Can Payment Reform Be Social Reform? The Lure and Liabilities of the “Triple Aim”

Abstract

Introduction

In 2008, Donald Berwick and colleagues proposed that the US health care system adopt the “triple aim”: to improve health care, to improve population health, and to reduce per capita health care costs (Berwick, Nolan, and Whittington 2008). Berwick became administrator of the Centers for Medicare and Medicaid Services (CMS) in 2010 and described the pursuit of these three goals as the focus of his term (Fleming 2010). This was possible not least through the implementation of the Affordable Care Act (ACA), specific provisions of which are considered to embody the triple aim. The formulation of the triple aim responds to three problems facing the US health care system: high cost, low quality, and poor health status (which frustrates efforts to contain cost and improve quality). It is the latest and most ambitious of a succession of health care goal statements, which progressed from cost containment in the 1970s, to cost containment while/by improving quality in the 1980s and 1990s, to cost containment while/by improving quality and population health most recently. From CMS's perspective, the triple aim is vital to Medicare, and new payment mechanisms, such as the accountable care organization (ACO), have been deployed to achieve it. These mechanisms partake of the enduring economism of US payment policy, that is, the belief that precisely calibrated financial incentives will produce socially desirable ends.

The purpose of this article is to analyze the potential of the health care system to achieve the triple aim and, specifically, the attempt to improve population health by rewarding providers who contain costs. The article considers if health care payers can and should incentivize providers to take responsibility for population health, even when it requires nonmedical services and the remediation of social problems, and whether the health of populations should be implicated in providers’ cost containment. The evaluation has two prongs, assessing on the one hand if current payment reforms will succeed in improving population health, and on the other if this approach to population health improvement is normatively and politically sound. Following Casalino et al. (2015), the article distinguishes patient population health from geographic population health when this is helpful. It also acknowledges that population health improvement runs along a continuum from providing health care to addressing the social components of health care (e.g., transportation to medical appointments) to providing social services (e.g., housing and education) to addressing the social determinants of health (economic and environmental conditions). Ultimately, the article argues that current payment reforms are unlikely, with some notable exceptions, to effect major improvements in geographic, or even patient, population health. Furthermore, despite the lure of health care dollars for population health, there are significant normative and political liabilities to seeking social reform in the cost containment efforts of health care providers.

The first section of the article will consider the task of improving population health through the health care system. Reformers have long recognized that many determinants of health status have nothing to do with health care, and the article will describe some efforts to involve the health care system in meeting nonmedical needs. All entail integrating health and social services/social determinants, but none operates under CMS's “alternative payment models,” such as the ACO. These examples are meant to illustrate the impetus toward integration and the variety of integrative schemes currently in place, but also to note their limited reach and idiosyncratic trajectories. They set the stage for CMS's alternative payment methodologies, devised to create health care resources and a health care constituency for population health.

The second section of the article will discuss CMS's efforts to pay providers to achieve the triple aim, that is, to improve health care and population health while containing cost. These include Maryland's Global Revenue Budget model, bundled payments, and ACOs, and they highlight the extent to which this version of integration is underwritten by savings achieved by providers for the Medicare program. This arrangement has implications for both the feasibility and desirability of these reforms. The health care system has deep pockets, but providers may be reluctant to accept financial risk and, if they do, may contain cost in ways unrelated to population health. There are more fundamental liabilities as well. These payment reforms run the risk of pathologizing poverty and furthering an instrumental logic of social reform.

The conclusion section of the article will consider the politics of payment reform as social reform. It will address proposals that health care payers and providers lead in addressing the social contributors to ill health (Shortell 2013; Casalino et al. 2015), and urge payment reformers to appreciate more fully the politics and policies of other sectors and the dynamics of their inclusion in population health improvement.

Health Care and Population Health

The factors contributing to population health extend from micro-determinants such as a lack of transportation to medical care to macro-determinants such as high levels of economic inequality and concomitant chronic stress. What these factors share is that they are nonmedical factors in health outcomes, and their effects are most pronounced in low-income populations.

