35 A recent evaluation of the program found that the model reduced nonstandardized payments by more than $550 million between October 1, 2018, and December 31, 2019.36 Although the model resulted in an estimated net loss of $65.7 million to Medicare, the program did generate net savings on surgical clinical episodes during this time period.37
The Comprehensive Care for Joint Replacement (CJR) model is a hip and knee replacement bundled payment demonstration. It pays for joint replacement episodes, beginning with hospital admission and ending 90 days after discharge. 38 Notably, CJR was the first CMMI demonstration with mandatory participation by hospitals, although the mandatory component was scaled back by the Trump administration after two years.39 CJR was retrospective, comparing spending to benchmarks and rewarding providers with a share of the savings if they were lower than the benchmark level. After accounting for these reward payments, the program saved Medicare an estimated $76 million over four years—about 1.82 percent of the baseline estimate for spending on these services during that same period.40 As a result of net savings in the first three years—totaling $61.6 million—attributed primarily to lower utilization of post-acute care,41 CMS issued a final rule in April 2021 extending the model through the end of 2024.42
CMMI has also developed two cancer-focused models, the Oncology Care Model (OCM) and the Radiation Oncology Model. As with the other bundled payment models, the OCM demonstration pays providers a portion of the savings achieved through better care for chemotherapy patients.43 Additionally, it provides participants with a flat monthly payment to offset some of the administrative costs associated with coordinating their patients’ care.44 A recent analysis of OCM outcomes through December 2019 found that the program drove use of higher-value support drugs and significantly reduced payments for four cancer types, but net spending rose relative to the model’s benchmark.45 The Radiation Oncology Model is an upcoming demonstration designed to improve the quality of radiation therapy for cancer.46 The model is a prospective bundle and, importantly, a mandatory demonstration.47 Although a radiation therapy model was required as part of the 2015 Patient Access and Medicare Protection Act, CMS finalized a rule on August 29, 2022, that indefinitely postpones implementation of the demonstration following delays from Congress.48
Patient-centered medical homes
CMS has also conducted a variety of patient-centered medical home (PCMH) demonstrations. PCMHs are redesigned primary care practices that focus on preventive care, patient education, and care coordination.49 Under a PCMH approach, primary care providers coordinate the care a patient receives from other specialty providers to avoid duplicative services, as their other providers are aware of what care a patient is receiving or has received. These models have some of the highest numbers of participants among Medicare demonstrations.
In the Comprehensive Primary Care (CPC) model, CMMI paid PCMHs a monthly fee to defray the cost of developing their care management practices and staff, but it also allowed participating practices to share in the savings accrued to Medicare.50 The program was intended to improve the quality of care delivered and lower costs as well as to improve providers’ experiences.51 An evaluation of the program by Mathematica Policy Research found that it reduced emergency department visits among patients at participating practices, especially in the later years of the program.52 The same analysis found that health spending for participants also grew more slowly than it did for nonparticipants but that these savings were offset by the management fee paid to providers.53
The CPC model also served as the foundation for the Comprehensive Primary Care Plus (CPC+) model, which began in 2017—after the initial CPC demonstration ended—and ran through December 2021.54 A recent evaluation, also by Mathematica Policy Research, found that over the first four years of the program, CPC+ reduced acute hospitalizations and expenditures on acute inpatient care.55 Despite this, however, the report found that these reductions were offset by increases in expenditures on other services, meaning the program had no significant impact on Medicare spending.56 According to the same report, the model also failed to demonstrably improve the quality of care, although practices consistently reported that their participation improved the quality of the care they delivered to their patients.57
Although a number of the models examined above failed to deliver savings or improve care, with some racking up high costs to the Medicare program, APMs are still a critical tool in CMMI’s efforts to transform care delivery. A recent evaluation conducted by Avalere found that while CMMI models generated $1.7 billion in Medicare losses from 2017 to 2021, continuing and expanding existing successful models—including BCPI Advanced and CJR—will create $2.5 billion in net savings from 2022 to 2026.58 With a decade’s worth of lessons to draw from, CMMI now has the opportunity to implement reforms that will draw on past successes and avoid the pitfalls of less successful programs.
