top of page

Insights InHealth


Looking Back to Move Forward: Lessons on Provider Compensation from Innovative Payment Models

Updated: Sep 26, 2023

Recently, the Centers for Medicare and Medicaid Services (CMS) released studies synthesizing the degrees of success of the value-based care (VBC) payment models established by the CMS Innovation Center (CMMI). One comprehensive study published by CMS (the 2012-2020 Synthesis)[1] evaluated the payment models based on their ability to reduce costs, limit unnecessary care, and improve care quality; it looked at both targeted payment models (acute or specialty care and targeted population models) and broad-based models (primary care and population management models). Another study (the Primary Care Lessons Paper)[2] published in the JAMA Health Forum summarized lessons learned from the primary care payment models established by CMMI. Both of the studies provide a glimpse into the future of Medicare and Medicaid reimbursement for professional services. This edition of Insights InHealth seeks to summarize the findings of these studies, highlight the anticipated changes to payment models based on the findings, and detail ways organizations can proactively align provider compensation with future innovative payment models today.

Overview of Lessons Learned from CMMI Payment Models

Based on the information contained in the 2012-2020 Synthesis and Primary Care Lessons Paper, a number of key lessons can be derived from CMMI’s various payment model experiments. The table below provides a broad summary of the key takeaways that we identified, and that are explored in more detail below:

Key Takeaways from CMMI Payment Models

1. Few models were able to achieve ALL three of the Triple Aims

2. The more targeted the patient population and costly the underlying disease, the higher the likelihood of success reducing costs

3. A long time horizon is needed to evaluate the effectiveness of primary care payment models

4. Voluntary participation in models makes evaluations challenging

5. Coordination and data sharing with providers outside of the models is vital

The 2012-2020 Synthesis evaluated the CMMI payment models based on shared characteristics including: “gross and net Medicare spending and measures of utilization (e.g., inpatient admissions, emergency department visits, post-acute care, inpatient readmissions) and quality of care (e.g., self-reported satisfaction with care, mortality).” The spending, utilization, and quality measures touch on the Triple Aims framework adopted by CMS, specifically: better care for individuals, better health for populations, and lower costs.[3] Unfortunately, of the 21 models evaluated in the 2012-2020 Synthesis, only 3 (less than 15%)[4] resulted in improvement across the three major domains of: net spending reductions, lower utilization, and improved quality. Generally, the models appeared to offer a tradeoff between the ability to meaningfully reduce costs and improve the quality of care (whether in terms of patient experienced or reduced mortality). While all of the models pertaining to acute or specialty care and targeted populations were able to reduce gross expenditures, the materially high incentive payments attached to most of these programs often offset or exceeded these savings.

According to the 2012-2020 Synthesis, the models pertaining to acute or specialty care and targeted populations generally reported more success in furthering the Triple Aims than those pertaining to primary care or population health. This difference is attributed to a few factors, including: (1) most targeted programs were focused on high-cost conditions that offered a greater potential for meaningful savings, (2) the degree of targeting facilitated the development of meaningful quality metrics and initiatives within the control of providers, and (3) targeted models were subject to less noise given the higher likelihood of patient turnover in the primary care and population health models.

Separately, both the 2012-2020 Synthesis and Primary Care Lessons Paper discuss the challenge of evaluating the effectiveness of primary care payment models given the broad and healthier profile of these patients. These models are designed to maintain health and proactively avoid the development of costly conditions, with the benefit of such efforts often accruing outside of the two- to five-year evaluation time horizons associated with the innovative payment models. Relatedly, both papers highlighted the fact that organizational capacities grown by primary care practices (e.g., improved EHR data measurement and reporting, care-team staffing models, better coordination with outside specialists) yielded the most meaningful benefits towards the end of the evaluation periods when they hit their stride. Indeed, CMS cautions that the short time horizons might show deceptive increases in spending “…suggesting that the screenings and disease management activities could have created more engagement with the health care system, prompting more visits in the short-term as beneficiaries became more aware of their symptoms and conditions.”

The 2012-2020 Synthesis underscored the difficulty evaluating payment models when participation was voluntary. Some of the key problems of voluntary participation included: (1) selection effects, whereby the practices best able to implement the payment models participated the most, and (2) timing issues, whereby practices dropped out before the end of the evaluation period once they determined participation was not financially beneficial. The Primary Care Lessons Paper also highlighted substantial variations in care by geography, using the inconsistent availability of state programs that address health-related social needs as an example of ways that coordinated primary care may differ by location.