Bradley and Taylor (2013) speak to the social contributions to health in claiming to have solved the “American health care paradox”—that the United States spends lavishly on health care but achieves only middling health outcomes. The United States ranks first in per capita spending but ranks twenty-fifth in maternal mortality, twenty-sixth in life expectancy, and thirty-first in infant mortality. The authors analyze cross-national data and find that countries with better outcomes spend less than the United States on health care but more on what they call “social service expenditures.” These include public and private spending on income and social supports for older people and people with disabilities; family support; employment programs, including not only unemployment benefits but supportive housing and rent subsidies; and other social services. According to (thirty-four-country) Organization for Economic Cooperation and Development (OECD) data, the United States spent 16.3 percent of GDP on health care in 2007 but only 9.1 percent on social service expenditures, compared with, say, France, which spent 11.9 percent of GDP on health care and 21.3 percent on social expenditures. Of the OECD countries, only Chile, Estonia, Turkey, Korea, and Mexico spent a lower percentage of GDP on social expenditures, and in Chile and Estonia, this percentage still surpassed that expended on health care (Bradley and Taylor 2013).

Health care programs designed to address social factors are not new, but they have mostly struggled to succeed. “Social medicine,” which emphasized the role of social factors in creating and perpetuating disease, dates to the mid-nineteenth century, but it was eventually overtaken by an emphasis on localized pathology and attention to particular organs (Starr 1982). The health center movement, active between the 1910s and 1930s, sought to integrate health care and public health for a given population; it was effectively opposed by organized medicine as an encroachment by government on private medical practice (Starr 1982). (Neighborhood health centers would reappear during the War on Poverty in quite a different guise and remain important to low-income Americans [see below] [Starr 1982].) In the area of long-term care, Medicare has allowed integration of health and social services in social health maintenance organizations (S/HMOs) and the Program of All-Inclusive Care for the Elderly (PACE). Evaluated in 2002, S/HMOs did not demonstrate reduced utilization of hospital services, except in a small group of high-utilizing enrollees, and nursing home admission was too rare to evaluate (Thompson 2002). On the other hand, PACE was shown to reduce medical and nursing home utilization and improve quality of life and functional status (Chatterji et al. 1998). Growth of the program, however, has been limited by a host of obstacles, including the substantial capital investment required to begin operations (Gross et al. 2004).

Bradley and Taylor (2013) profile a number of current programs that integrate medical care and social interventions for vulnerable populations. One is the Errera Community Care Center in West Haven, Connecticut, which is funded by the Veterans Administration and provides a full array of medical and social services, including legal, housing, and job-training services, under one roof. In contrast, the Ocean Park Community Center serves homeless people in Santa Monica by practicing the “housing first” model, but it also operates a satellite clinic where program participants can be seen by a physician and receive basic medications and referrals. In one case, then, a medical facility broadens its offerings to include social services, and in the other, a homelessness program brings medical services to its participants.

Some programs integrate health care and social services by broadening the duties of existing health professionals. The Nurse-Family Partnership (NFP) program, for example, provides home visitation by registered nurses to low-income, first-time mothers. Nurses support these women not only in prenatal and infant care but also in planning for their futures by guiding them to existing educational and employment opportunities. The program has been studied in multiple randomized controlled trials nationally and has been shown to improve birth outcomes and child health. In addition to better health outcomes, the program also reduces the number of pregnancies, increases employment rates, and decreases reliance on Supplemental Nutrition Assistance Program (SNAP) and Temporary Assistance for Needy Families (TANF) (Williams et al. 2008). Although home visiting by nurses is not a new intervention, in 2010 the Maternal, Infant and Early Childhood Home Visitation (MIECHV) program was created by the ACA. It funds the NFP program and others like it. In 2015, the federal grant to states for the MIECHV program was $386 million (HHS 2015).

Perhaps the largest-scale example of health and social service integration is the neighborhood or community health center, noted above. As federally qualified health centers (FQHCs), they operated almost 8,000 sites and served 18 million patients in 2009 (Katz et al. 2011). These FQHCs are required to provide low-income patients with comprehensive health services, including environmental health services, and social services, such as transportation, that ensure access to health care. One category of service, “enabling services,” is meant to support a patient's access to “non-medical, social, educational or other related services (e.g., child care, food banks/meals, employment and educational counseling . . .)” (HRSA, n.d.), and in 2014, the federal Health Resources and Services Administration (HRSA) added civil legal services to the list of enabling services (NCMLP 2014). The FQHCs are viewed as an important resource for newly insured patients under the ACA, especially for those eligible under the Medicaid expansion.

