5
POLICY TO SUPPORT THE SPREAD AND SCALE OF CARE MODELS
Fictional patient vignette: Andy is a 75-year-old man whose arthritis, anxiety, and heart disease make it difficult for him to be on his feet and out of the house for long periods of time. He has frequent doctor appointments, and he feels lucky that between his Medicare and Medicaid benefits, most of his costs—for his general practitioner, pain management specialist, psychiatrist, and cardiologist—are covered. Often, the most difficult part of his health care routine is trying to figure out what is covered under Medicare and Medicaid, and by whom. Even though the staff at his various doctors’ offices are willing to help him, Andy still spends hours trying to figure out what he is eligible for, and whether Medicare or Medicaid or both will pay for it. Andy doesn’t understand why his Medicare and Medicaid coverage are so separate. They’re both part of the federal government, aren’t they?
While a range of programs have been shown to improve care for high-need patients, a variety of barriers have prevented successful programs from expanding beyond a single site or led them to be discontinued after an initial trial. These barriers are complex and span a range of factors: health system fragmentation, high implementation start-up costs with uncertain returns on investment; the challenge of integrating (and paying for) social and other nonmedical services with medical care; the difficulty of replicating care models developed in one setting across disparate settings; workforce training issues; and the need for appropriate quality measures and a data infrastructure to inform those measures. A number of barriers, however, have actionable solutions, with the key foundational issue often being federal-, state-, and health-system-level policies that exacerbate the challenges of caring for high-need patients. This chapter explores areas in which policy initiatives could accelerate the spread
and scale of care models for high-need patients—particularly the programmatic integration of social supports and medical care—through developing a workforce to deliver comprehensive health care, expanding and realigning payment policies, reexamining quality measurement, and improving the data infrastructure.
Although the committee recognizes that prevention of the chronic conditions associated with high-need patients—through both public health and medical interventions—is a critical pursuit, this publication does not address policies that focus on population health and prevention. Many elements central to population health strategies (e.g., integration of social services and medical care) are relevant to the issues in this publication, however, and considering steps to increase prevention efforts could perhaps be the focus for future work by those concerned with improving the lives of high-need patients.
SPREADING AND SCALING SUCCESSFUL CARE MODELS
In his presentation at the second workshop, Arnold Milstein from the Clinical Excellence Research Center at Stanford University noted that an important barrier to spreading and scaling care models is the complexity of health care delivery systems. In Milstein’s experience, system leaders are interested in adopting a new care model if the model would affect only one area of operations, such as primary care or neurology. Any idea for lowering the cost of providing better care that required cooperation among and across multiple departments, however, was typically rejected by system executives. “We are still in a situation where systems are challenged by the complexity of the job they face simply in delivering care, let alone improving it,” he said. He also noted the challenge and cost of adapting a model developed to serve one particular subpopulation of high-need patients in one specific setting to another subpopulation in a different health care setting.
One approach Milstein suggested to dealing with these challenges would be to create a network of improvement communities that would bring together parties interested in scaling models of care to chart what does and does not work for various settings. Creating a network of improvement communities could accelerate the transfer of insights about better care methods and scaling approaches, and he credited organizations such as the Institute for Healthcare Improvement for starting to engage in these types of learning activities. Milstein also suggested that spreading and scaling efforts might benefit from a research effort to apply simulation modeling, using continuously generated patient data, to identify actions to improve care and lower cost. The resulting simulation models would then be available as a national asset.
Though the challenges to spreading and scaling models of care are significant, research has identified helpful tactics for spread and scale. During the first workshop, Deborah Peikes from Mathematica Policy Research discussed some of the factors for successful scaling that she and her colleagues found in studies conducted for the Centers for Medicare & Medicaid Services (CMS). The identified success factors included substantial financial incentives; support from multiple payers, such as coordination and aligning spending, technical assistance, data feedback, staff support, and reporting requirements; adapting data and technical assistance to reflect the considerable diversity of practices, health systems, markets, and patients; and monitoring or auditing, particularly if the funder bears risk, to ensure that programs are implemented as intended (Dale et al., 2016; Taylor et al., 2015a).
Despite Milstein’s observation that health system leaders are reluctant to adopt models that require widespread changes in a health system, Peikes and colleagues found that practices that spread a model broadly throughout the practice were the most successful at implementing the model. Strong and consistent leadership is also critical for successful model adoption, and technical assistance on leadership and teamwork may help spread interventions. She noted that implementing a care model piecemeal in an organization puts too much burden on clinician champions, leads to unclear roles and responsibilities, and does not encourage the development of a learning health system.
At the first workshop Lisa Mangiante from the Pacific Business Group on Health (PBGH) discussed lessons learned from efforts to spread the Intensive Outpatient Care Program (IOCP) model that Milstein and Alan Glaseroff developed for high-need patients in California’s Health Homes program. Between May 2012 and July 2015, this high-touch, care-coordinated, patient-involved, and team-based care model (see Appendix A) has been spread to 23 delivery system partners in Arizona, California, Idaho, Nevada, and Washington, and it encompasses some 500 practices and 15,000 patients. Of the 23 health system partners who participated in this CMS-funded scaling project, 20 are sustaining this model in their operations by integrating the model into their overall population health strategies. Mangiante explained that there is a great deal of payment and organizational variation among the partners, including the Medicare Shared Savings Program, Pioneer accountable care organizations (ACOs), Medicare Advantage Plans, and fee-for-service operations. Partners included independent practice associations, medical foundations, and both integrated and nonintegrated systems.
When discussing what was involved in scaling the IOCP model, Mangiante said PBGH started with what it called the A List of medical groups in California: those that already had track records of innovation, did well with innovation,
had an infrastructure in place to implement this model, and had supportive leadership. Those A List groups had IOCP operational within 10 months. Once Mangiante and her colleagues had successful experiences working with the A List groups, they added less aligned and less sophisticated providers who required ongoing support.
