Creating Incentives to Move Upstream: Developing a Diversified Portfolio of Population Health Measures Within Payment and Health Care Reform

Abstract

I examined the feasibility of developing a balanced portfolio of population health measures that would be useful within the current deliberations about health care and payment reform.

My commentary acknowledges that an obstacle to the selection of population health metrics is the differing definitions of population health. Rather than choosing between these definitions, I identified five categories of indicators, ranging from traditional clinical care prevention interventions to those that measure investment in community-level nonclinical services, that in various combinations might yield the most promising results.

I offer concrete examples of markers in each of the categories and show that there is a growing number of individuals eager to receive concrete recommendations and implement population health pilot programs.

Wellness and prevention are receiving more and more attention during the implementation phases of the Affordable Care Act (ACA).¹ Supported by the federal government, with additional efforts such as the State Innovation Models Initiative grants and by nonprofit groups such as the Robert Wood Johnson Foundation, movements are afoot to alter the current incentives that focus the health care system on illness rather than wellness—incentives that encourage health care clinicians to test, prescribe, and treat even when there is little apparent health value in doing so but reimbursement is guaranteed.^2,3

Often, discussions about how to focus on prevention within clinical settings have been limited to the consideration of medically oriented interventions, such as immunization. However, in a surprising number of settings this limited framework is being expanded to think outside the box and even outside the clinical setting itself. Sometimes referred to as “primary prevention” or “upstream prevention interventions,” such approaches attempt to alter the conditions in people’s lives so that the easier behaviors are the healthier ones.^4,5

The atmosphere is ripe for considering the ways to link prevention to clinical care because of the momentum to move away from fee-for-service to value-based global payment for health care.^6,7 In these new models, global payment is tied to the fulfillment of a set of prioritized health indicators, including those that emphasize wellness and the avoidance of the utilization of expensive and sometimes unnecessary hospitalizations, emergency department visits, and medical interventions. If the measures are not met, an accountable care organization or other provider group may lose a financial bonus and even risk future contracts with payers. Other innovative payment approaches also create opportunities for the promotion of wellness. For instance, many states are implementing delivery system reform incentive payment programs, partnerships between the federal government and state Medicaid programs, to support safety net providers who care for patients with complex socioeconomic barriers to improved health.

Coinciding with these changes is the expanded prevalence of practice models that may lend themselves to support for a population health focus. The integrative care approach of patient-centered medical homes, including the chronic disease–focused Medicaid health homes, is more likely than is a conventional model to recognize the importance of addressing the conditions of patients’ lives.

These changing circumstances have created a new openness to nontraditional approaches to keeping people and communities healthy. But although the moment exists for meaningful change, it may not last long. The ACA’s implementation and the Centers for Medicare and Medicaid Services’ distribution of hundreds of millions of dollars in funding for state innovation model (SIM) grants have created an environment in which unprecedented discussions are taking place on a relatively tight timeline. In state after state efforts are under way to engage public and commercial payers in an overhaul of the payment system. In the case of the SIM grants, the states are given just four years to shift to value-based contracting as the predominant reimbursement mechanism with aligned and consistent quality measures across insurers. As these discussions occur, public health leaders have to be prepared to participate. More importantly, they need to offer concrete and specific suggestions that fit into an insurance-oriented framework and have a demonstrable benefit in cost and quality.

Various approaches to identify meaningful population-based health measures are being examined. These include state-specific exercises such as the Massachusetts Statewide Quality Advisory Committee, a legislatively mandated body charged with compiling the proven and most promising health indicators to assist in the state’s innovative cost control and quality promotion initiatives. Although initially prioritizing the selection of at least a few population health indicators, Massachusetts postponed this task in part because of the lack of concrete, evidence-based measures of prevention and wellness and the enormity of the effort to sift through the more conventional treatment-oriented indicators.⁸

There have been national high-profile efforts as well. Starting in 2012 the Institute of Medicine Roundtable on Population Health Improvement convened several stakeholders’ dialogues to examine what was needed to improve population health. It produced a useful summary report regarding population health improvement that highlighted the importance of collaborations involving multiple stakeholders and the use of financial incentives to align the interests of physicians and hospitals.⁹ The participants in these meetings offered thoughtful insights into the groundwork needed to establish the incentives for population health and highlighted numerous positive examples of such activities around the country.