Efforts to integrate health care and civil legal services have been underway for more than twenty years. Under the medical-legal partnership model, attorneys work in teams with patients’ health care providers and serve as advisers and advocates on health-related legal questions. Low-income patients especially may benefit from legal assistance in negotiating complex systems of social provision and in preventing the legal crises that may prove deleterious to health (Williams et al. 2008; Zuckerman et al. 2008). At present, there are nearly three hundred medical-legal partnerships in thirty-six states (NCMLP 2015). The UMass Memorial Medical Center in Worcester, for example, has partnered with Community Legal Aid since 2003 to address legal problems affecting the health of pediatric patients. In 2014, the partnership began a pilot project dedicated to addressing housing conditions, including pest infestations, that contribute to patients’ high rates of asthma and are considered remediable through legal action (Kremer 2015).

A final example of health and social service integration sees health care payers addressing homelessness, an obvious threat to health. In 2013–14, New York State redirected $86 million in state Medicaid funds to supportive housing for “high-risk and unstably housed” patients. CMS has rejected New York's request for a waiver to allow federal matching funds for housing services, but proponents hope—and research suggests—that much of the cost to the state will be recouped in reduced health care spending by supportively housed beneficiaries (Doran, Misa, and Shah 2013). In Texas, UnitedHealthcare, which serves Medicaid and non-Medicaid populations, is using company funds to provide housing to homeless enrollees on a very small scale. The firm calculates that housed clients will utilize fewer services than homeless ones and ultimately reduce cost (Walters 2015). Officials in New York make a broader case: “We envision a Medicaid system in which spending on social determinants of health . . . is not only allowable but recognized as a best practice” (Doran, Misa, and Shah 2013: 2376).

In spite of these and other efforts to address the social factors in ill health, financing remains a serious obstacle to the work. As Bradley and Taylor (2013) discover, whereas other OECD countries spend an average of two dollars in social expenditures for every one spent on health care, in the United States the figure is sixty cents per health care dollar. Lawyers involved in medical-legal partnerships, for example, often work pro bono, as there are far too few legal aid attorneys to meet the needs of low-income Americans (LSC 2009). Despite new funding under the ACA, the capacity of FQHCs varies significantly across communities (Katz et al. 2011), and they can refer patients only to existing food banks and feeding programs. Similarly, NFP nurses can only link first-time mothers to available employment and educational services. The CMS is unlikely to approve Medicaid funding for housing services. As former Health Care Financing Administration (HCFA) Administrator (and New Yorker) Bruce Vladeck puts it, “Medicaid is supposed to be health insurance, and not every problem somebody has is a health problem” (Aronczyk 2014).

Some analysts propose attracting private money for the improvement of population health. Charitable foundations play an important role in programs such as the ones described above, but foundation funding is highly contingent and not sustainable. The IOM (Institute of Medicine) (Shortell 2013) and others (Audrey 2013) look to community development bonds to improve neighborhood conditions contributing to poor health status, and “social impact bonds” (SIBs) represent a new mechanism for raising funds (Ollove 2014). The SIBs are designed to fund government agencies to undertake (or contract for) programs to achieve social ends, say, improved Medicaid birth outcomes. If the program succeeds, and thereby reduces state spending on ill newborns, the bondholders will recoup their principal and a healthy rate of return (and presumably invest again). Critics argue that the bonds may cost government more than they raise, especially once administrative costs are paid, and because SIBs are a relatively new concept, there are currently few data by which to evaluate them (Ollove 2014; Popper 2015). More importantly, although SIBs may raise private money for public purposes, investors will have a personal stake in how much public money is saved and may exert undue influence on the conduct of the program. The overall scarcity of resources to improve population health surely contributes to the lure of the triple aim. Despite continuous consternation about the rising costs of health care, these expenditures are substantial and continue to grow.

Paying for the Triple Aim

The lure of the triple aim is, in part, the current enthusiasm for bringing social factors into the medical encounter. According to the Robert Wood Johnson Foundation, physicians believe that social determinants jeopardize their patient-care efforts, and those seeing low-income patients wish they could order (but not necessarily pay for) prescriptions for employment assistance, adult education, and housing assistance (RWJF 2011). Williams et al. identify the “important challenge” of creating among providers “personal and institutional ownership of the problem of health disparities,” which can be traced to social determinants such as poverty and race (2008: 10). Again, Sayer and Lee, writing in the New England Journal of Medicine, propose that health care organizations pay other groups to undertake activities related to patients’ social determinants of health. The authors imagine the integration of “personnel who can help address social needs” (Sayer and Lee 2014: 1276) into a clinical setting, although they do not specify how this will be achieved. In their cross-national study, Bradley and Taylor (2013) demonstrate the extent to which health care spending outpaces social services expenditures. The lure of paying for the triple aim, then, is also that health care payers with deep pockets might, however circuitously, divert some funding to meet existing community needs. The prospects for significant new social spending or greater market-generated family incomes seem dim. By pairing improved population health with health care cost containment, the triple aim promises political cover for some amelioration of social determinants of ill health.