Mangiante noted the following key characteristics that enable this model to scale and launch successfully at new sites, many of which overlap with the framework of attributes for successful care models described in Chapter 4: provider-hospital integration and integration into a larger population health strategy; adapting to the local environment after meeting core requirements; a strong analytical capability enabling aggressive patient monitoring with regular feedback; hiring effective care coordinators; identifying physicians and nurses who welcome disruptive innovation; training staff in didactic, interactive, and peer-to-peer learning; giving physicians a role in patient selection; developing intensive local patient outreach with close contact between physician and patient; ensuring strong support from senior leadership; having dedicated physician champions; and targeting those most likely to benefit from this program. With regard to sustaining the model once it is operational, she said it is critical for the programs to involve multiple payers as a means of creating a solid revenue stream.
Given these types of lessons, an important consideration for spreading and scaling successful models of care for high-need patients could be having payers and health systems work both separately and in tandem to more aggressively implement these models. Gerard Anderson from Johns Hopkins Bloomberg School of Public Health reported at the third workshop that payers could help foster success by tying payments to improving the patient experience, improving the health of populations, and reducing per capita cost of care; they could also target resources to populations most likely to benefit from these models as well as in a manner consistent with best practice. Health systems, meanwhile, could work with peers to identify promising models and work with payers to develop alternative contracts that pay for services not covered by fee-for-service arrangements (see page 86 for more on payment policies). Health systems could also commit to adequate investments in training, practice redesign, and information technologies. Working together, payers and health systems could explore the use of the patient taxonomy as a tool to match patient groups to tailored care models that better meet individual characteristics, needs, and challenges.
Anderson also pointed to the need for more research on developing programs that can be adopted widely in a variety of settings. To help inform efforts to spread and scale effective models of care, he suggested more research in areas such as identifying people at high need in actual practice settings, identifying
individuals who are likely to be high-need patients in the future, identifying the best methods of care coordination, developing cost-effective implementation practices, and developing methods for effectively integrating medical and social services. As he noted, the importance of integrating social services and medical care is embedded in the other four policy areas—workforce development, payment policy, quality measures, and data infrastructure.
INTEGRATION OF SOCIAL SUPPORTS AND MEDICAL CARE
A recent analysis by McCarthy and colleagues found that comprehensive transitional care and case management involving patients and their caregivers after hospitalization is an important integrative feature of successful care models for high-need patients (McCarthy et al., 2015). As part of an effort to provide a framework for understanding the nature and extent of integration in programs that integrate long-term services and supports (LTSS) with medical care and behavioral health, the Long-Term Quality Alliance concluded that a critical element of a fully integrated model includes having a plan for health systems to accept responsibility for integrating medical care, postacute care, behavioral health care, pharmaceutical care, transitional care, and LTSS, including transportation and housing (Long-Term Quality Alliance, 2016).
As noted in Chapter 2, functional limitations are an important contributor to the disproportionate share of health care spending in the United States associated with high-need patients. In his presentation at the first workshop, Bruce Chernof from The SCAN Foundation noted that addressing a person’s medical needs without also addressing functional limitations will have little impact on the cost or quality of care for these high-need patients. He also pointed out that social determinants of health,13 which have been largely considered beyond the purview of the medical system, can have the biggest effect on ameliorating functional limitations. This is why successful models for improving care for high-need patients, such as those described in Chapter 4, often include the integration and delivery of social services in addition to better coordinated medical care. Enacting policies to facilitate the integration of social services and medical care is crucial for obtaining better outcomes for high-need individuals.
__________________
13 The Department of Health and Human Services’ Office of Disease Prevention and Health Promotion defines social determinants of health to be social, economic, and physical conditions of an individual’s life and surrounding environment, such as income, house, and nutritional factors, that impact the health outcomes of individuals (Secretary’s Advisory Committee on Health Promotion and Disease Prevention Objectives for 2020, 2010).
A report prepared by Taylor and colleagues for Blue Cross Blue Shield of Massachusetts Foundation (Taylor et al., 2015b) points to the extensive scientific literature showing that nonmedical factors can play a substantially larger role than medical factors in health and health outcomes. On page 3, the report states there is “strong evidence that increased investment in selected social services as well as various models of partnership between health care and social services can confer substantial health benefits and reduce health care costs for targeted populations.” Hayes and colleagues at the Bipartisan Policy Center (BPC) recently made a similar observation with regard to high-need Medicare patients and dual-eligible patients who receive care from both Medicare and Medicaid programs (Hayes et al., 2016). As an example, Hayes and colleagues (2016) noted that a recent evaluation of the Minnesota Senior Health Options program, which coordinates care for dual-eligible patients, found that this program increased the use of home- and community-based LTSS while decreasing hospitalizations for treatment of chronic illnesses and days spent in nursing homes (Anderson et al., 2016).
Two studies from Bradley and colleagues at the Yale School of Public Health further emphasize the importance of providing social services to improve health outcomes. The first study compared national spending on health services and social services among Organisation for Economic Co-operation and Development (OECD) countries and found that a larger ratio of social expenditures to health expenditures was significantly associated with better health outcomes (Bradley et al., 2011). OECD data used in this study (OECD, 2009) show that, compared to the United States, most highly developed countries spend a greater percentage of gross domestic product on social support and a smaller percentage on health care. In the second study, a state-to-state comparison of spending on social services and health care between 2000 and 2009 found that states with higher ratios of social service spending to health care spending14 had better health outcomes and fewer days with functional limitations (Bradley et al., 2016). This team’s analysis of 74 studies examining the impact of various health outcomes and health care costs found that three types of services were particularly important: supportive housing, nutritional support such as in-home meals for older adults, and case management and outreach programs.
While these and other studies reveal the important role social supports play in achieving the best health outcomes for high-need patients, it can be difficult to integrate social service and medical care programs into a seamless system, often because there are separate funding streams that hinder integration. For
__________________
14 This ratio was calculated as the sum of social service spending and public health spending divided by the sum of Medicare spending and Medicaid spending.
example, for the 10.3 million dual-eligible beneficiaries—many of whom are also high-need patients—Medicare is the primary payer for acute and postacute care services, while Medicaid covers services not included in Medicare benefits, such as case management services, transportation to medical appointments, personal-care services to help patients with functional limitations, and other LTSS. Too often, according to the work from Hayes and colleagues (2016), the separation of Medicare and Medicaid benefits and the “carving out” of certain Medicaid benefits from managed care contracts can lead to a fragmented care model in which the beneficiaries and their family caregivers must navigate multiple plans or payers depending on the type of service provided (Hayes et al., 2016a).