Policy options to advance delivery system reform
The role of CMMI is to develop and identify delivery system changes that reduce costs and improve quality.59 Accordingly, a long-run goal of CMS should be to scale up successful interventions to maximize savings and benefits for patients and taxpayers. With that goal in mind, demonstrations at CMMI should focus on generating results to support potential large-scale implementation—specifically by implementing a smaller number of large-scale demonstrations and making participation mandatory.
Sharpen the focus of CMMI demonstrations
To date, CMMI has implemented more than 50 payment models.60 While the proliferation of APMs is an encouraging sign of enthusiasm for reform, it also presents challenges. As the size and number of demonstrations and other APMs in effect increases, providers become more likely to participate in multiple APMs, and beneficiaries’ care may be delivered through multiple APMs. Overlapping models can complicate payments and make it difficult for evaluators to attribute changes in outcomes to specific interventions.61 At the same time, larger demonstrations can be more informative because their results are more likely to be generalizable, rather than particular to the set of providers participating in the project. Ideally, demonstration projects should involve a diversity of providers and beneficiaries, spanning both rural and urban areas and including patients from a variety of socioeconomic and racial/ethnic backgrounds.
To balance these trade-offs, CMS should consider a more purposeful approach to CMMI, concentrating efforts on a smaller number of large-scale models. Such a change, according to the Medicare Payment Advisory Commission (MedPAC), would reduce the “complexity and uncertainty that providers face when deciding whether to participate in an APM” and would thus increase both provider participation and performance.62 Luckily, a revamp of this nature appears to be already underway. The current leadership of CMS—Chiquita Brooks-LaSure, Elizabeth Fowler, Meena Seshamani, and Daniel Tsai—laid out their vision for payment innovation in a 2021 Health Affairs blog post.63 They articulated six goals: centering equity, reducing the number of concurrent models, reexamining the role of financial incentives for participation, supporting providers in taking downside risk, improving benchmarking to avoid overpayment, and focusing on encouraging long-term “transformation” and “quality investments.”
The Biden administration’s vision echoes recent advice from past CMS leaders. In 2021, former CMS Administrator Donald Berwick and former CMMI Director Rick Gilfillan argued in the Journal of the American Medical Association that “models that CMMI tests should link tightly to a shared vision” of improvements to health and the health care delivery system and that CMMI should support models aimed at improving health equity. MedPAC has also consistently urged CMS to move away from temporarily testing many small-scale models, unanimously agreeing in 2021 that CMS must “implement a more harmonized portfolio of fewer alternative payment models that are designed to work together to support the strategic objectives of reducing spending and improving quality.”64 In its June 2022 report, MedPAC provided specific suggestions on how to streamline and harmonize APMs—including reducing the number of population-based model tracks, moving away from the regular benchmark rebasing using actual spending that penalizes providers successfully lowering spending, and integrating episode-based and population-based APMs to maximize the quality of care provided to Medicare beneficiaries.65
Make CMMI model participation mandatory
In addition to the more intentional approach to models described above, CMMI should shift demonstrations toward mandatory participation. Mandatory models have been used by every administration since the creation of CMMI to test policy interventions,66 and they are essential to determining the true impact of payment changes. Voluntary demonstration projects are more prone to selection bias, with outcomes confounded by the types of providers, geographic areas, or patient populations involved in the model.67
Taking a more intentional approach with models would improve provider participation and performance, and greater mandatory participation in demonstrations would improve model evaluation.
Foremost among these concerns about selection is that providers with healthier patient mixes would participate in a program at a higher rate than those with less healthy patients. Because these patients represent a lower risk of having a poor outcome than less healthy patients, their providers are incentivized to enroll in the program, especially if it is a shared savings program. Voluntary models also run the risk of drawing participation rates too low to properly evaluate the impact of a demonstration.68
The reforms suggested above would represent a sea change in CMMI policy, with great potential to improve the quality of models. Taking a more intentional approach with models would improve provider participation and performance, and greater mandatory participation in demonstrations would improve model evaluation.