The CMS and other health care payers are not, on the other hand, searching out new spending opportunities. Rather they have, for decades, been focused on containing their expenditures. Paying providers for the third aim represents the recognition that health status affects health care costs: an obese Medicare patient, for example, costs the program more than one at a healthy weight (NRC 2010). The same is true for more distant predictors of ill health: a low-SES (socioeconomic status) Medicare patient is more expensive to the program than an affluent one (NRC 2010). Berwick and colleagues (2008) view the triple aim as three interdependent, and not merely simultaneous, goals. They acknowledge that in order to achieve all three, the health care system must confront a “commons” problem, specifically that current incentives lead rational actors to meet one or two of the aims but not all three. It is necessary, then, “to change what is rational for them to do” (Berwick, Nolan, and Whittington 2008: 761). Central to the authors’ vision is the creation of an “integrator,” who accepts responsibility for the triple aim and is motivated to do so through incentives that have been “aligned.” Specific payment mechanisms such as bundled payments and ACOs incentivize such integration at the provider level (Berwick, Nolan, and Whittington 2008), as does the Maryland CMS waiver for a global hospital budget (Butler, Grabinsky, and Masi 2015).

In a recent discussion paper, Brookings Institution researchers profile Washington Adventist Hospital as a “hub” in the improvement of community health. The hospital is located in Maryland, where the state's hospital payment system is unique and operates under a waiver from CMS. In 2014, Maryland adopted the Global Budget Revenue model, which pays each hospital an annual global budget based on historical levels of service and the number of people in the community. This is paired with a reduction in the global budget for any readmission (above a threshold) to any hospital within thirty days (Butler, Grabinsky, and Masi 2015). The Global Budget Revenue model is supported by CMS, but unlike bundled payments and ACOs, it is available only to hospitals in Maryland, a state that has enacted and implemented innovative payment models for decades. Washington Adventist's efforts can be summarized as: identifying and acting on “hotspots,” where residents are most likely to phone 911 for emergency care; leading the creation of community-wide networks to coordinate health and social services for a specific patient; collecting information about patients’ social circumstances and linking them with appropriate social service agencies; and developing formal partnerships with such agencies, an FQHC, and an antipoverty organization.

At the national level, Medicare-bundled payment seeks to integrate services delivered during a single episode of care by paying providers a single sum for the episode. Bundling is not new, but it has recently become the subject of a large-scale Medicare pilot program sponsored by the Centers for Medicare and Medicaid Services’ (CMS's) Center for Medicare and Medicaid Innovation (CMMI). The program features four models of episodic payment for any of forty-eight medical or surgical episodes with the opportunity for “gainsharing” by demonstration participants, that is, for sharing with the Medicare program any savings achieved. The most expansive model includes all services during hospitalization and for a period of thirty, sixty, or ninety days following. This episodic bundling includes care delivered in settings outside the hospital or physician's office, say, in rehabilitation facilities and nursing homes, and delivered by physicians, nurses, physical therapists, nursing assistants, etc. (CMS, n.d.). There are currently almost 7,000 providers receiving bundled payment from Medicare. Recently, CMS accelerated its efforts to promote bundling by proposing such payment for joint replacement in seventy-five health care markets across the country (Hammer 2015). Bundling does not yet seem to inspire attention to social determinants of health. It is possible, however, that providers of joint replacement (or their discharge planners) will find it financially advantageous to improve a patient's housing conditions rather than pay for additional nursing home days.