While this chapter addresses the negative effect that current payment policies have on integration in more detail below, one step the federal government, states, and payers could consider taking would be to revise financial incentives and organizational systems in a manner that recognizes the importance of providing social supports in conjunction with medical care. Currently, said Anderson, health systems and payers invest money for social services primarily when doing so saves money for the medical care system, even though providing social services for high-need patients has importance aside from cost savings. Moreover, savings accrued from social service investment are often not reinvested in social services, missing an opportunity to provide even better care. One caution to exercise when integrating social services and medical care is to not “medicalize” social services by making them the responsibility of health care systems. Doing so would create the risk that all services aimed at improving outcomes for high-need patients become “health care” and therefore subject to the administrative and payment rules that govern health systems.
While there are many system constraints to integration, there are opportunities nonetheless to better link medical and social services. For example, the Office of the Assistant Secretary for Planning and Evaluation could take the lead in overseeing integration efforts, perhaps in conjunction with an interagency task force involving the Departments of Agriculture, Health and Human Services, Transportation, and Urban Development that would work to embed health in all policies.15 The federal government will likely need to engage in a strategy coordinated with state leadership to incentivize provision of evidence-based social support services in conjunction with the delivery of medical services. In addition, the nation would be well-served if the federal government studied the
__________________
15 Health in all policies is a strategy for addressing the complex factors that influence health and equity, including educational attainment, housing, transportation options, and neighborhood safety. Additional information is available at: http://www.naccho.org/uploads/downloadable-resources/Programs/Community-Health/HiAP-FAQ-Final-12-04-24.pdf.
impact of providing social services on health outcomes for high-need patients and encouraged states to support integration of social support services through “no wrong door” approaches that link patients to needed services regardless of how or where they enter health care or social services systems.
State governments, which control Medicaid spending, can also play a role in fostering the integration of health and social services. McGinnis and colleagues at The Commonwealth Fund developed a policy framework to help states move beyond isolated pilot efforts and establish the infrastructure necessary to support ongoing integration of health and social services, particularly for Medicaid beneficiaries (McGinnis et al., 2014). Their framework focuses on creating a statewide integrator to assume responsibility for ensuring coordination and communication across state-level services, establishing a robust set of tools to measure health outcomes and costs and share data among health systems, and developing long-term financing sources and payment models with incentives to encourage ongoing integration.
EXPAND AND ALIGN PAYMENT POLICIES
As multiple speakers at the first two workshops noted, payment policies that misalign financial incentives—particularly those that reimburse providers on a fee-for-service basis and that fail to pay for social services benefiting high-need patients—are perhaps the most prominent barrier to the widespread adoption of successful models of care for high-need patients. Many workshop participants stated the need for new payment policies that incentivize integration of social services and medical care and improved outcomes for high-need patients: Melissa Abrams from The Commonwealth Fund; Alan Glaseroff from Stanford Coordinated Care and Stanford School of Medicine; Bruce Chernof from The SCAN Foundation; Lisa Iezzoni from Harvard Medical School and the Mongan Institute for Health Policy at Massachusetts General Hospital; Robert Master from Commonwealth Care Alliance; John O’Brien from CareFirst Blue Cross Blue Shield; Peter Long from the Blue Shield of California Foundation; and Rahul Rajkumar from the Center for Medicare & Medicaid Innovation. A research synthesis compiled by The Commonwealth Fund also concluded that a lack of reimbursement under fee-for-service payment policies for providing care coordination and social supports is a major obstacle to spreading and scaling patient-focused care models for high-need patients (McCarthy et al., 2015).
Significant improvements have been made in paying for care coordination, and there is an increasing recognition that social supports are important components of effective care plans for high-need patients. Many insurers, including
states and the federal government, are starting to embrace value-based purchasing that includes paying for care delivered outside of the traditional medical silo (Bachrach et al., 2014; Hamblin et al., 2011). In a recent perspective on the urgency of caring for high-need, high-cost patients, Blumenthal and colleagues point out, for example, that the Affordable Care Act catalyzed the formation of 838 ACOs covering more than 28 million people (Blumenthal et al., 2016b). At least some of these ACOs have allocated independent resources—not reimbursed by Medicare—toward providing short-term housing upon hospital discharge as a means of reducing hospital readmissions for vulnerable patients and keeping Medicare per-member spending below predetermined spending benchmarks (Viveiros, 2015). They also note that under the Medicare Access and Children’s Health Insurance Program Reauthorization Act (MACRA), physicians will face strong incentives to participate in alternative, value-based payment models.
Fee-for-service Medicare Advantage Plans now pay for care coordination, and Medicare managed care plans have recognized the importance of care coordinators for high-need patients. In addition, CMS has granted an increasing number of Medicaid demonstration waivers for states that want to provide greater flexibility in covering community-based services as a means of reducing health care costs. Anderson noted that state and local public agencies are developing programs and task forces to support integration of social service and medical care programs. Examples include housing-related interventions such as the National Governors Association’s Housing as Health Care program16 and state-led Balancing Incentives Program and no wrong door approaches to provide access to LTSS for all populations and payers.17
The analysis conducted by Hayes and colleagues at the BPC and the recommendations they developed were intended to accelerate delivery system reform (Hayes et al., 2016a). Their analysis focused on the 10.3 million dual-eligible patients. Though only 20 percent of Medicare beneficiaries and 14 percent of Medicaid beneficiaries are dual-eligible individuals, they account for 35 percent of Medicare spending and 33 percent of combined federal and state spending on Medicaid. According to BPC’s analysis, some 87 percent of dual-eligible beneficiaries have multiple chronic conditions, 54 percent have at least one behavioral health condition or cognitive impairment, and 29 percent have two or more limitations affecting activities of daily living. BPC’s team noted that provider organizations seeking to improve care integration for high-need patients
__________________
16 For more information, see https://www.nga.org/cms/home/nga-center-for-best-practices/center-publications/page-health-publications/col2-content/main-content-list/housing-as-health-care-road-map.html (accessed on July 31, 2017).