Combined, these reforms would help maximize the cost-saving potential of demonstration models, furthering CMMI’s goal of achieving Medicare savings. Additionally, they would ensure greater diversity among participants and help CMS identify policies that can advance health equity. These outcomes align with CMMI’s strategic objective to build a health system that achieves equitable outcomes through high quality care.69
Policy options to improve competition and reduce wasteful spending
In addition to reforming how Medicare pays health care providers, improving competition within the program and eliminating waste can help to ensure that patients are served well and that tax dollars are spent efficiently. There are numerous policy options that could create greater competition in Medicare, including expanding the existing competitive bidding program for durable medical equipment, halting the rise of coding intensity in Medicare Advantage, and introducing competitive bidding in Medicare Advantage.
Expand competitive bidding for medical equipment and supplies
As required by the Medicare Modernization Act of 2003, CMS implemented a competitive bidding process to set rates for durable medical equipment, prosthetics, orthotics, and supplies (DMEPOS).70 The program was first implemented in 2011 and expanded in 2013.71 A claims analysis conducted by CMS in the first year of the program found no change in beneficiary health status outcomes, highlighting that competitive bidding can achieve savings without harming care.72
Another analysis of the competitive bidding program found a similar result. Researchers at Stanford and Northwestern examined DMEPOS claims data from 2009 to 2015, evaluating both spending and clinical benefit to patients.73 These researchers found that spending on DMEPOS included in the program fell by more than 47 percent and that this effect persisted throughout the program’s operation.74 After examining the financial impact, the researchers focused on continuous positive airway pressure (CPAP) machines to determine whether the program reduced patient access to needed DMEPOS. They found that the competitive bidding program reduced the price of the machines by 45 percent and lowered utilization by about 4 percent.75 Importantly, however, use of CPAP machines by patients with a sleep apnea diagnosis was “virtually unchanged,” with the majority of the reduction in use concentrated among patients who benefited less from the machines.76 A more recent study evaluating the impact of the competitive bidding program on spending, utilization, and adverse health events from 2009 to 2016 similarly found that the program lowered Medicare expenditures without reducing access by beneficiaries or increasing adverse outcomes.77 These studies show that the competitive bidding program has been successful at utilizing market forces to reduce costs while preserving the quality of care.
To further maximize Medicare savings, the federal government should expand the competitive bidding program to include, at minimum, additional DMEPOS products and laboratory tests. A June 2018 report by MedPAC found that Medicare expenditures for DMEPOS products not included in the bidding program increased by 23 percent from 2010 to 2015, as compared with a 42 percent decrease in expenditures among products included in the program.78 Payment rates for the top 10 DMEPOS products not subject to competitive bidding were, on average, one-third higher than those of private insurers.79
Currently, CMS has the authority to include high-cost products such as chest wall oscillation devices, ventilators, and off-the-shelf orthotics in its competitive bidding program.80 Among the most expensive noncompetitively bid DMEPOS products, these three categories alone made up more than $800 million in Medicaid expenditures in 2015.81 An expansion to all prosthetics and orthotics, plus certain other supplies proposed in the HHS budget for fiscal year 2017, was estimated to save $3.8 billion over a decade.82 CMMI should take care to ensure that access is not hindered by such an expansion. For example, excluding custom prosthetics would ensure that patients who need them retain access but would likely result in lower overall savings.