Beginning in 2011 the National Quality Forum began a two-phase process that led to their endorsement of 24 clinically focused population health measures. These measures provide a valuable building block for the development of a coordinated approach to population health and its relationship to clinical care. The National Quality Forum recently formed a Health and Well-Being Standing Committee to review specific health measures that thus far have had a traditional clinical focus. In addition, it formed a Health and Human Services Department–funded Population Health Framework Committee and a Population Health Task Force to further its consideration of this work.¹⁰

The Centers for Medicare and Medicaid Services’ Innovation Center funding of SIM has led states to transform health care systems by creating and testing new models of care delivery and payment. The recent second round of the SIM grant awards required states to include a population health component that was integrated with value-based contracting and patient-centered medical homes. Many of the funded states have expressed an interest in guidance on the development and use of population health measures.^7,11

In each of these settings, there has been general agreement that the study of population health measures is a relatively nascent field and lacks an optimal reservoir of information and experience. Nonetheless, in these various endeavors there has been an oft-repeated recognition that the current period offers a window of opportunity that should not be missed.

The specific proposals for population health metrics that have been offered for consideration in these various settings have differed—often dramatically—on the basis of the definition of population health (e.g., is it limited to a clinical practice or focused on the larger geographic community?), the size of practice (e.g., does it apply only to accountable care organizations [ACOs] and the largest practices or does it also apply to smaller medical groups?), and the intervention setting (e.g., do the action steps occur only in clinical settings or do they also occur in the patients’ homes, neighborhoods, or cities?). This definitional variation has led to some confusion as the discussions have proceeded and likely impeded progress in the effort to identify meaningful metrics.

Rather than choosing between approaches, it may be useful to conceptualize a portfolio of proposals that in various combinations—depending on the circumstances—might yield the most promising results. Based on the types of proposals that have been offered in recent years, such a portfolio might include the following five categories:

1. traditional clinical prevention interventions;

2. health indicators of the universe of patients treated in a particular practice or health system;

3. health indicators of the residents within a community, a catchment area, or the total population;

4. indicators of support for ancillary services that are patient centered and evidence based but often delivered outside a clinical setting or community-level care; and

5. indicators of support for improving the living conditions in the neighborhoods, workplaces, and schools of the patient population or, as Centers for Disease Control and Prevention director Tom Frieden, MD, MPH, refers to them, context-changing interventions that encourage healthy behaviors.⁴

Traditional clinical prevention interventions and practice-based population measures involve data that can be captured in the routine manner of a one-to-one encounter of a patient with a clinician and then aggregated across a practice or system. Total population measures involve data predominantly collected by local, state, or federal health officials. Community-level care and context-changing interventions often involve what may be referred to as process measures as well as outcome measures. These process measures are systemwide resource and policy action steps that extend beyond the one-on-one provision of patient care. They aim to positively influence the conditions in specific communities or in certain settings, such as prisons, resulting in the widespread prevention of illness and injury. Such measures are most feasible for sizable organizational entities, such as ACOs, vertically integrated health care systems, and large hospitals. Requirements for process measures could be written into licensing or certification rules or regulations.

The indicators selected may vary from state to state, taking into account the diseases and risk factors that are most prevalent and costly and those for which there are clearly identifiable effective actions steps. Whenever possible, the time frames for results for the indicators in each of these five categories should be both short and long term. Some short-term cost savings (or at least cost neutrality) and short-term positive health outcomes are necessary to guarantee continued support from elected officials and from health plans under pressure to perform. However, at the same time, many population health measures will take years to demonstrate measurable positive health outcomes—and some may turn out to be costly but deemed necessary because of the impact they have on the health and well-being of the population. A case can be made that all five areas are needed to optimize health and wellness.