17 For more information, see http://www.balancingincentiveprogram.org (accessed on July 31, 2017).
frequently run into the complex maze of federal and state reimbursement rules that preclude payment for, and in some cases coverage of, services that health providers believe could avert costlier emergency or hospital inpatient visits, which are major driving forces for the high costs associated with high-need patients.
To best appreciate the challenges arising from dual-eligible status, it is necessary to understand how dual-eligible patients receive their benefits from these two distinct programs. Although both Medicare and Medicaid are authorized under the Social Security Act, the federal government administers Medicare, while federal and state governments jointly finance Medicaid. States cover certain mandatory benefits under Medicaid, while other services are optional and coverage is determined on a state-by-state basis. As Hayes explained in her presentation at the third workshop, Medicaid covers LTSS, including many services that deal with functional limitations. As of June 2015, only some 20 percent of dual-eligible individuals were enrolled in the type of organized systems of care that blend social services and medical care, such as Medicare managed care plans, Program of All-Inclusive Care for the Elderly (PACE), and Dual Eligible Special Needs Plans (D-SNPs).
In their report, Hayes and colleagues (2016) state that the specific care delivery model and state implementation of the model will likely determine whether full integration of Medicare and Medicaid services will improve quality and lower the total cost of care for dual-eligible beneficiaries. There is mounting evidence, however, that integration does improve quality and value. As noted earlier in this chapter, for example, dual-eligible patients enrolled in Minnesota’s Senior Health Options program had fewer hospitalizations and emergency department visits and increased use of home- and community-based LTSS compared to individuals receiving benefits through Medicare fee-for-service plans (Anderson et al., 2016).
BPC’s team pointed out that there is much to learn about integrating care for dual-eligible beneficiaries. Hayes and colleagues (2016) explained that only a relatively small number of states have more than a few years of experience fully integrating Medicare and Medicaid services for dual-eligible beneficiaries over age 65. Even fewer states have experience with the under-65 population, which has higher-than-average rates of untreated behavioral health issues and/or prevalence of homelessness according to Medicare managed care sponsors. Summarizing BPC’s recommendations, Hayes explained that they include changing existing reimbursement structures, consolidating regulatory authority for dual-eligible programs within the Medicare-Medicaid Coordination Office at CMS, and building on lessons learned through implementation of existing programs and demonstrations to develop a consolidated framework for programs serving dual-eligible beneficiaries. Critical to that framework is the ability to
combine Medicare and Medicaid financing streams into an integrated benefit structure that allows flexibility in benefit design to address patient needs. See Box 5–1 for selected excerpts of BPC’s recommendations.
While BPC’s recommendations aim to harmonize Medicare and Medicaid benefits to improve care of dual-eligible individuals, other efforts are under way to take advantage of Medicare and Medicaid programs that enable providers, payers, and state agencies with opportunities to test delivery system innovations that improve outcomes and patient experiences while increasing the value of care. Health Homes, Patient-Centered Medical Homes, Community Health Teams, and Transition Care Models are among the many programs using value-based strategies to replace traditional fee-for-service payment models. Recently, the Center for Health Care Strategies outlined approaches to value-based payments (Houston, 2016), some of which supplement fee-for-service payments, and others that replace them. Foundational payments, for example, are a flat or per-member per-month (PMPM) fee to reimburse providers for upfront investments they make to better coordinate care. Meaningful-use payments from the Office of the National Coordinator for Health Information Technology, which aim to reimburse practices for installing electronic health record systems, are an example of a foundational payment. Pay-for-performance models supplement fee-for-service payments by rewarding providers with performance incentive payments linked to outcomes and patient satisfaction metrics. The Medicare Physician Group Practice Demonstration is an example of this type of value-based payment model. Other approaches include bundled payments for a set of services for a specific care intervention, a common mechanism that state Medicaid programs use; shared savings programs that manage the total cost of care based on risk, such as the Medicare Shared Savings Program and state Medicaid ACOs; and capitation and global payments that pay a single PMPM fee to a provider to cover all of an individual’s care, a model used by Oregon’s Coordinated Care Organizations.
These types of payment methodologies can incentivize care investments in evidence-based best practices for high-need individuals and reduce the incentives that lead to ineffective and uncoordinated care. To increase their effectiveness, however, such payment models could be constructed to account for the increased financial risk associated with caring for high-need patients (Barnett et al., 2015). Health systems that focus exclusively on high-need patients can be at a market disadvantage and may be financially unstable.
They may be further disadvantaged by value-based purchasing arrangements that do not recognize the unique requirements of the high-need population. It may also be the case that capital markets would be less likely to finance organizations to better serve high-need patients because of the potential for competitive
disadvantage compared to organizations that do not serve large numbers of high-need patients.
One issue, addressed by Blumenthal and colleagues in a discussion paper from the National Academy of Medicine’s series of discussion papers Vital Directions for Health and Health Care (Blumenthal et al., 2016a), is that most ACOs and performance- and risk-based plans still pay clinicians on a fee-for-service basis (Bailit et al., 2015). The authors of this discussion paper note that if individual providers or practice sites do not feel accountable for health outcomes, population health, and value, the diffusion of promising practices and models of care will be slow. Another concern the authors of this paper noted is the misalignment between investment and savings: too often, the savings realized by a successful care model accrue to payers, even though it is the providers who are expected to cover the up-front costs of staff training and other investments a program requires (Hong et al., 2014). Even if savings are shared with providers, the time for these complex delivery system reforms to produce savings can be several years (Jones et al., 2016), which can discourage providers from making the necessary initial investments. Ashish Jha from the Harvard T.H. Chan School of Public Health noted at the third workshop that realizing a return on investment with even good models is a long process. Google, he pointed out, took 8 or so years to become profitable, and to expect these models to yield large returns in 2 to 3 years will lead to disappointment. Similarly, Peikes and colleagues stated, “The providers we speak to report that it takes a year and a half or longer for interventions to really click.” In the third workshop, David Atkins of the Department of Veterans Affairs underscored the need for support for the long-term experiments to demonstrate meaningful returns on investments, particularly given the reluctance of health system administrators to maintain programs that are not yielding short-term benefits.