Halt the rise of coding intensity in Medicare Advantage plans
Eliminating the prevalence of upcoding within Medicare Advantage plans is another reform that would yield cost savings. Upcoding refers to the practice of making beneficiaries appear less healthy than they actually are by adding diagnoses, coding more severe diagnoses, or fraudulently altering records.83 Because the payments that Medicare Advantage plans receive from the federal government per enrollee are risk-adjusted based on enrollees’ health, plans are incentivized to manipulate how healthy their enrollees appear by coding as many diagnoses as possible.84
Multiple studies have found evidence of upcoding in Medicare Advantage plans. A 2020 report by MedPAC found that in 2018, Medicare Advantage enrollees’ risk scores were 8 percent higher than similar patients enrolled in traditional Medicare.85 Even after adjusting these risk scores to make them more consistent with traditional enrollees, MedPAC found that Medicare Advantage risk scores and associated payments were about 2 to 3 percent higher than they would have otherwise been.86 This trend continued in 2020, with MedPAC estimating that Medicare Advantage risk scores were 9.5 percent higher (3.6 percent with CMS coding adjustment) than those of traditional enrollees, which generated approximately $12 billion in excess spending on Medicare Advantage plans.87 A 2019 study by researchers at the University of Minnesota found a similar result: significantly higher risk scores for Medicare Advantage patients, driven by greater diagnostic code intensity.88
These small changes in how beneficiaries’ diagnoses are recorded add up to billions in wasteful Medicare expenditures. According to a 2017 study in Health Affairs, increases in coding intensity could raise Medicare expenditures by $200 billion over a decade,89 while a separate study found that upcoding results in excess payment of $2 billion per year.90 In addition to overcharging the government, these plans can also deny patients, particularly older adults, access to vital medical care.91
According to a 2017 study in Health Affairs, increases in coding intensity could raise Medicare expenditures by $200 billion over a decade, while a separate study found that upcoding results in excess payment of $2 billion per year. In addition to overcharging the government, these plans can also deny patients, particularly older adults, access to vital medical care.
One approach that would reduce Medicare Advantage plans’ ability to upcode their beneficiaries is to exclude diagnoses created during health risks assessments (HRAs), which are often used by Medicare Advantage plans to search for additional diagnoses that can be used to artificially inflate enrollees’ risk scores.92 A recent study by the HHS Office of Inspector General found that several Medicare Advantage plans were leveraging HRAs to increase beneficiaries’ risk scores without providing any care for the diagnoses found during the assessments.93 Removing these assessments from how risk scores are calculated would remove the incentive for Medicare Advantage plans to assign multiple diagnoses to patients without ever intending to treat them for these additional health conditions. Plans would still be able to conduct these assessments, but they would be incentivized to use them for their intended purpose—improving the quality of care provided to patients.
Another way to reduce the impact of upcoding is to perform regular reviews of Medicare Advantage plans’ use of HRAs, as suggested by the inspector general’s report. By evaluating the patient outcomes associated with HRA diagnoses, CMS would be able to determine whether plans are abusing the practice to receive greater federal funds.94 When a plan has overused HRAs to drive up its enrollees’ risk scores, CMS should subject them to more frequent reviews and lower payments. Taking this approach would also incentivize proper use of HRAs, as plans’ payments would be at risk if they are found to have upcoded their patients.
Addressing upcoding has the potential to save the federal government billions of dollars a year. In its 2018 report on policy options to reduce the deficit, the Congressional Budget Office (CBO) estimated that increasing the downward adjustment to Medicare Advantage risk scores to correct for the differences in coding intensity relative to fee-for-service (FFS) Medicare would save the federal government $47 billion over a decade.95 CBO also examined a separate option, which would base risk scores on two years’ worth of FFS diagnostic data and exclude diagnoses captured during health risk assessments; this two-pronged reform to risk scores would reduce federal spending by more than $67 billion over 10 years.96
Introduce competitive bidding for Medicare Advantage plans
In 2022, more than 28 million people received coverage through a Medicare Advantage plan, accounting for 48 percent of all Medicare beneficiaries, and enrollment in Medicare Advantage continues to grow.97 One reform that would generate billions annually in federal savings is bringing Medicare Advantage plan payments in better alignment with the actual cost of benefits. Currently, plans are paid based on their bid relative to a benchmark set administratively by CMS that represents the cost of providing Part A and B care under Medicare FFS.98 Plans with bids below the benchmark receive rebates that they are required to use to provide supplemental benefits to enrollees, which can include reducing cost-sharing or premiums.99 In 2022, average Medicare Advantage rebates reached an all-time high of $164 per enrollee per month.100
By contrast, under a competitive bidding process, a plan’s payment would be based on the bids of other plans, rather than the benchmark. In practice, competitive bidding could be implemented in a variety of ways, such as pegging payment to the average bid or the median bid. While the exact savings from competitive bidding would depend on the design of the program, the Committee for a Responsible Federal Budget estimates that paying Medicare Advantage plans via competitive bidding could save the federal government between $30 billion and $100 billion over 10 years.101 A Brookings Institution working paper from 2018 estimates savings of a similar magnitude, finding that reforms to standardize Medicare Advantage plan design and tie payment to competitive bids would save $10 billion annually.102
A variation proposed by HHS in the president’s budget for FY 2017 would have introduced competitive bidding without eliminating rebates completely.103 It would pay plans the lesser of the FFS-based, administratively set benchmark or the average bid, plus a 5 percent buffer to partially protect rebates. That change, combined with reforms to the bonus payments available to Medicare Advantage plans, is estimated to save $77.2 billion over a decade.104
The reforms explored above have a number of benefits that make them both feasible and likely candidates for success. Introducing greater competitive bidding to medical equipment purchasing and establishing competitive bidding for Medicare Advantage plans outright will allow CMS to take lessons learned from successful past efforts and scale up these established solutions. Addressing upcoding in Medicare Advantage will allow CMS to revolutionize incentives within Medicare Advantage, moving away from the current quantity-over-quality paradigm. Together, these changes could allow CMS to recoup billions in lost revenue while increasing the quality of care.