TRADITIONAL CLINICAL PREVENTION INTERVENTIONS

Numerous indicators already routinely collected by clinical providers measure quality, including the Healthcare Effectiveness Data and Information Set and Centers for Medicare and Medicaid Services hospital process measures for congestive heart failure, pneumonia, and surgical infection prevention.¹² These generally count certain units of service that are delivered in hospitals and outpatient settings. A subset of these units of service is particularly suited for inclusion in the list of preventive measures that could be incentivized by insurance reimbursement in the future. Examples include the percentage and number of patients who have had seasonal influenza vaccines or colonoscopies and the percentage and number of patients who have been screened for obesity and tobacco use. Even if many of these interventions are already reimbursed, there is room for improvement by increasing the weight with which insurers incentivize them. And some of these interventions work best when paired with another category of action indicators. For example, screening for tobacco use and counseling against smoking or prescribing medication to assist with smoking cessation may work well when paired with support for community-based smoking-cessation groups; measuring body mass index, a standard Healthcare Effectiveness Data and Information Set metric, may work well in conjunction with community-level interventions to attract stores that sell fresh produce to low-income neighborhoods.

Examples of prevention-oriented traditional clinical measures include the following: annual influenza vaccination, use of aspirin for those at increased risk of a cardiovascular event, screening for tobacco use, screening for substance abuse, and screening for domestic or other violence.

PRACTICE-BASED POPULATION MEASURES

Health information technology makes it possible to look at certain diagnoses, conditions, or risk factors across all the patients in a clinical practice or system. Providers should identify a few conditions or risk factors that provide insight into the need for enhanced prevention efforts for their patients. For example, an analysis of patients who are overweight might reveal that patients from particular geographic subsections of neighborhoods are more likely to need assistance in eating well or exercising. Such information would help shed light on the challenges patients face in following clinical advice related to behavior change. The Camden Coalition of Health Providers in New Jersey identified a disproportionate number of asthma and otherwise chronically ill patients who lived in two buildings, suggesting the need for a concentrated effort to reduce environmental triggers in these settings.^13,14 This type of information can also be linked to (1) health indicators in total population measures to better understand a practice-based population in the context of the total population; and (2) the targeted resource allocation indicators of support for improving the living conditions in the neighborhoods, workplaces, and schools of the patient population. This type of information can also inform specific ancillary services in community-level care, such as home visits of community health workers.

Examples of practice-based population measures include percentages of the following: patients with asthma and diabetes, patients who are overweight, age- and gender-appropriate patients who have had a colonoscopy or other cancer screenings, and patients with a newly diagnosed sexually transmitted disease, including HIV.

TOTAL POPULATION MEASURES

Within a prevention-oriented system, clinicians have a financial interest in improving the health of the larger population, not simply their own patients. This includes their patients’ families, friends, and neighbors. One way to incentivize this outcome on a broad scale is to link global payment to the improvement of specific health indicators in the community population.⁶ For example, if new HIV infections are increasing in the catchment area, providers might have a financial interest in understanding the causes and thus support larger efforts—some of them led by public health agencies—to reduce these infections. Care would have to be taken in selecting these population-wide indicators. But an undeniably good place to start is the Centers for Disease Control and Prevention–supported “ABCS” community-wide measures of success in ensuring aspirin for people at increased risk of cardiovascular events (A), blood pressure control for people with hypertension (B), cholesterol management (C), and smoking-cessation support for those trying to quit (S).¹⁵

Total population measures are best addressed if there is agreement for their inclusion in Medicaid and all major commercial insurance programs in a state. It is more difficult to show that a single ACO should be held accountable for an entire community’s health when it has a limited percentage of the market share. But if all ACOs and large providers are held accountable for the same total population health measures, their collective partnership is likely to yield results from which they all benefit.

Examples of total population measures include the percentages of the following: residents with asthma and diabetes; residents who are overweight; age- and gender-appropriate residents who have had a colonoscopy or other cancer screening; residents with a newly diagnosed sexually transmitted disease, including HIV; and residents whose ABCS have been met.