When discussing payment policies, Anderson said in the third workshop, it is important to remember that just as there is not one kind of high-need patient, corresponding flexibility will be needed when it comes to payment models that incentivize high-value care for high-need patients. In particular, reimbursements for care coordination will have to reflect the different levels of patient need that require different levels of care coordination and that entail different degrees of risk. One issue that a breakout group in the second workshop raised was the need to allow organizations to have some flexibility in the benefits they offer as long as they can demonstrate that they are providing high-quality care for all of the high-need individuals in their care, not just a selected few. Flexibility could allow providers and health care organizations to target individuals who are most likely to benefit from particular delivery
models if the focus is on improving quality of care rather than squeezing cost savings out of the system. In that regard, said John O’Brien from CareFirst Blue Cross Blue Shield, payment models should incentivize targeting patients who are most likely to benefit from the right interventions.
WORKFORCE FOR COMPREHENSIVE HEALTH CARE
Both Peikes and Mangiante noted the importance of education and training in successful scaling efforts and the integration of services. Many clinicians, however, are not well trained to address the needs of high-need patients. Anderson commented that medical schools tend to emphasize “body parts” instead of the whole person and that nurses, who are often the care coordinators in these model programs, have little training in care coordination. In fact, he pointed out, care providers of all types—physicians and nurses, medical paraprofessionals, social workers, and housing and employment professionals—need to receive training on caring for and interacting with high-need individuals.
Thomas-Henkel and colleagues, in a study commissioned by the Robert Wood Johnson Foundation (Thomas-Henkel et al., 2015), noted that barriers to the spread and scale of care models for high-need patients include gaps in the training of current and newly graduated clinicians, a lack of interprofessional education among team members, low reimbursement rates that may limit recruitment efforts, and the need to develop more effective models for preventing and managing staff burnout given the professional and emotional challenges this work can entail. They highlighted the opportunity for academic health centers and professional societies to collaborate on developing new training and certification opportunities, particularly those that encompass team-based approaches and training in behavioral health, substance use disorders, and complex psychosocial factors. They also raised the point that there are new models of supervision involving the entire spectrum of traditional and nontraditional health care team members that care models are drawing upon to better serve high-need patients.
There is, therefore, a need and an opportunity for education and training to be integrated into the process of spreading and scaling any given model. Many models that have proven successful at improving care for high-need patients already put an emphasis on social supports, a trend noted by Molly Coye from the Network for Excellence in Health Innovation. As mentioned in Chapter 4, adoption of these models can lead to substantial changes in workforce roles as evidenced by assigning important roles in the care teams that integrate the broad range of social and behavioral health services high-need patients require to professionals who are often not considered key members of a health care team.
Credentialing programs, particularly for nontraditional health workers such as community health workers and peer support providers, could be developed to encourage workforce development to support high-need patients. Research has shown that properly trained community health workers can play a unique role in helping high-need patients navigate the health system, obtain necessary supportive resources, and build self-efficacy and health literacy; by doing so, they can improve patient experiences and outcomes and reduce hospital readmissions (Adair et al., 2012; Adair et al., 2013; Davis, 2013; Kangovi et al., 2014). For example, an Oregon program for high-need Medicaid patients, in which care teams were led by a nurse and two community health workers, reduced emergency department utilization from 78 percent in 2011 to 59 percent in 2013 (Takach and Yalowich, 2015).
Academic institutions, health systems, and other educators could develop curricula on the treatment and social support needs of high-need patients, including training on team-based care, patient engagement, care coordination across health and social sectors, and the social determinants of health. Key workforce sectors in need of training would include clinicians, nurses, physician assistants and other medical paraprofessionals, mental health professionals, social workers, pharmacologists, substance abuse providers, community health workers, peer providers, law enforcement officers, and housing and employment service providers. In particular, anyone involved in case management would benefit from special training that would include field training to observe delivery of evidence-based practices. Rajesh Davda from Cigna noted that physicians and nurses participating in Cigna’s model program for high-need patients were generally poorly informed about care coordination when the program started. Once staff members were trained on matters of care coordination, they became the most effective instructors for training additional team members. This result prompted Cigna to develop learning collaboratives to foster workforce training. Anderson commented that high-need patients would also benefit from improved training for students in health care fields, which he believes would reduce the stigma associated with people with complex illnesses among clinical trainees, active professionals, and health system administrators, and produce a culture shift in the world of health care.
REEXAMINE QUALITY MEASUREMENT
As value-based purchasing becomes more common, it will be essential to use quality measures—and the data that inform those measures—to improve care and offer incentives for providers to treat high-need patients. As Shari Ling from
CMS noted in her presentation at the third workshop, the proper use of measures creates transparency on cost and quality of care. Most quality measures currently in use, however, focus on specific conditions and whether methods of care for those conditions are effective. Ling pointed out that meaningful quality measures are too often setting-specific rather than being aligned with patient-centered outcomes that span various settings (Conway et al., 2013). Discussion during the second workshop noted that the current system of metrics is not designed in a way that encourages providers to organize care in the most effective manner.
While condition-specific measures are important, high-need patients are more than the sum of their individual diseases, and they have additional concerns beyond the appropriate care for each specific condition. To better reflect this reality, measures for assessing the performance of care models for high-need patients could indicate the degree of care coordination, quality of life, independence, and overall mental and physical health status. In her presentation at the third workshop, Helen Burstin from the National Quality Forum highlighted the need for measures based on patient-reported outcomes, patient involvement in the decision-making process regarding their care, and the quality of home and community-based services. In his remarks at the third workshop, Rick Kronick from the University of California, San Diego, added that measures should assess whether systems are stinting on some aspects of care, whether patient preferences are elicited and respected, and whether the communication between clinician and patient is of adequate quality. During the same workshop, Richard Frank from Anthem Blue Cross Blue Shield wondered if it would be possible to measure patient behavior in some manner that would provide a better understanding of engagement and motivation to change.