Medicare also offers providers the opportunity to form and be paid as ACOs, and Berwick believes these organizations will “accelerate progress” toward achieving the triple aim (2011: n.p.). As ACOs are designed to integrate care for Medicare patients, hospitals, physicians, and other health care providers may join in the organizational structure. The ACOs are eligible for gainsharing if they reduce costs while satisfying quality metrics, and they may choose to assume greater or lesser financial risk. Specifically, they may choose a guaranteed gainsharing rate of 50 percent with no risk or a gainsharing rate of up to 60 percent with a risk of financial penalty if cost targets are not met (Abrams et al. 2015). The ACOs are paid on a fee-for-service basis, not a bundled one, but they are incentivized to integrate care by the high cost of fragmentation. In 2015, there were over 400 Medicare ACOs serving about 14 percent of the program's beneficiaries (Abrams et al. 2015). The 2015 legislation repealing the sustainable growth rate (SGR) formula also set out future changes to Medicare reimbursement policy that increase incentives for physicians to join ACOs and make clear CMS's commitment to these organizations and other alternative payment mechanisms (Oberlander and Laugesen 2015). At least some ACOs have been attentive to population health. In New Jersey, the Camden Coalition of Healthcare Providers (CCHP) is strengthening the capacity of primary care practices and improving health education and community-based care for diabetic patients. Unlike some other ACOs, however, CCHP is located in an area where CMS dominates the payer mix and where local providers had formed a nonprofit organization devoted to improving care delivery as early as 2002 (Brenner 2010).

The Global Revenue Budget, bundled payments, and ACOs are meant to align the incentives of health care providers with those of patients and payers: high-quality care delivered to patients with better health status should ultimately contain health care costs, and providers will deliver this care and improve this status if they share in the savings. If improving population health saves money, however, CMS might achieve the triple aim by addressing social determinants, such as housing quality, directly rather than through medical providers. This is what New York Medicaid is doing, without CMS participation, for homeless beneficiaries. Similarly, Los Angeles County is using health care dollars to house over two thousand homeless individuals in spring 2016 (Gorman 2015). In the private sector, a 2015 report was commissioned by Blue Cross Blue Shield of Massachusetts to evaluate the cost of unmet social needs. It summarizes the “substantial” evidence that spending more on social services and health and social partnerships improves health status and reduces health care costs (Taylor et al. 2015). Whether Blue Cross Blue Shield will act on this analysis remains to be seen, but in Texas, as noted, the private UnitedHealthcare is already paying to house a small number of homeless enrollees (Walters 2015).

At CMS, the Medicaid program has a history of attending to enrollees’ social needs directly. Medical transportation, for example, is a mandatory service for every participating state. Under the home- and community-based waiver programs, an individual eligible for Medicaid nursing home benefits may be maintained at home, under a budget, with a range of social services—including some not ordinarily covered—if these forestall entry into a nursing home. Some of the original waiver offerings, such as personal care, became optional services for possible inclusion in the Medicaid State Plan, and even if CMS withholds matching dollars for New York's homelessness initiative, Medicaid has long funded some group living arrangements for people with developmental disabilities.

Medicare, of course, is different and hews more closely to a medical model. Program benefits have been quite slow to expand, even to medical staples such as prescription drugs, and this insurer of older Americans still offers very little in the way of long-term care. Medicare has, however, had enormous success as a regulator of health care reimbursement (Oberlander 2003). Paying providers to achieve the triple aim is, or seems to be, within the program's repertoire, although in 2014, 45 percent of ACOs cost Medicare more than projected, and ACOs overall created a nearly $3 million loss to the program. An analyst with The Advisory Board Company concludes, “It's turning out to be tougher to transform care and realign delivery than people expected” (Rau and Gold 2015). In fact, providers actually participating in bundled payment and ACOs are notably hesitant to accept financial risk, diminishing the incentive to address population health in the first place. As of spring 2015, there were almost seven thousand providers involved in CMS's bundled payment initiative, but only 3.5 percent bore any financial risk. Phase 1 of the initiative required the gathering of information alone, and providers agreed to do this for an average of forty-five of the possible forty-eight episode types to which bundled payment could be applied. Phase 2 requires the assumption of financial risk—providers are penalized for cost overruns as well as rewarded for cost savings—and in this phase, they are accepting bundled payment for only an average of eleven out of forty-eight (Eckstrom 2015). Similarly, of the approximately 350 Medicare ACOs operating in 2014, only 7 percent chose the high-risk/high-reward payment option; the others opted for smaller potential bonuses so as to avoid any possible penalties. Financial risk is so unpopular with ACOs, in fact, that CMS declined to phase out the no-risk option and extended its availability to a period of six years (Rau and Gold 2015).