COMMUNITY-LEVEL CARE INDICATORS

This category would involve a process measure, or specific system-wide action step, such as a commitment of resources, demonstrating an organization’s investment in total population health. Agencies would need to support the provision of secondary preventive care outside the clinical setting. For example, such an investment might involve the routine use of community health workers to supplement and amplify the messages offered in a clinical visit. In Massachusetts, we have seen strong evidence of the cost and quality benefits of community health worker home visits in reducing clinical symptoms of children with severe asthma.¹⁶ In addition, certain evidence-based community interventions complement patient care in clinical settings, where they may be impractical to provide. In the case of smoking cessation, for example, we know that talk therapy combined with prescription medication may yield the best results.¹⁷ To be accessible, however, cessation groups must be offered at multiple times and in various geographic locations. Because all but the largest practices would find it challenging to field groups on this large a scale, such services might best be provided at a community-wide level with pooled dollars from ACOs or large practices in a catchment area. Similarly, chronic disease self-management training may become more practical if it is offered at the community level, with multiple clinical practices providing pooled funding.

An example of this approach is the Massachusetts Prevention and Wellness Trust. Created by legislation, this $60 million trust is funded by an assessment of health plans and the largest clinical providers. Its funding is supporting regional efforts to link clinical providers to neighborhood-based social services and to promote community-wide health improvement.

Examples of community-level care indicators include the following: the routine use of community health workers and funding for smoking-cessation groups and chronic disease self-management groups in the larger community.

CONTEXT-CHANGING INTERVENTIONS

This category would also involve an action-oriented process measure, demonstrating the commitment of a large organization, such as a hospital or an ACO, to total population health. Because the authority and expertise to improve the health of the general population exist within local and state public health agencies, this category would promote and incentivize the development of strong public–private partnerships.⁶ One way to achieve that would be to require an annual meeting between ACOs and public health leadership to establish uniform prioritized health campaigns. In Boston, Massachusetts, for example, the Public Health Commission has identified three priority issues it aims to improve—the rates of low birth weight, obesity, and chlamydia.¹⁸ In this example, each of the city’s hospitals and ACOs would work with the commission to devote attention and resources to meeting these objectives. This link with public health would also provide a forum for considering the post-ACA feasibility of integrating into the private sector those services currently provided in the public sector.

An additional requirement might be the demonstration of support for community-level planning with the use of a model such as the Robert Wood Johnson Foundation’s Roadmaps to Health.¹⁹

Examples of context-changing interventions include the following: a financial contribution to a pooled population health fund, an assigned employee or funded position to work on community-level health improvement, and a required annual meeting with local public health leadership to review public health priorities and develop a plan to address them.

The federal Community Preventive Services Task Force provides useful resources to determine which context-changing approaches should be incentivized for reimbursement. This expert panel makes prevention-oriented, evidence-based recommendations on the basis of scientific reviews. However, its recommendations do not currently result in reimbursement, unlike those given the A or B grade by its sister entity, the US Preventive Services Task Force, which focuses on clinical care.

CONCLUSIONS

Health care and payment reform is occurring in one form or another across the nation. Change of this magnitude does not occur very often. Public health and prevention professionals and advocates need to fully participate in the process with the goal of improving the health of the nation through prevention and wellness promotion, as well as access to high-quality clinical services. To participate meaningfully, public health officials and providers need to have concrete, specific, and defensible prevention-related proposals for inclusion in the emerging payment models and systems. The best approach may be one that uses a diverse portfolio of approaches combining prevention-oriented, practice-focused clinical care and analysis with total population improvement initiatives. Such an approach relies on the use of carefully selected quantitative data that have already been collected by clinical and public health entities and process measures that demonstrate a commitment to collaboration with public and private partners in the larger community to maximize health and health outcomes.

Now is the time to roll up our sleeves and hammer out the best mixture of measures and incentives. We will learn the most if we innovate and test out various approaches. If we wait until all the data are in and the answers are clear, we will have missed a once-in-a-generation opportunity.