The Health and Medicine Division of the National Academies of Sciences, Engineering, and Medicine has undertaken a study of the relationships between social risk factors and performance measurement (National Academies of Sciences and Medicine, 2016a, 2016b), which will also inform metrics focused on the social determinants of health for the high-need population. According to Burstin, the availability of more appropriate measures of care relevant to high-need patients will enable health care financing to move beyond reliance on claims-based risk adjustment and instead enable differentiation of risk within clinical conditions and risk-based grouping of multiple chronic conditions. Such measures could also better account for unmeasured clinical complexity, such as patient frailty, disability, poor functional status, and multiple chronic conditions. Among existing measures, recent evidence suggests that because reliable case mix accommodation approaches are still evolving, hospitals caring for a higher proportion of patients with complex medical problems tend to fare worse on certain quality measures,
such as readmissions (Joynt, 2013, 2017), and may experience high penalties under value-based purchasing programs, potentially creating a disincentive to caring for these individuals.
Burstin and other participants at the third workshop voiced their concern that the proliferation of measures and “measurement for measurement’s sake” has become a burden to providers. A 2016 National Academy of Medicine Perspective, Observations from the Field: Reporting Quality Metrics in Health Care (Dunlap et al., 2016), offered the same concern. As David Dorr from the Oregon Health & Science University noted, it takes discipline to be parsimonious with measures. It is important for payers and health systems to choose measures that reflect realistic quality and accountability goals and to understand that programs may not demonstrate marked improvements for several years. Kronick remarked that measures should not be the only means used to improve quality of care. In his opinion, public policies related to quality improvement should emphasize methods of enhancing professional intrinsic motivation while recognizing the role of organizations to promote and facilitate that motivation by providing systematic feedback to physicians, technical assistance, and opportunities for providers to collaborate on projects to improve care.
IMPROVE DATA INFRASTRUCTURE
Research shows that high-quality data and analytics are an essential component of effective models of care for high-need patients in that they are used to match high-need individuals with specific interventions (Bates et al., 2014; Bradley et al., 2016; Dale et al., 2016; Rajkumar et al., 2015). High-quality data are also needed to inform the types of measures discussed in the previous section. One major challenge Anderson noted is that there are many disparate systems that cannot easily share information, making it difficult to assess the requirements of high-need individuals and whether they are getting appropriate medical and social care. During the first workshop, Lisa Iezzoni from Harvard Medical School and the Mongan Institute for Health Policy at Massachusetts General Hospital said that reliable data are needed when identifying high-need patients to overcome the limitations of the diagnostic data by which chronic conditions are identified. International Classification of Diseases-Clinical Modification diagnostic codes may not fully capture disability, functional limitations, or frailty, while other measures of frailty and disability can carry biases, including cultural ones, or have gaming potential once reimbursements start being based on a particular measure. In addition, diagnostic claim codes may fail to capture the health of persons who have not received adequate care.
Iezzoni illustrated the problem of trying to use diagnostic claim codes to capture the health of persons who have not yet received adequate care by recounting the experience of the One Care program in Massachusetts, a dually capitated program begun on October 1, 2013, for Medicare and Medicaid enrollees ages 21 to 64. Capitation for Medicaid payments was set using rating categories that were initially assigned based on prior year MassHealth Medicaid claims for the program’s participants. As soon as the One Care program started, it became obvious that many enrollees had higher needs than were indicated in their assigned rating category, in large part because they had not had Medicaid claims from a prior year since they had not received services in that prior year. Program administrators went to MassHealth and explained that they were not able to provide needed services because the capitated payments were too low because of the improper ratings. The solution was to conduct a functional status assessment once a person was enrolled in the One Care program and upgrade the rating based on that assessment. Before the upgrade, 59 percent of the enrollees had the lowest rating—a minimal care level for this high-need population—while after the upgrade only 40 percent of the enrollees fell into that category. At the same time, the percentage of independently living individuals in the highest-need categories rose from 19 percent to 35 percent. At the time of the second workshop, reimbursements were matched closely to costs, allowing One Care to provide the services its clients required.
Electronic health record (EHR) data, combined with claims data, can provide some additional insights about high-need patients that can help with managing complex patient populations. At the second workshop, Paul Bleicher from OptumLabs, a division of UnitedHealth Group, described how his organization uses these combined data sources to characterize the natural history of disease and identify specific issues and conditions associated with the biggest costs. Researchers at OptumLabs have also been using these data to examine patient clustering. They created a model that identifies patients who are at the highest risk of hospitalization and uses machine-learning clustering technologies to segment the top 10 percent of these patients. This analysis, he said, can support efforts to personalize care based on specific patient profiles.
Data from EHRs can provide a finer-grained picture of different groups of patients. For example, EHR data analysis reveals a substantial difference between prescriptions written and prescriptions filled. In the case of patients with hypertension who are not following established guidelines, this type of analysis can show whether the problem rests with the patient or the physician. Natural language processing of EHR data can create structured variables that provide detailed pictures of laboratory test results as they relate to care management, leading to
the identification of possible drivers of hospital readmission, said Bleicher. He noted that significant amounts of granular information can be extracted from the EHR with natural language processing and used to gain a better understanding of patient outcomes. This value-added information includes clinical findings that are not available in claims data, such as preadjusted diagnostic and procedure information and temporal data about a patient’s stay in the hospital. In addition, clinical notes can be mined for details, such as the risk of falling, that are not available in the EHR’s structured data.
OptumLabs has been using this type of data analysis and mining to create predictive models that can help reduce hospitalizations. For example, a congestive heart failure predictive model uses a patient’s prior health care use and clinical findings such as blood oxygenation, laboratory results, and vital signs to predict the risk of future hospitalization over the following 6 months. Individuals in the 95th percentile of risk were contacted and brought into the clinic for further assessment and treatment, with the result that hospital admissions for all patients with congestive heart failure were reduced by 60 percent from a year earlier. Bleicher explained that with claims data alone, the model was inadequate.
