Are Value-Based Models Delivering More Value?

Published 2020.9.22

The Centers for Medicare & Medicaid Services (CMS) released two new payment models on Friday: one for kidney disease promoting home hemodialysis and kidney transplantation and the other for radiation oncology including 16 different cancers. The End-Stage Renal Disease Treatment Choices model will adjust some payments to nephrologists and other clinicians managing patients with end-stage renal disease (adjustments to prospective payment system per treatment and a monthly capitation payment). The kidney disease model is expected to save $23 million over 5.5 years. The radiation oncology model uses 90-day episodes of care with bundled payments to radiotherapy providers and radiotherapy suppliers. It is expected to save Medicare $230 million over the five-year evaluation period. Both models require provider participation and expect to enroll nearly a third of all eligible Medicare beneficiaries across the country.

Since its inception in 2010, the CMS Innovation Center has been churning out a dizzying number of payment and service delivery models to achieve the triple aim (better care for patients, better health for communities and lower costs for the healthcare system). Models must either reduce spending without reducing the quality of care, or improve the quality of care without increasing spending, and must not deny or limit the coverage or provision of any benefits. The Innovation Center also helps to implement the Quality Payment Program, a program to replace Medicare’s Sustainable Growth Rate formula, to encourage the adoption of advanced alternative payment models.

The CMS Innovation Center 2018 report to Congress highlighted the following models that demonstrated savings:

  • Pioneer ACO model (started with 32 ACOs in 2012 and ended with nine ACOs in 2016) generated two-year savings nearing three percent of beneficiary spending.1
  • Y-USA (Young Men’s Christian Association [YMCA] of the USA) Diabetes Prevention Program generated nearly 14% Part A and Part B spending for Medicare beneficiaries2
  • Prior Authorization Model: Repetitive Scheduled Non-Emergent Ambulance Transport saved nearly four percent of Part A and Part B Medicare beneficiary spending3
  • Maryland All-Payer model showed almost three percent Medicare savings over the first three years as compared to a comparison group of non-Maryland hospitals
  • Next Generation ACO model decreased Medicare spending 1.1 percent after adjusting for shared savings/loss payments (see below for a more recent update showing no savings)

1In May 2015, the CMS Chief Actuary certified the Pioneer ACO model to be eligible for expansion, the first Innovation Model to do so. The Pioneer ACO Model learnings were incorporated into Track 3 of the Medicare Shared Savings Program (MSSP). In July 2019, MSSP was divided into the BASIC (levels A-E [C-E include the possibility of shared losses with limits on maximum shared savings and losses] with ACOs moving up one level annually until they reach level E) and ENHANCED (higher savings and loss opportunities with higher limits). As of January 1, 2020, 192 of the 517 MSSP ACOs (37%) were in a two-sided risk arrangement.
2The Y-USA Diabetes Prevention Program was the second CMS Innovation Center Model to meet expansion criteria. The expanded program is called the Medicare Diabetes Prevention Program.
3As of November 2019, CMS was still evaluating if this program meets expansion criteria.

Two models that did not show savings, but did inform future models are:

  • Bundled Payments for Care Improvement (BPCI) did not generate savings after deducting reconciliation payments (click here for the sixth evaluation report published in June 2020). The new model, BPCI Advanced, revised target prices to incorporate risk adjustment for patient complexity and reflect peer performance a higher discount than BPCI.
  • Comprehensive Primary Care model increased Medicare expenditures by one percent. The new model, Comprehensive Primary Care Plus (CPC+), requires tracking individual practice performance instead of regional performance and taking on two-sided risk.

CMS provides a five percent incentive for achieving threshold payments or patients in advanced alternative payment models through its Quality Payment Program. Eligible advanced alternative payment models through payment year 2020 are:

Advanced Payment Model Description 2017 2018 2019 2020
Comprehensive End-Stage Renal Disease (ESRD) Care (CEC) Two-Sided Risk (ESRD, October 2015 – December 2020)1 Identify, test, and evaluate new ways to improve care for Medicare beneficiaries with ESRD. Y Y N N
Comprehensive ESRD Care (CEC) Model (LDO arrangement and Non LDO Two-Sided Risk Arrangement, October 2015 – December 2020) Identify, test, and evaluate new ways to improve care for Medicare beneficiaries with End-Stage Renal Disease (ESRD). Through the CEC Model, CMS partners with health care providers and suppliers to test the effectiveness of a new payment and service delivery model in providing beneficiaries with person-centered, high-quality care. N N Y Y
Comprehensive Primary Care Plus (CPC+, January 2017 – December 2022)2 Primary care medical home model that aims to strengthen primary care through regionally based multi-payer payment reform and care delivery transformation Y Y Y Y
Next Generation ACO Model (January 2016 – December 2020)3 Sets predictable financial targets, enables providers and beneficiaries greater opportunities to coordinate care, and aims to attain the highest quality standards of care. Y Y Y Y
Medicare Accountable Care Organization (ACO) Track 1+ Model (January 2018 – December 2020) Time-limited model for Track 1 Medicare Shared Savings Program (Shared Savings Program) ACOs. The Shared Savings Program is a voluntary program that encourages groups of doctors, hospitals, and other health care providers to come together as an ACO to provide coordinated, high-quality care to their Medicare patients. Track 1+ Model ACOs assume limited downside risk. N Y Y Y
Medicare Shared Savings Program – Tracks 2 and 3 Share in savings or repay Medicare losses depending on performance. Track 2 ACOs may share in a greater portion of savings than Track 1 ACOs. Track 3 ACOs take on the greatest amount of risk but may share in the greatest portion of savings if successful. Y Y Y Y
Oncology Care Model (OCM) – Two-Sided Risk (July 2016 – June 2021)4 Physician practices have entered into payment arrangements that include financial and performance accountability for episodes of care surrounding chemotherapy administration to cancer patients. Y Y Y Y
Bundled Payments for Care Improvement (BPCI) Advanced (31 inpatient episodes and four outpatient episodes, October 2018-December 2023)5 Links payments for the multiple services beneficiaries receive during a clinical episode of care N Y Y Y
Comprehensive Care for Joint Replacement (CJR) Payment Model (Track 1-CEHRT, April 2016 – December 2020)6 Support better and more efficient care for beneficiaries undergoing the most common inpatient surgeries for Medicare beneficiaries: hip and knee replacements (also called lower extremity joint replacements or LEJR). Y Y Y Y
Vermont Medicare ACO Initiative (as part of the Vermont All-Payer ACO Model, January 2017 – December 2022) Most significant payers throughout the entire state – Medicare, Medicaid, and commercial health care payers – incentivize health care value and quality, with a focus on health outcomes, under the same payment structure for the majority of providers throughout the state’s care delivery system and transform health care for the entire state and its population. N N Y Y
Maryland Total Cost of Care Model (Care Redesign Program and Primary Care Program, January 2019-December 2026)7 Voluntary program within the Maryland All-Payer Model that advances efforts to redesign and better coordinate care in Maryland. The CRP provides hospitals participating in the Maryland All-Payer Model the opportunity to partner with and provide incentives and resources to certain providers. In exchange, suppliers offer activities and processes that aim to improve quality of care and reduce the growth in total cost of care for Maryland Medicare beneficiaries. The Primary Care Program (PCP) is an advanced primary care medical home model that aims to strengthen primary care through multi-payer payment reform and care delivery transformation. N N Y Y

1The September 2019 evaluation report included two years of data. The evaluators found a 1.8% reduction in total Medicare Part A and Part B spending. The largest savings were in institutional post-acute care.
2In a July 2020 report including two years of data, evaluators found Medicare expenditures increased in CPC+ practices when including CMS’ enhanced payments.
3After two years of evaluation, the January 2020 Next Generation Accountable Care Organization evaluation report did not identify any Medicare savings.
4In a May 2020 report including the first three performance periods, evaluators found that OCM resulted in net losses for Medicare.
5The first annual BPCI Advanced evaluation report, published in June 2020, did not include any information about savings or quality improvements.
6The June 2019 annual evaluation report that included two years of evaluation data showed a no reduction in comprehensive care for joint replacement payments compared to controls after reconciliation payments were made.
7The November 2019 final evaluation report showed a 2.8% slower growth in total expenditures relative to the comparison group, with most of the relative savings from a slower growth in total hospital expenditures.

As of November 2019, there were many Alternative Payment Models that were not eligible for the five percent incentive, including: Kidney Care Choices: Comprehensive Kidney Care Contracting (Graduated Option 1, Graduated Option 2, Professional Option and Global Option); Direct Contracting Professional model (Professional and Global PBP); Emergency Triage, Treat and Transport (ET3); ESRD Treatment Choices (ETC); Home Health Value-Based Purchasing Model (HHVBP, after two performance years, the December 2019 evaluation report showed a 0.9% reduction in annual Medicare spending); Integrated Care for Kids (InCK); Medicare Patient Intravenous Immunoglobulin (IVIG) Access Demonstration Project; Medicare Advantage Value-Based Insurance Design (VBID); Primary Care First (General and Seriously Ill Population); and Transforming Clinical Practice Initiative (TCPI).

The 2018 report summarized its multiple program evaluations as follows:

  • The six primary care initiatives (including Comprehensive Primary Care) did not show a difference from comparison groups among core outcomes (Medicare expenditures, outpatient ED visits, hospital admissions and 30-day readmissions).
  • Episode payment models (BPCI, CJR and OCM) are associated with a reduction in costs and utilization without an impact on quality or functional status. Most cost reductions come from reduced post-acute care utilization.
  • Multi-payer state-based initiatives seem to have the strongest impacts on utilization and spending.

I believe the CMS Innovation Center to issue their next report in the spring of 2021.