Both the 2012-2020 Synthesis and the Primary Care Lesson Paper detailed the importance of care coordination and data sharing. Despite being a key goal for primary care and population health programs, the Primary Care Lesson Paper noted “…the effectiveness of these tactics depended highly on informal and organizational relationships and resources, local referral patterns and practices, and workforce availability.” The financial incentives offered in the CMMI payment models were not suited to address these items or, as noted in the 2012-2020 Synthesis, “…model incentives payments were not always large enough to compensate for the amount of work required in the model.”

Features of Future Payment Models

In addition to the insights gained from evaluating the success of CMMI payment models, CMS also offered suggestions regarding refinements to future payment models based on prior experience. The table below identifies a few of the features that we found notable:

Key Future Refinements to Payment Models

1. Greater coordination between ALL payors (Medicare, Medicaid, and Private)

2. Continued targeting of high-cost conditions with bundled payment

3. Financial incentives tied to health equity and social risks

4. More downside risk to providers

5. Reduced number of voluntary models with increased mandatory models

CMS has identified that participation in innovative payment models can be inhibited by the upfront costs required to implement the infrastructure needed for success. According to the Primary Care Lessons Paper, coordination across payors is essential to “…align payment models, quality measurement, and data feedback across payers to reduce practice burden and operational costs.” As a result, CMS is encouraging multi-payor alignment through the involvement of state Medicaid agencies and commercial payors early in the design of new payment models.

The 2012-2020 Synthesis highlighted the widespread success of the acute or specialty care and targeted population models in reducing spending, primarily through fewer inpatient admissions and post-acute care; conversely, the changes in emergency department visits and readmissions were more ambiguous. Furthermore, the greater the degree of targeting of the patient population, the greater the success in reducing spending. As stated in the 2012-2020 Synthesis, “The patient populations within each of these [targeted] models are fairly homogenous relative to the broader, general Medicare population covered in the population-based models, which allows for a discrete set of care delivery practices and quality measures for health care providers to focus on.” As a result, we expect an increased focus on bundled payment models for targeted and high-cost populations that incentivize savings through reduced utilization of resources throughout the course of care.

Historically, the healthcare industry has not prioritized the promotion of health equity and reimbursement incentives to address the social determinants of health. Beginning with CMMI’s strategy refresh in 2021 and new CMS priorities put forth by the Biden Administration, these items will become core features of the national healthcare landscape. The Primary Care Lessons Paper highlighted the significant overrepresentation of White beneficiaries, relative to Black and Hispanic beneficiaries, in value-based primary care payment models. As a result, innovative payment models, like ACO REACH, and existing programs, such as Medicare’s Shared Savings Program, have incorporated financial incentives to provide care to underserved populations. Furthermore, CMS has targeted the collection of data related to social determinants of health across a variety of payment programs (e.g., Inpatient Prospective Payment System). These financial incentives and data initiatives represent a first step along the path to redressing health disparities. It is therefore unsurprising that the Primary Care Lessons Paper explicitly stated that one of its essential lessons is that “Equity Must Be an Explicit Aim of Primary Care Models.”

Though not discussed in the 2012-2020 Synthesis and Primary Care Lessons Paper, the experience of CMMI hints at the impact that meaningful incentives have on influencing the costs and quality of care. One of the most controversial topics in VBC is the degree of downside risk required to make payment programs a success. Fortunately, enough time has elapsed to allow for a meaningful review of evidence. The results are fairly conclusive (and validate the findings on net spending detailed in the 2012-2020 Synthesis): payment models with downside risk do a better job promoting the Triple Aims than those with upside-only risk. A literature review looking at commercial payor VBC payment models[5] and a study looking at nearly 500,000 Medicare Advantage beneficiaries[6] both determined that models with downside financial risk outperformed fee-for-service and upside-only risk models in reducing utilization / spending. The commercial payor literature review found similar results in improving the quality of care. As a result, we expect prevalence of downside risk payment models to increase, as commercial payors contain costs to keep premiums manageable and Medicare seeks to improve the depletion of its trust fund.