There are policy impediments—particularly with regard to sharing behavioral health and substance abuse information—that act as barriers to coordinating care for high-need individuals. Bleicher noted that standardizing EHR data across different systems is a major challenge, as is the fact that individual physicians capture and record data differently, and coders will code records and turn them into claims differently. For example, Bleicher stated that his team found from mining EHR data that between 11 and 31 percent of patients who had no billing code for diabetes over a 3-year period were in fact being treated for diabetes based on EHR-recorded laboratory results and prescriptions for diabetes medications. Jha added that claims data are limited when it comes to identifying which individuals either use or have used long-term care. Medicare data, for example, do not include long-term care; there is not a good national dataset with any granularity on long-term care services and supports and social services for the Medicare population. Federal, state, and local governments could identify barriers that currently inhibit data flow among the clinicians and organizations treating high-need populations and work to minimize those barriers while respecting patient privacy and data security.
In her presentation at the second workshop, Sandra Wilkniss from the National Governors Association said that access to data is one of the main challenges that states face in crafting effective policies to support better care for high-need, high-cost patients. In fact, some states have difficulty accessing even their own claims data to conduct necessary analyses, in part because state laws are barriers to
effective data sharing. For the most part, explained Wilkniss, governors are taking a data-driven strategy to identify target populations primarily using Medicaid claims data, pharmacy data, and other types of high-level data to segment patient populations. A significant piece of this strategy involves what Wilkniss called “geospatial hot-spotting,” or identifying communities with a disproportionate share of high-need, high-cost patients. She noted that high-value health care systems with which she has interacted are using data to conduct rapid-cycle evaluations of their programs’ performance to continuously improve care and reduce costs.
In their work toward defining a state policy framework for integrating health and social services, McGinnis and colleagues at The Commonwealth Fund suggest that state- and community-level data-sharing tools could include integrated claims databases that link and share information across payers, service sectors, and provider networks (McGinnis et al., 2014). One example of such a tool is the Predictive Risk Intelligence System (PRISM) that Washington State developed to support care management for high-risk Medicaid patients (Court et al., 2011).
Integration of medical, behavioral, and social data, along with improved data sharing, is paramount to improving care for high-need patients. Integration and data sharing, however—whether across health and social services systems or within different components of the health care system—is a challenge that federal, state, and local partners could work together to address. One possible first step would be to identify the barriers to data flow among and within agencies and providers and then invest in activities to optimize information exchange. As noted at the third workshop, some of those barriers include federal confidentiality regulations (42 CFR Part II) that restrict sharing information about patients’ history of alcohol and substance abuse; misinterpretation of data sharing rules in the Health Insurance Portability and Accountability Act of 1996 (HIPAA); and integration of EHRs and Prescription Drug Monitoring Program data.
Data could inform the strategic deployment of health care and social services resources. Toward that end, public and commercial payers could lead efforts to identify and share information about target populations and the potential for different models to positively affect the care of those populations. Incorporating screening tools for social risk and behavioral health variables into EHRs could serve as a source of data on patients that could be used to inform program targeting. Other sources of data could include claims, administrative data, data from patient encounters with health and social services systems, and patient-related geographical information. Medicare data, collected by CMS, could serve as a rich source of information about patients and program effectiveness. If these disparate sources of data could be integrated, they could be used to align targeting strategies across payers and to inform benefits, care delivery, and payment models.
As multiple speakers over the course of the three workshops noted, achieving the type of policy changes discussed here will not happen without the involvement of all stakeholders—federal, state, and local governments; providers; payers; and patients and caregivers. Moreover, changing policies and allowing organizations to adapt to those changes will not happen quickly. As Mary Anne Sterling from Connected Health Resources said at the conclusion of the third workshop: “I think what we are doing is culture change on a grand scale, probably [on a scale] that has never been done before in this country. I think we all need to encourage our peers that it is going to take some patience, maybe one or two do-overs, maybe a left or right turn along the way, but it is definitely doable.”
REFERENCES
Adair, R., J. Christianson, D. R. Wholey, K. White, R. Town, S. Lee, H. Britt, P. Lund, A. Lukasewycz, and D. Elumba. 2012. Care guides: Employing nonclinical laypersons to help primary care teams manage chronic disease. Journal of Ambulatory Care Management 35(1):27–37.
Adair, R., D. R. Wholey, J. Christianson, K. M. White, H. Britt, and S. Lee. 2013. Improving chronic disease care by adding laypersons to the primary care team: A parallel randomized trial. Annals of Internal Medicine 159(3):176–184.
Anderson, W. L., F. Zhanlian, and S. K. Long. 2016. Minnesota managed care longitudinal data analysis. Washington, DC: HHS Office of the Assistant Secretary for Planning and Evaluation.
Bachrach, D., S. Anthony, and A. Detty. 2014. State strategies for integrating physical and behavioral health services in a changing medicaid environment. New York, NY: The Commonwealth Fund.
Bailit, M. H., M. E. Burns, and M. B. Dyer. 2015. Implementing value-based physician compensation: Advice from early adopters. Healthcare Financial Management 69(7):40–47.
Barnett, M. L., J. Hsu, and J. M. McWilliams. 2015. Patient characteristics and differences in hospital readmission rates. JAMA Intern Med 175(11):1803–1812.
Bates, D. W., S. Saria, L. Ohno-Machado, A. Shah, and G. Escobar. 2014. Big data in health care: Using analytics to identify and manage high-risk and high-cost patients. Health Affairs 33(7):1123–1131.
Bennett, I. M., J. Chen, J. S. Soroui, and S. White. 2009. The contribution of health literacy to disparities in self-rated health status and preventive health behaviors in older adults. Annals of Family Medicine 7(3):204–211.
Blumenthal, D., G. Anderson, S. Burke, T. Fulmer, A. K. Jha, and P. Long. 2016a. Tailoring complex-care management, coordination, and integration for high-need,
high-cost patients. Vital Directions for Health and Health Care Series. Discussion Paper, National Academy of Medicine, Washington, DC.
Blumenthal, D., B. Chernof, T. Fulmer, J. Lumpkin, and J. Selberg. 2016b. Caring for high-need, high-cost patients - an urgent priority. New England Journal of Medicine 375(10):909–911.
Bradley, E. H., M. Canavan, E. Rogan, K. Talbert-Slagle, C. Ndumele, L. Taylor, and L. A. Curry. 2016. Variation in health outcomes: The role of spending on social services, public health, and health care, 2000–09. Health Affairs 35(5):760–768.