Policy options to align payment for misvalued care
A third set of policies that would improve value in Medicare center around aligning payment rates for care that is currently misvalued. Policymakers can achieve realigned payment by reformulating how Medicare determines the value of a procedure, reducing bad debt payments, addressing high payments for post-acute care, and aligning payments for comparable services across facility types.
Reform value determinations for Medicare services
About 20 percent of the nation’s health expenditures are for physicians’ services, with Medicare payments to physicians accounting for about $195 billion in 2020 and a projected $313 billion in 2021 and $345 billion in 2022.105 Traditional Medicare pays a separate amount for each service that corresponds to one of more than 7,000 codes under the physician fee schedule.106 Each payment amount is adjusted to reflect the relative resources needed to perform that particular service compared with other physician services. Federal law requires CMS to review all relative values at least every five years and identify services that are likely to be misvalued on an ongoing basis.107
The current system has two fundamental problems. First, it is rife with conflicts of interest due to CMS’ reliance on recommendations submitted by the American Medical Association’s (AMA) Relative Value Scale Update Committee (RUC) when determining a service’s value.108 As the Government Accountability Office has noted, RUC “is made up of practitioners who have a financial stake in” Medicare’s final payment decisions, it operates largely out of view, and its members’ subjective assessments comparing the relative ease or difficulty of various tasks are confidential.”109
Second, there are substantive problems with how the RUC values nonprocedural work by physicians. The relative value of a service accounts for the value of physicians’ work—both the time needed to provide the service as well as the intensity—including the cognitive effort and judgment, technical skill, and psychologic stress associated with performing the service.110 The RUC develops its recommendations on the relative values of services based on subjective physician surveys that estimate both the time and intensity of a specific service and that also compare those values relative to other specified services.111
This conflict-riddled, subjective process results in the chronic underpayment of services that are critical to ensuring optimal health, in particular primary care and behavioral health. Surveyed physicians have historically placed far higher values on procedural services that involve physical and mechanical skills, such as colonoscopies or cataract extractions, instead of services that require “critical thinking involved in data gathering and analysis, planning, management, decision making, and exercising judgment in ambiguous or uncertain situations.”112 Because Medicare’s relative values are also used by Medicaid and private insurers, these flaws “ripple through the entire health care system.”113
Policymakers must reform how Medicare assigns value to physician services. First, payment for nonprocedural services such as primary care and mental health services should not be set based on arbitrary comparisons to procedure-based services. Second, payment for procedural services should reflect empirical data instead of subjective assessments by RUC physicians as to how difficult and how long procedures take. Overall, there should be far more transparency in the value-setting process.
Reduce bad debt payments
Currently, Medicare reimburses eligible providers—hospitals, skilled nursing facilities, various types of health care centers, and facilities treating end-stage renal disease—for 65 percent of the allowable “bad debt” owed by fee-for-service beneficiaries.114 This “bad debt” is essentially the unpaid and uncollectible cost-sharing from covered services provided to beneficiaries.115 The original purpose of this coverage was to prevent the outstanding costs of Medicare beneficiaries from being shifted to other non-Medicare patients.116
As coverage increases, particularly Medicaid coverage, and uncompensated care at medical facilities declines sharply, including in rural areas, there should be less need for the federal government to subsidize providers through methods such as the Medicare bad debt policy.
