Jun 17, 2022

Pathways for Specialty Care Coordination and Integration in Population-based Models

By: Sarah Fogler, PhD; Meghan O’Connell, MPH; Jacob Quinton, MD, MSHS; Chris Ritter, PhD; Brian Waldersen, MD, MPH; Purva Rawal, PhD

In support of the Centers for Medicare & Medicaid Services’ (CMS) vision to advance health equity, expand coverage, and improve health outcomes, the CMS Innovation Center launched a strategic refresh in October 2021. This included setting a bold goal to have 100% of beneficiaries in Traditional Medicare and the vast majority of Medicaid beneficiaries in accountable care relationships by 2030, where their providers are accountable for the quality and the total cost of their care. Beneficiaries will experience accountable care relationships mostly through advanced primary care or accountable care organizations (ACOs), and these entities are expected to coordinate with or fully integrate specialty care to deliver whole-person care. Achievement of this goal will make the health system more responsive and affordable for the people it serves.

While primary care remains central to a high-functioning health system, Medicare beneficiaries and their care providers are facing greater clinical and health system complexity. To complement population-based models, fully achieving whole-person care requires the additional depth and scope of services offered by specialty care and the effective coordination of primary and specialty care providers. Between 2000 and 2019, researchers have found that the portion of beneficiaries seeing five or more physicians annually increased from 18 to 30%, and the mean annual number of specialist visits increased by 20%. Moreover, they found that the average number of physicians with which a primary care provider needs to coordinate increased from 52 to 95 physicians from 2000 to 2019 —an 83% increase.[i]

In addition to the growing volume of visits, specialty care plays an outsized role in overall medical spending and offers important opportunities to increase the value of care. Between 2002 – 2016, Martin et al found that total annual health care spending increased by $806 billion, according to data from the Medical Expenditure Panel Survey (MEPS). Direct spending for specialists accounted for 18%, while primary care accounted for 4%, of this increase.[ii] Looking at low-value care, defined as services and supplies that offer no or limited benefit to patients, researchers have found that an estimated 75% of low-value care is neither provided by nor referred by the beneficiary’s attributed primary care provider.[iii] Further, Mafi et al have found that despite a national educational campaign and growing awareness of the problem of low-value care, spending on these services has only decreased marginally.[iv]

The Innovation Center’s model tests have shown improvements in lowering expenditures and enhancing quality. The episode-based payment models, including the original Bundled Payments for Care Improvement (BPCI) Initiative, the BPCI Advanced model, and the Comprehensive Care for Joint Replacement (CJR) model focusing on specialty care, have reduced Medicare FFS payments for the majority of episodes in the models while maintaining quality for beneficiaries.[v],[vi]  These models provide an important foundation for increasing access to coordinated and integrated specialty care.

It seems that substantially reducing low-value care further requires increased focus across specialty and primary care. The Innovation Center undertook a listening tour covering a diverse set of perspectives to understand the challenges to specialty care integration and how population-based payment models could better integrate specialists given limited experience with incentives of specialty integration in previous models. The listening tour included forty semi-structured interviews with stakeholders, in order to formulate a path to better align specialty care with primary care and ACO models.  The following section summarizes general themes from these discussions to inform the Innovation Center’s goal to integrate specialty care with primary care and ACO models going forward.

Increasing Access to Coordinated and Integrated Specialty Care

Stakeholders reported that advanced primary care and ACO initiatives have not successfully integrated specialty care into their organizations and/or care delivery infrastructure due to a complex set of barriers and disincentives. Four themes emerged from interviews that can inform model designs to better integrate specialty care and are described in detail below (see Table 1).

1.  Providing Data on Specialist Performance and Enhancing Data Sharing across Practices would Facilitate Integration with Primary Care

Actionable data on specialist cost and quality performance was cited by many stakeholders as critically needed to inform referrals of high clinical value. Without data to understand practice patterns and spending, some ACOs noted that it is hard to align or include specialists in the ACOs - especially if spending on specialty care was concentrated in a small subset of the beneficiaries aligned to an ACO. Referring clinicians across health systems reiterated the importance of data on specialist performance, including both cost and quality of care.

For example, as required by Congress, CMS has developed episode-based cost measures to evaluate specialist physician performance, in terms of resources used for specific conditions and treatments. Currently, 23 procedural, acute inpatient medical, and chronic condition episode-based cost measures are included in the Merit-based Incentive Payment System (MIPS), five more are in field testing, and two more are in early development.[vii] Most measures demonstrate acceptable reliability at the national provider identifier (NPI) and/or taxpayer identification number (TIN) level.[viii] Aggregating quality and cost data from Medicare, Medicaid, and multiple commercial payers, if feasible, could generate even more reliable and meaningful specialist profiles that could inform referrals of high clinical value and drive broader health system transformation.

Stakeholders said that enhancing data sharing on referred beneficiaries across primary care and specialist practices is critical to effective collaboration. Increasing specialists’ focus on ensuring value and appropriateness before scheduling a referral visit and closing the referral loop back to primary care post-visit would support accountability, shared risk, and team-based care. The Innovation Center is considering how model design could facilitate the flow of data, such as improving electronic consultation to efficiently access specialists’ expertise, to support co-management of patients through collaborative care agreements or care compacts, and to confirm appropriateness of the referral.