The CMS Innovation Center has done more to measure the effect of alternative payment models than any other organization in the country. Most of the Center’s models have not resulted in a reduction in care quality or member satisfaction. Unfortunately, their approach to engage providers and hospitals at a payment rate that they will agree to participate in voluntary models may have forced the Center to accept incremental savings over transformational change. The fact that most APMs only reduced post-acute care utilization reflects the conservative approach of existing healthcare partners in driving higher value. The only intervention with double-digit measured percent savings, the Diabetes Prevention Program, was deployed outside traditional healthcare settings.

Shrank et al. suggested 25% of all healthcare spending is waste. The authors identified six waste domains: failure of care delivery, failure of care coordination, overtreatment or low-value care, pricing failure, fraud and abuse, and administrative complexity. The authors reviewed the published literature to estimate savings from interventions that address waste. The top 10 (by the upper end of the interventions’ estimated savings) are:

Intervention Savings
Integration of behavioral and physical health (failure of care delivery) $31.5-58.1B
Insurer-based pricing interventions (pricing failure) $31.4-41.2B
Laboratory and office visit pricing transparency (pricing failure) $29.7B
Prevention initiatives to address diabetes, obesity, smoking and cancer (failure of care delivery) $4.0-25.8B
Legislative, administrative and integrity strategies (fraud and abuse) $20.6-25.6B
Optimizing medication use (overtreatment/low-value care) $8.8-21.9B
Drug pricing interventions (pricing failure) $20.3B
Care coordination in accountable care organizations (failure of care coordination) $8.3-13.1B
Interventions to address adverse hospital events and hospital-acquired infections (failure of care delivery) $5.4-9.4B
Transitional care programs (failure of care coordination) $9.2B

The authors expected shared decision-making to reduce unnecessary procedures to save $3.2 billion, expanding hospice access to save $395 million – $3 billion, health information exchanges to save $205-410 million, prior authorization procedures to save $250 million, and incentives to increase physician efficiency to save $47.5 million.

Pricing failures account for a substantial proportion of the dollars that could be recaptured by applying the suggested interventions. The authors acknowledge that pharmaceutical pricing failures are not addressed by new payment models. Our efforts to use pricing transparency and other healthcare consumer decision aids have not resulted in meaningful savings to-date.

Instead of investing more into health information exchanges and prior authorization, CMS and other payers may have more success increasing healthcare value by coordinating care delivery outside the traditional office visit (e.g., optimizing medication use by using pharmacists who connect with the primary care team, facilitating care coordination across care settings). Other interventions might include integrating behavioral and physical health through new benefit design and launching prevention initiatives to address non-communicable diseases (e.g., diabetes, smoking-associated lung cancer).

There are other ways to conceptualize healthcare value. In a JAMA Internal Medicine editorial last week in response to a VA study about when deintensification might be appropriate, Mehta and Lehman suggested a different model to view low-value care. They focus on diagnosis, unnecessary treatment and futile/dangerous drug prescribing. Their paradigm is an update to the Institute of Medicine’s approach to diagnostic errors. Patients and providers interact in a diagnostic journey and then assess if a recommended treatment addresses the presumptive diagnosis. Spending additional dollars to confirm a diagnosis should result in improved healthcare value, especially if the suggested treatments are expensive or poorly tolerated. Deprescribing becomes a reassessment of the treatment’s risks and benefits as the patient changes over time. Another implications of the Institute of Medicine diagnostic error model is that reducing the cost of treatment (e.g., reducing monitoring costs and long-term prescription costs for chronic disease; reducing unnecessary evaluations prior to a definitive, therapeutic maneuver; and identifying clinical care delivery partners who are more efficient) is likely to increase the value of the healthcare delivered.

Our current efforts to deliver higher value care are not generating the savings we expect given the assumed waste in the healthcare system. Shifting care from high-cost settings to low-cost settings (e.g., hospital to outpatient, outpatient to home) seemed like a reasonable approach, but those savings do not appear to slow the rise in healthcare spending without a concurrent increase in quality scores when compared to other high-income countries. Although public health initiatives like Y-USA may show marked savings by delaying or preventing the onset of diabetes among high-risk members, we have yet to generate similar results among the other four leading causes of early death and disability in America (i.e., ischemic heart disease, lung cancer, COPD, and low back pain). Buxbaum et al. suggest public health interventions and pharmaceuticals were responsible for most of the gain in life expectancy between 1990 and 2015. Medical care accounted for less than 15%. If the gains from medical interventions are small, the primary ways to increase value are:

  1. Double down on public health interventions for select health conditions (e.g., ischemic heart disease, lung cancer, COPD, diabetes, opioid use disorder)
  2. Maximize medication adherence among those conditions where medications make a meaningful difference (e.g., ischemic heart disease, cerebrovascular disease, breast cancer and HIV)
  3. Instead of reducing the cost of each medical intervention, reduce the utilization of medical interventions that are not frequently beneficial (e.g., low back pain, neck pain), and
  4. Develop cost-effective interventions to address conditions that cause a high degree of premature mortality or excess morbidity (e.g., low back pain, major depression, migraine)

Targeting these approaches may have a higher likelihood of generating meaningful value for the American healthcare system.