Notably, increases in health insurance coverage through the Affordable Care Act (ACA) have led to significant declines in the overall uncompensated care absorbed by medical providers,117 with reductions up to 55 percent for some hospitals.118 The enhanced ACA marketplace subsidies introduced through the American Rescue Plan Act, and extended by the Inflation Reduction Act, have helped nearly 2 million more adults enroll in marketplace coverage—contributing to a record-low 8 percent uninsured rate in early 2022.119
As coverage increases, particularly Medicaid coverage, and uncompensated care at medical facilities declines sharply, including in rural areas, there should be less need for the federal government to subsidize providers through methods such as the Medicare bad debt policy.
CMS has already taken some efforts to tighten its bad debt policy through its and FY 2022 Hospital Inpatient PPS’ final rules, 120 though many of the changes merely codified CMS’ long-standing practices.121 Changes include specifying the length of time and reasonable collection efforts providers must follow before writing off a balance as bad debt and documentation requirements as well as requiring state Medicaid programs to facilitate timely enrollment of valid Medicare providers so that providers can submit claims for dual eligibles before claiming bad debt reimbursement.122 However, these changes ultimately leave Medicare’s bad debt policy intact and will not produce meaningful savings.
The CBO has scored several policy options that achieve cost savings by reducing Medicare’s bad debt coverage. The first would be a more modest reduction in coverage, from 65 percent of allowable debt to 45 percent, yielding savings of $21 billion by 2030. A reduction in coverage to 25 percent of allowable debt would yield nearly $43 billion in savings by 2030. Finally, the elimination of allowable “bad debt” coverage would result in $69 billion saved by 2030. Alternatively, CMS can perform a study to determine whether the bad debt policy is a needed and effective solution to cost shifting in the wake of recent coverage expansions and, if so, what rate of reimbursement is best.
Address expensive post-acute care
Post-acute care refers to a variety of medical services that support a person’s continued recovery after a hospital stay.123 Long-term care hospitals (LTCHs) treat patients with severe recovery needs; inpatient rehabilitation facilities (IRFs) treat patients who need intensive rehabilitation in combination with hospital-level care; skilled nursing facilities (SNFs) treat patients with lower care needs; and home health agencies (HHAs) provide care to homebound patients.124 In 2020, FFS Medicare spent around $58 billion on post-acute care,125 and the Medicare Payment Advisory Commission (MedPAC) has described payments for post-acute care in SNFs, HHAs, and IRFs as high relative to the cost of care for more than a decade.126
There are a variety of policy options that can be used to improve the accuracy of post-acute care provider rates. First, CMS should maintain or reduce the payment rates for post-acute care facilities in future fiscal years to better reflect the cost of care in these settings. In a recent report to Congress, MedPAC described Medicare Advantage payments for post-acute care as being “considered attractive by many SNFs” despite being much lower than FFS payments.127 The same trend of overpayment persisted for two other facility types, HHAs and IRFs.128 As a result, MedPAC recommended no change to the SNF payment rate and a 5 percent reduction in payment to both HHAs and IRFs.129 By adjusting payments to these facilities to avoid needless increases in payments, HHS would bring payments more in line with the actual cost of providing post-acute care to patients and ensure that Medicare patients are provided with valuable care. This is also a foundational step for further reforming post-acute care payment rates, as these rates are used as the baseline for alternative payment models.