As noted above, CMS indicated that the voluntary nature of many programs impeded accurate evaluation of their success. Beyond the shortcomings already listed, the 2012-2020 Synthesis highlighted that some voluntary models resulted in patient population sizes that were too small to generalize the conclusions to a broader population. For example, one of the most unambiguously successful programs (the Medicare Care Choices Model), was limited by a small population (only 1% of those eligible participated) and a high dropout rate (60%). Accordingly, we predict that CMS will increase the degree of mandatory participation in innovative payment models. As concluded by the 2012-2020 Synthesis: “A mandatory model with random assignment to the treatment and control group would ensure that all types of organizations participate, enhancing the generalizability of the findings and ensuring that the health care system as a whole is being transformed by [CMMI] initiatives.”

Designing Aligned Provider Compensation Models

In the prior two sections, we sought to encapsulate the compelling story CMS is telling regarding its future direction in paying for healthcare. With that background in mind, let us guide you through some of the key design principles that will ensure success in VBC. The table below summarizes the compensation design principles that further alignment with the characteristics of future payment models:

Features of Aligned Compensation Plans

1. Increasing at-risk compensation tied to payment model goals

2. Financial rewards for furthering health equity and addressing social risks

3. Rethinking the production metrics utilized in compensation plans

4. Effective compensation design for physician leaders

Increasing At-Risk Compensation to Providers

As extensively documented in research on reimbursement models, programs that expose providers to both upside and downside risk are more likely to achieve objectives associated with the Triple Aims. As a result, at-risk compensation tied to the successful achievement of quality measures or payment program initiatives is well-suited to align the incentives of providers and payors. Though research on the topic is limited, the current percentage of at-risk compensation is relatively low (9% for primary care and 5% for specialists, according to RAND)[7] and may not be sufficient to motivate providers to focus on VBC.

Historically, the regulatory environment around provider compensation had limited the ability to pay directly for efforts designed at furthering VBC. For example, a compensation plan by a hospital employer that would provide a bonus to a physician for improving the cost effectiveness of care to inpatients could be said to have violated the prohibition on basing provider pay on the value of referrals imposed by various statutes. In recognition of this misalignment, regulatory agencies have sought to introduce flexibilities in the form of safe harbors to the Anti-Kickback Statute and exceptions to the Stark Law, which permit compensation structures that further VBC. In a prior edition of Insights InHealth, we explored Value-Based Enterprises (VBEs), which offer extensive flexibilities in provider compensation design. At-risk compensation meeting the criteria for the VBE safe harbors/ exceptions offer a new and powerful tool to align provider pay with VBC.

In our experience, successful at-risk compensation structures should strive to attach payment to quality measures and initiatives that meet the following criteria:

Financial Rewards for Furthering Health Equity and Addressing Social Risks

As noted above and in other editions of Insights InHealth, a larger proportion of Medicare payment programs are tying reimbursement to efforts to further health equity and address social risks. For example, the most recent proposed rule for the Medicare Shared Savings Program recommends additional payments to providers that treat a high share of patients in underserved areas, while penalizing providers that do not track patient information pertaining to social determinants of health. Traditionally, providers received no financial benefit for treating underserved patients; indeed, providers faced disincentives to treat underserved communities given comorbidities/complexity of care, lack of access to transportation/support resources, and poor reimbursement from safety-net payors.

As payors increasingly reward care directed to underserved patient populations, financial incentives to promote health equity have become a desirable feature of compensation plans. As with quality measures and initiatives, financial rewards tied to health equity and addressing social risk must meet the criteria detailed in the figure from the prior section. Examples include:

  • At-risk compensation tied to the partial or total elimination of health disparities among underserved populations (while holding constant or improving the quality of care to non-underserved patients).

  • (For Primary Care) Incorporating risk adjustments into panel sizing that increase the risk associated with patients from underserved communities.

  • Fixed bonuses for exceeding a threshold of patients that are designated from an underserved population.

  • Providing stipends for underserved patients treated by a provider and referred to community resources addressing health-related social needs (e.g., food banks).

The last example highlights a compensation term that might formerly have been prohibited by law, but which may be allowable under the new VBE structure promulgated by CMS.

Rethinking Production Metrics in Compensation Plans

For decades, providers have become accustomed to compensation plans tied to production as measured by work relative value units (wRVU). While such plans work well in a fee-for-service payor environment, they often undermine the goals of VBC. For example, a provider paid per wRVU has every incentive to maximize the utilization of diagnostic resources they personally have access to, if doing so yields more wRVUs. As discussed in the first section, a key goal of VBC is the reduction of wasteful utilization in the healthcare system, which makes exclusively wRVU-based payment plans a potential stumbling block to achieving that goal. Further, healthcare quality measures tied to patient experience can also be diminished by wRVU-based plans if the marginal effort of the provider is focused on producing more, as opposed to spending face-to-face time with the patient. This theme was echoed in Medicare’s proposed change to split-share billing rules related to inpatient care performed by advanced practice professionals and supervised by physicians, whereby Medicare sought to attribute reimbursement to the provider with the greatest amount of face-to-face time with the patient.