Bradley, E. H., B. R. Elkins, J. Herrin, and B. Elbel. 2011. Health and social services expenditures: Associations with health outcomes. BMJ Qual Saf 20(10):826–831.
Conway, P. H., F. Mostashari, and C. Clancy. 2013. The future of quality measurement for improvement and accountability. JAMA 309(21):2215–2216.
Court, B. J., D. Mancuso, C. Zhu, and A. Krupski. 2011. Predictive risk intelligence system (prism): A decision-support tool for coordinating care for complex medicaid clients. In Comprehensive care coordination for chronically ill adults: John Wiley & Sons, Inc. Pp. 349–359.
Dale, S. B., A. Ghosh, D. N. Peikes, T. J. Day, F. B. Yoon, E. F. Taylor, K. Swankoski, A. S. O’Malley, P. H. Conway, R. Rajkumar, M. J. Press, L. Sessums, and R. Brown. 2016. Two-year costs and quality in the comprehensive primary care initiative. New England Journal of Medicine 374(24):2345–2356.
Davis, A. C. 2013. Leveraging community health workers within california’s state innovation model: Background, options and considerations. Sacramento, CA: California Health and Human Services Agency.
Dunlap, N. E., D. J. Ballard, R. A. Cherry, W. C. Dunagan, W. Ferniany, A. C. Hamilton, T. A. Owens, T. Rusconi, S. M. Safyer, P. J. Santrach, A. Sears, M. R. Waldrum, and K. E. Walsh. 2016. Observations from the field: Reporting quality metrics in health care. Discussion Paper, National Academy of Medicine, Washington, DC.
Hamblin, A., J. Verdier, and M. Au. 2011. State options for integrating physical and behavioral health care. Washington, DC: Center for Health Care Strategies, Inc.
Hayes, K., G. W. Hoadland, N. Lopez, M. Workman, P. Fise, K. Taylor, R. Meltzer, and S. Seong. 2016. Delivery system reform: Improving care for individuals dually eligible for medicare and medicaid. Washington, DC: Bipartisan Policy Center.
Hong, C. S., M. K. Abrams, and T. G. Ferris. 2014. Toward increased adoption of complex care management. New England Journal of Medicine 371(6):491–493.
Houston, R. 2016. Maintaining the momentum: Using value-based payments to sustain provider innovations. http://www.chcs.org/maintaining-the-momentum-using-
value-based-payments-to-sustain-provider-innovations/ (accessed November 2, 2016).
Jones, C., K. Finison, K. McGraves-Lloyd, T. Tremblay, M. K. Mohlman, B. Tanzman, M. Hazard, S. Maier, and J. Samuelson. 2016. Vermont’s community-oriented all-payer medical home model reduces expenditures and utilization while delivering high-quality care. Popul Health Manag 19(3):196–205.
Joynt, K. E., A. A. Gawande, E. J. Orav, and A. K. Jha. 2013. Contribution of preventable acute care spending to total spending for high-cost medicare patients. JAMA 309(24):2572–2578.
Joynt, K. E., J. F. Figueroa, E. J. Orav, and A. K. Jha. 2016. Opinions on the Hospital Readmission Reduction Program: Results of a National Survey of Hospital Leaders.The American Journal of Managed Care 22(8): 287–294.
Kangovi, S., N. Mitra, D. Grande, M. L. White, S. McCollum, J. Sellman, R. P. Shannon, and J. A. Long. 2014. Patient-centered community health worker intervention to improve posthospital outcomes: A randomized clinical trial. JAMA Intern Med 174(4):535–543.
Long-Term Quality Alliance. 2016. Taxonomy of long-term services and supports integration. Washington, DC: Long-Term Quality Alliance.
McCarthy, D., J. Ryan, and S. Klein. 2015. Models of care for high-need, high-cost patients: An evidence synthesis. Issue Brief (Commonw Fund) 31:1–19.
McGinnis, T., M. Crawford, and S. A. Somers. 2014. A state policy framework for integrating health and social services. New York, NY: The Commonwealth Fund.
National Academies of Sciences, Engineering, and Medicine. 2016a. Accounting for social risk factors in medicare payment: Criteria, factors, and methods. Washington, DC: The National Academies Press.
National Academies of Sciences, Engineering, and Medicine. 2016b. Accounting for social risk factors in medicare payment: Identifying social risk factors. Washington, DC: The National Academies Press.
OECD. 2009. Health at a glance 2009. Geneva: OECD Publishing.
Olin, G., and D. D. Dougherty. 2006. Characteristics and medical expenses of adults 18 to 64-years old with functional limitations, combined years 1997–2002. Rockville, MD: Agency for Healthcare Research and Quality.
Rajkumar, R., M. J. Press, and P. H. Conway. 2015. The cms innovation center—a five-year self-assessment. New England Journal of Medicine 372(21):1981–1983.
Secretary’s Advisory Committee on Health Promotion and Disease Prevention Objectives for 2020. 2010. Healthy people 2020: An opportunity to address the societal determinants of health in the United States. Washington, DC: US Department of Health and Human Services.
Takach, M., and R. Yalowich. 2015. Transforming the workforce to provide better chronic care: The role of a community health nurse in a high-utilizer program in Oregon. Washington, DC: AARP Public Policy Institute.
Taylor, E. F., S. Dale, D. Peikes, R. S. Brown, and G. Arkadipta. 2015a. Evaluation of the comprehensive primary care initiative: First annual report. Princeton, NJ: Mathematica Policy Research.
Taylor, L. A., C. E. Coyle, C. Ndumele, E. Rogan, M. Canavan, L. Curry, and E. H. Brandley. 2015b. Leveraging the social determinants of health: What works? Boston, MA: Blue Cross Blue Shield of Massachusetts Foundation.
Thomas-Henkel, C., A. Hamblin, and T. Hendricks. 2015. Opportunities to improve models of care for people with complex needs. Princeton, NJ: Robert Wood Johnson Foundation & Center for Health Care Strategies.
Viveiros, J. 2015. Addressing housing as a health care treatment. Washington, DC: Center for Housing Policy and the National Housing Conference.