2.  Episode-based Payment Models Can Be Designed to Align Incentives between Specialists and ACO Initiatives

Episode payment models have successfully increased the number of providers in value-based payment models. For instance, in CY 2022, BPCI Advanced includes over 700 model participants—with an estimated 430 acute care hospitals and 400 physician group practices initiating episodes. In FY 2022, CJR includes 324 hospitals initiating lower joint replacement episodes. In the future, episode-based payment models can be deployed to facilitate expansion of accountable care within Traditional Medicare.

Right now, ACOs, hospitals, and physician group practices, however, may target the same savings opportunities within and across payment models, such as reducing historical inefficiencies in post-acute care utilization, if model overlap policies are not carefully designed and deployed. Stakeholders expressed confusion about Innovation Center model overlap policies. Currently, CMS policies on overlap between ACOs and episode-based payment models aim to avoid financial competition between ACOs and episode initiators for the same savings, either through precedence rules (i.e., Innovation Center ACO models take precedence over episode-based payment models) or by allowing overlap without direct adjustments (i.e., overlap between the Medicare Shared Savings Program ACOs and Innovation Center bundled payment models). However, historical confusion on overlap policies led to the misperception that episode-based payment models lead to fewer savings opportunities for Shared Savings Program ACOs - and looking forward, this confusion could weaken incentives for ACO participation. Current policies that allow full overlap between models without recouping payments retrospectively, such as between Shared Savings Program and the Bundled Payments for Care Improvement Advanced (BPCI Advanced) Model, mitigate duplicate payments through retrospective trends, but do not include active incentives to strengthen coordination across providers and care settings.

Bundled payments for specialty care should complement care transformation in advanced primary care and ACO initiatives. Strategic implementation of episode-based models can help fill the geographic and demographic gaps where accountable entities have yet to extend their reach and can help move the system toward accountability. To avoid adverse selection in provider participation and benchmarking complexity associated with voluntary models, stakeholders suggested that the Innovation Center also examine whether mandatory models can increase quality and access for beneficiaries, as well as increase provider participation in value-based payment. Stakeholders did note however, such approaches would have to ensure that those caring for underserved populations are not negatively impacted.

Where accountable entities can take on higher levels of accountability for a broader range of specialty care, CMS-run bundled payment models may be less necessary as these entities may have the tools, financial means, and influence to engage specialists and implement bundles directly. The Innovation Center could facilitate data analytic support for episode construction or other data aggregation tools.

3.  Many ACOs and Primary Care Practices Recognize Value in Specialty Care assuming Primary Responsibility for Special Populations and Beneficiaries with Specific Conditions

Most interviewees identified the near term need for models focused on special populations and beneficiaries with specific conditions that require intensive specialty care (e.g., certain cancers, end-stage renal disease). Stakeholders report that many ACOs recognize the value of specialty capabilities to manage special populations, and specialists often take over for primary care in a mutually beneficial arrangement. For example, cancer care requires high visit frequency and expertise in a rapidly changing clinical area. Together, these challenges may make comprehensive specialty care management by a primary care practice inefficient and less effective.

Models that are testing care delivery to special populations or those with specific conditions, such as the Oncology Care Model and the Kidney Care Choices model, can enable effective management of beneficiaries’ needs, which differ significantly from the broader beneficiary population. The long-term approach to supporting care for these populations—whether attributed to an accountable entity that can manage the total cost of care using the right data analytic supports or managed through a longitudinal, specialty accountable arrangement designed for these populations - depends to a great extent on how the delivery system continues to reorganize over the next five to ten years.

4.  Beneficiaries with Complex Conditions May Benefit from Specialists Integrated into Primary Care Delivery Pathways

As ACOs are able to assume greater accountability, they will also have greater flexibilities to manage complex populations, such as beneficiaries with serious illness. This will likely require integration of specialists into their primary care delivery system. The extent of integration may vary to reflect the ACO’s capabilities and the clinical needs of the beneficiaries they serve. Integration is becoming more common in areas such as primary care and behavioral health integration. Building specialty care within ACO or primary care pathways, potentially using mechanisms such as behavioral health integration billing codes or collaborative care codes, may also prove successful for other specialty care domains, such as cardiology, neurology, and palliative care. We look to ACOs to further integrate primary and specialty care in population-based models in a way that is less burdensome for beneficiaries. 

Looking Forward to a Future of Person-Centered Care

The CMS Innovation Center is building on these lessons, challenges, and barriers to test models that provide tools, supports, and financial incentives that will enable greater integration of primary and specialty care to meet the needs of an increasingly complex population of beneficiaries. Integrated and coordinated care for beneficiaries are essential features of a health system that achieves equitable outcomes through accountable, high-quality, affordable, person-centered care.