Another policy option is expanding the use of APMs for post-acute care. For example, in its FY 2017 budget for the U.S. Department of Health and Human Services (HHS), the Obama administration proposed implementing bundled payments for post-acute care providers, including LTCHs, IRFs, SNFs, and HHAs.130 The proposal would have required at least half of the total payments to such providers to be bundled, with a goal of reducing payments by 2.85 percent after the first two years of implementation.131 The budget proposal estimated that this would save nearly $10 billion over 10 years, but it was not included in the final appropriations legislation.132
HHS has tested some bundling for post-acute care through CMMI demonstration models, as discussed in the first section of this report. Two models in the BCPI demonstration, Model 2 and Model 3, included post-acute care services within the episode of care.133 Both Model 2, which included both acute and post-acute care, and Model 3, which solely included post-acute care, resulted in reductions in total payments.134 Evaluations of these models, however, found that the benchmarks were set too high, which resulted in the shared savings payments to providers being far too high.135 These benchmarks were based on estimates of spending growth, and the benchmarked estimates ranged from 2.7 to 4.4 percentage points higher than evaluation results for similar facilities.136 Because incentive payments were based on these overestimations of how much spending would have grown, providers received inflated payments, resulting in a net increase in spending.137
In addition to the BPCI bundle, the CJR model—a demonstration testing bundled payments for hip and knee replacements—also addressed post-acute care. The CJR episode of care included 30 days after discharge from the hospital, which would include post-acute care for these surgeries. The CJR model likely achieved savings to the Medicare program, driven by participating hospitals reducing their level of post-acute care delivered through SNFs, as hospitals overwhelmingly made shorter stays at SNFs a goal.138 Hospitals worked to send more patients home, paired with outpatient physical therapy, as well as to ensure that patients were sent to higher-value SNFs.139 While CJR patients with hip fractures reported less improvement in functional status than non-CJR patients, overall measures of unplanned readmissions and complication rates improved during the model for all patients, and ER visits and mortality rates were unchanged.140
When implementing future bundled payment demonstrations, CMMI should ensure that episodes of care are properly defined and the benchmarks reflect the actual cost of the care being provided. In a 2019 presentation, MedPAC described two approaches to paying for post-acute care, one based on stays at a post-acute facility and one based on episodes that could include multiple stays related to the same condition.141 The commission found that an episode-based approach would standardize payment for different conditions but also incentivize shorter facility stays over longer.142 This has the potential to encourage providers to avoid higher-need patients, an outcome that should be avoided. CMMI should conduct demonstration models of these approaches to determine how best to structure post-acute bundles.
Finally, addressing post-acute care payment should also ensure that facilities are properly being evaluated for the care they provide. Under the Protecting Access to Medicare Act of 2014, CMS was required to implement a value-based purchasing program for SNFs.143 Under this program, however, SNFs were evaluated solely based on the level of hospital readmissions that their patients had, which captures an incomplete picture of the quality of care these patients received.144 For example, patients readmitted to a hospital while still at an SNF were treated identically to patients readmitted after discharge, even though these readmissions stem from different shortcomings in the care delivered. MedPAC recommended adding additional claims-based performance measures, such as successful discharges to the community, as well as including measures of patient experience, to develop a more holistic and accurate understanding of care quality.145
In addition to issues with how the quality metric was evaluated, MedPAC also found that the program disincentivized SNFs with higher proportions of low-income and medically complex patients from participating, as social risk factors were not included in evaluations of these facilities.146 This is particularly important for post-acute care, as medically complex patients will require more care, increasing the likelihood that these patients will receive a higher-than-necessary volume of low-value care. Intentional design is needed to ensure that accounting for this increased risk does not result in these patients receiving worse care. In the same report, MedPAC recommended establishing a social risk factor score to ensure that performance is accurately compared without the risk of lower-quality care to these patients. CMS should consider applying this type of risk adjustment to future bundled payment designs to avoid the patient selection concerns associated with episode-based bundling.
Implement site-neutral payments
Another policy option to address misvalued care is to implement site-neutral payments for Medicare patients. Site neutrality refers to paying the same rate for a given service, regardless of the setting in which it was provided.147 This has the potential to drive significant amounts of spending, as for some services, Medicare provides substantially different payment rates based on the setting.148 Depending on the service, this can result in wasteful spending if a service can be safely and effectively provided by other providers or in other facility types.