In discussing the lessons learned above, we note an emerging understanding that provider time is a precious and limited resource, and that the most effective use of this time is central to delivering on the promises of VBC. As a result, careful thought is needed to allocate provider compensation across the various types of payment terms. To facilitate this process, we encourage organizations to consider how their compensation plans reward providers for effort directed to production, quality, availability, and administration. For example, successful VBC programs require providers to dedicate time to managing care quality through tasks such as utilization review, EHR report analysis, and keeping current with specialty best-practices. A provider paid exclusively based on wRVU-production receives no remuneration for this work. An organization that must adapt to VBC payment models will not succeed in the long run if this misalignment is not addressed.

Effective Compensation Design for Physician Leaders

Beyond reevaluating compensation for all providers, it is especially important for organizations to focus attention on how physician leaders are paid. Traditionally, these providers were designated as “medical directors” and received an hourly stipend based on documented time spent on administrative duties. With the growth of VBC, we have observed a substantial widening of the scope of what constitutes effective physician leadership. In order to deliver higher quality care at lower costs, engaged physician leadership is a necessity. For example, industry best practices for accountable care organizations recommend leadership dyads involving a physician and administrator in order to “…facilitate formal decision-making processes at both the market and central level that are critical to ensuring the success of complex, multi-pronged value transformations.”[8] Physician leaders can serve as force multipliers by functioning as “value ‘champions’ at all levels of the organization to maintain momentum for change.” It is not a stretch to claim that physician leaders of organizations involved in VBC are the lynchpins to sustainable success.

With all that said, the industry has yet to adapt to this reality. Many medical directors are selected (often by their peers) because they are the providers that produce the most wRVUs or because they have served with the longest tenure. Though valuable traits, it is also essential to evaluate their ability to: synthesize large data reports into actionable clinical strategies, enthusiastically engage peer clinicians in VBC, and advocate for alignment between patients, providers, administration, and payors. A compensation plan driven by wRVUs is unlikely to direct the energies of a talented physician leader to furthering VBC. We encourage organizations to think about how payment must be structured to attract and retain physician leaders, as opposed to high volume producers. As payors increasingly tie reimbursement to reduction in costs and improvement in quality, this task will become central to long-term financial viability.

Learning from CMMI Models
Download PDF • 781KB


[1] Centers for Medicare and Medicaid Services: Center for Medicare and Medicaid Innovation. Synthesis of Evaluation Results across 21 Medicare Models, 2012-2020. Last accessed August 23, 2022, from:

[2] Finke, Bruce, MD; Davidson, Kathryn, LCSW; and Rawal, Purva, PhD. “Addressing Challenges in Primary Care—Lessons to Guide Innovation.” JAMA Health Forum (August 19, 2022). Last accessed August 23, 2022, from:

[3] Last accessed from the CMS website on August 23 ,2022, from:

[4] These are Medicare Care Choices Model, ACO Investment Model, and Pioneer ACO Model.

[5] Milad, Marina A;, Murray, Roslyn C.; Navathe, Amol S.; and Ryan, Andrew M. “Value-Based Payment Models in the Commercial Insurance Sector: A Systematic Review.” Health Affairs, Volume 41 Issue 4, published April 2022. Last accessed August 23, 2022, from:

[6] Gondi, Suhas, BA; Li, Yong, PhD; and Antol, Dana Drzayich, MS. “Analysis of Value-Based Payment and Acute Care Use Among Medicare Advantage Beneficiaries.” JAMA Network Open (March 17, 2022). Last accessed August 23, 2022, from:

[7] “Despite Push to Reward Physicians for Quality and Value, Most Health Systems Base Pay on Volume.” RAND Corporation Press Release (January 28, 2022). Last accessed August 23, 2022, from:

[8] Health Care Transformation Task Force. Transformation to Value: A Leadership Guide. Last accessed August 23, 2022, from:


Subscribe to our newsletter • Don’t miss out!

Thanks for subscribing!

bottom of page