Table 1. Summary of Key Findings, Challenges, and Considerations for Future Model Design


Key Findings

Major Challenges

Potential Design Considerations

Finding #1: Providing Data on Specialist Performance and Enhancing Data Sharing across Practices would Facilitate Integration with Primary Care


  • Performance data on specialty providers could inform referrals of higher clinical value by primary care providers, but data are often not readily available and/or actionable.
  • Sharing reliable, actionable data on the performance of specialists could inform referrals from primary care. 
  • Data sharing and aggregation from multiple payers could provide the volume of episodes needed to generate meaningful, reliable profiles..
  • Current payment incentives within Medicare fee-for-service place limited emphasis on the referral process and follow-up between primary care and the specialists, resulting in a lack of data and infrastructure to coordinate care and to close referral loops..
  • E-consults, platforms that support screening of beneficiary specialty care needs, and other tools that facilitate data exchange could increase the proportion of referrals of high clinical value and help ensure effective follow-up after the referral visit.
  • Incentive design may encourage use of informal collaborative care agreements between primary care providers and specialists that outline referral protocols and care transition and management responsibilities, among other processes.
Finding #2: Episode-based Payment Models Can Be Designed to Align Incentives between Specialists and ACO Initiatives
  • ACOs, hospitals, and/or specialty physician group practices often target the same savings opportunities in shared savings models within and across models, which can weaken or create competition for incentives among providers responsible for outcomes and spending for the same beneficiary.
  • Compensation for providers is often still tied to productivity with volume-based incentives that undermine value-based care goals that aim to move away from fee-for-service.
  • Episode-based payment models could be tested to fill gaps for beneficiaries where accountable care entities may not be available.
  • To avoid selective provider participation associated with voluntary models, examine whether mandatory models can increase quality and access for beneficiaries and increase provider participation, without negatively impacting those who care for underserved populations.
  • Implementation of episode-based payment models and assigning responsibility for the episode of care may need to vary according to the degree of risk an ACO has assumed.
Finding #3: Many ACOs and Primary Care Practices Recognize Value in Specialty Care assuming Primary Responsibility for Special Populations and Beneficiaries with Specific Conditions Certain populations of beneficiaries, including those with cancer, end stage renal disease (ESRD), and chronic kidney disease (CKD), often require intensive, longitudinal treatment by a specialty care provider that is their primary source of care.
  • Accountable care models for special populations may enable more effective management of certain beneficiaries than a primary care-based ACO.
  • As more ACOs form and can assume greater risk, special populations may be able to be more effectively integrated into ACOs.

Finding #4: Beneficiaries with Complex Conditions May Benefit from Specialists Integrated into Primary Care Delivery Pathways


  • Adding more providers to a beneficiary’s care team can create friction and lack of transparency over each provider’s roles and responsibilities..
  • Lack of flexibility in fee-for-service prevents the delivery of coordinated, person-centered care.
  • Lack of access to specialists, especially for Medicaid beneficiaries, prevents delivery of person-centered care
  • Until more ACOs can assume full risk, collaborative care codes, as currently used in behavioral health, could support integration and co-location of some specialty care, such as palliative care. 
  • Total cost of care or high-risk ACOs provide the flexibility to integrate specialists into care delivery pathways in a way that can increase access and best meet the needs of beneficiaries.



[i] Barnett ML, Bitton A, Souza J, Landon BE. Trends in Outpatient Care for Medicare Beneficiaries and Implications for Primary Care, 2000 to 2019. Ann Intern Med. 2021 Dec;174(12):1658-1665. doi: 10.7326/M21-1523. Epub 2021 Nov 2. PMID: 34724406; PMCID: PMC8688292.

[ii] Martin S, Phillips RL Jr, Petterson S, Levin Z, Bazemore AW. Primary Care Spending in the United States, 2002-2016. JAMA Intern Med. 2020 Jul 1;180(7):1019-1020.

[iii] Baum A, Bazemore A, Peterson L, Basu S, Humphreys K, Phillips RL. Primary Care Physicians and Spending on Low-Value Care. Ann Intern Med. 2021 Jun;174(6):875-878.

[iv] Mafi JN, Reid RO, Baseman LH, Hickey S, Totten M, Agniel D, Fendrick AM, Sarkisian C, Damberg CL. Trends in Low-Value Health Service Use and Spending in the US Medicare Fee-for-Service Program, 2014-2018. JAMA Netw Open. 2021 Feb 1;4(2):e2037328.

[v] Center for Medicare and Medicaid Innovation. Bundled Payments for Care Improvement (BPCI) Initiative, Model 2-4. Accessed March 27, 2022.

[vi] Center for Medicare and Medicaid Innovation. Bundled Payments for Care Improvement Advanced (BPCI Advanced) Model: Model Years 1& 2 (October 2018 through December 2019). Accessed March 27, 2022.

[vii] Centers for Medicare and Medicaid Services. Quality Payment Program. 2022 Call for Cost Measures Fact Sheet. Accessed March 27, 2022.

[viii] Duseja R, Andress J, Sandhu AT, et al. Development of Episode-Based Cost Measures for the US Medicare Merit-based Incentive Payment System. JAMA Health Forum. 2021;2(5):e210451.