Medicare payments have historically favored hospitals,149 contributing to both hospital acquisition of other providers and consolidation in the provider market, which in turn can drive up health care spending and patient cost-sharing.150
CMS has taken steps in the past to address payment disparities between facility types. In 2015, Congress passed the Bipartisan Budget Act of 2015, which required site-neutral payments for newly established off-campus, provider-based outpatient departments, though the 2016 21st Century Cures Act exempted additional departments from this approach.151 In 2018, CMS finalized a site neutrality rule for the Outpatient Prospective Payment System (OPPS) to lower payments to all off-campus hospital outpatient departments (HOPDs) for clinical visits to bring them in line with the payments to ambulatory surgical centers (ASCs) starting in 2019.152 The American Hospital Association and several hospitals sued in response to this regulation, and the D.C. Court of Appeals upheld CMS’ rule in 2020.153
Medicare payments have historically favored hospitals, contributing to both hospital acquisition of other providers and consolidation in the provider market, which in turn can drive up health care spending and patient cost-sharing.
The savings possible through site-neutral payments are substantial. In 2019 alone, Medicare could have saved more than $6.6 billion and beneficiaries could have saved $1.7 billion if the payment rates for care provided at HOPDs, ASCs, and doctors’ offices were the same.154 Over a longer period of time, expanding the 2018 changes to OPPS from clinical visits to all services was estimated by CBO in 2020 to save more than $14 billion over 10 years.155 The same CBO report found that making the payments for clinical visits performed at hospitals equivalent to those performed at physician offices—essentially the same policy that CMS finalized in 2018, applied to hospitals—would save more than $71 billion over the same time period.156
MedPAC has also explored implementing site neutrality to additional Medicare payments. In its November 2021 public meeting, MedPAC presented a proposal aligning rates for about 40 percent of OPPS services with payment rates that vary between HOPDs, ASCs, and physician offices.157 These low-complexity services, which MedPAC felt represented those that could reasonably be provided in lower-cost settings, would have their rates adjusted down to that of the highest volume setting. For example, if physician offices had the largest volume of cases, the new rate for HOPDs and ASCs would have their rates aligned with the physician office rate, and if ASCs had the largest volume, HOPDs would have their rates aligned with the ASC rate.
Reforms are urgent to remove the conflicts of interest and subjectivity that have come to define many aspects of Medicare payment.
MedPAC estimated that taking this approach would save Medicare $7.74 billion (about 3.6 percent of Medicare spending) in the fiscal year this change is implemented as well as $1.96 billion in reduced cost-sharing for patients.158 MedPAC also examined tailoring these policies to ensure that hospitals that serve vulnerable patients, such as critical access hospitals, were not negatively affected. Even when limiting the percentage reduction to hospitals that serve more than the median share of low-income patients, these policies would produce savings equivalent to 3.8 percent of Medicare’s total revenue.159 In all cases, the federal savings and cost-sharing reductions would be greater among patients served by rural hospitals and government-owned hospitals.160
Reforms are urgent to remove the conflicts of interest and subjectivity that have come to define many aspects of Medicare payment. The policy options detailed above would modernize Medicare, incorporate more objectivity and transparency in payment determinations, bring payment more in line with the private sector, and, most importantly, incentivize providers to deliver high quality, nonwasteful care.
Medicare has made progress in improving the value of the care that it supports. The program continues to shift away from fee-for-service payments in favor of value-based arrangements. More reforms are needed, however, to ensure that the care that Medicare beneficiaries receive is both high-quality and cost-effective.
The policy options presented in this report represent a variety of ways to improve the quality of care that Medicare beneficiaries receive, lower the price paid for that care, or both. Building on the successes of and learning from the drawbacks of previous demonstration projects and making more of these demonstrations mandatory will continue the transformation of Medicare from a volume-driven to a value-based delivery system. Meanwhile, increasing the amount of competition within Medicare by expanding existing competitive bidding programs and addressing manipulations of risk scores will reduce wasteful spending and avoid unnecessary care for patients. Finally, improving the value of Medicare will help ensure that the program serves beneficiaries and taxpayers better, all while paving the way for continued improvements throughout the U.S. health care system.
This publication was made possible in part by a grant from the Peter G. Peterson Foundation. The statements and the views expressed are solely the responsibility of the Center for American Progress.