Even if CMS—and other payers—are able to make these payment reforms work, that is, to incentivize health care providers to contain cost through population health, the practical liabilities of this approach remain significant. First, providers may not address social contributors to population health at the most effective point along the social continuum described above but rather intervene as close to the medical encounter as possible. In Maryland, where hospitals receive an annual global budget amount based on population, there arises “a general challenge facing organizations that undertake upstream activities in a community. . . . The challenge is how to identify and measure the positive externalities generated outside the hospital” so as to account for them in budgets and business models (Butler, Grabinsky, and Masi 2015: 12). Hospitals that address the social determinants of health might contribute to, say, a lower high school drop-out rate, but they are unlikely to capture the health benefit of higher SES, especially in the short run. This makes hospitals less likely to invest in population health except for patients whose immediate social circumstances lead directly to higher hospital costs—so-called “super-utilizers.” Casalino et al. (2015) point out that ACOs will be rewarded only for savings on the patients attributed to them, and in competitive health care markets they are unlikely to invest in social interventions that benefit not only their patients but also their competitors’.

Second, if providers assume the financial risk intended to improve health status, they may choose cost-cutting strategies with which they are more familiar and over which they have more control. Such strategies may even compromise population health. For example, an early report from CMS's bundled care initiative involves one of the Cleveland Clinic's community hospitals, where twenty-seven of twenty-eight joint replacement patients were treated for less than the bundle's target price. Data suggest this was accomplished mainly by decreasing discharges to a skilled nursing facility (32 percent as opposed to a previous 78 percent) and increasing discharges to home (68 percent as opposed to a previous 22 percent) (Froimson et al. 2014). This result may represent better patient care or the care most patients prefer, but it may also constitute a shift from formal to informal caregiving, with the costs to informal caregivers that it entails. Families in communities with poor population health can least afford (to miss work so as) to care for a convalescing relative or friend at home. More globally, hospitals routinely seek cost containment by resisting unionization, and this is unlikely to change despite the fact that unions typically bring higher wages and more job security to low-wage workers in the surrounding community (Moody 2014). The managed care organizations of the 1990s surely cut costs in ways not contributory to the “health maintenance” of their enrollees, and although ACOs are not HMOs, one difference is that ACOs are put at less risk and have less incentive to contain costs at all (Frakt and Mayes 2012).

There are also two normative liabilities of paying providers to address the poor health status of low-income people. One is that it will further pathologize poverty and individualize its remediation. United States social policy is strongly rooted in the Puritan association of material poverty with poor individual character (Morone 2003), and although the belief that people are poor because they do not help themselves is less common than it once was, 44 percent of all Americans and 61 percent of Republicans held that belief in 2014 (O'Connor 2014). Addressing the social determinants of health in the medical encounter may be an improvement on Puritanism, but adverse conditions will still be regarded as a personal deficit rather than a failure of public policy—as a medical datum such as a disease or an allergy. This may lead to some efforts on the patient's behalf, but it notably misconstrues social conditions as individual problems. Poverty, however much a health risk, is not fundamentally a personal affliction to be resolved in the doctor's office.

A second is that paying providers to achieve the triple aim reinforces an instrumental view of social provision. In other words, it holds that social determinants are a problem because they are an obstacle to “health,” and so to fewer hospital days or dollars of reimbursement. Instrumental arguments for social provision can be powerful and underlay many American social programs, including public education (Katz 1987) and public health (Buchanan 2008). Paying providers for the triple aim doubles down on that instrumentality by pairing better health not with, say, broad social goals such as better military preparedness, but with reduced program spending, some of which reverts to the providers themselves. Poverty, however, arises from a distribution of economic and political power (Graham 2004) that must be considered unjust regardless of any measurable effects on health: poverty's effects are anti-democratic regardless of their role in higher health care costs. Under the CMS payment reforms, what would befall a patient whose diabetes did not improve—and who continued to incur costs—even with access to healthy food or a safe place to exercise? There might be no justification for continuing to provide these benefits, and patients living in adverse circumstances could find themselves having to “earn” access to basic necessities by getting healthier in specific ways.

Conclusion: The Politics of the Triple Aim

Challenges to achieving population health derive, at least in part, from what Fox calls the “determinants of policy for population health” (2006: n.p.). Among the several political factors he includes is the “diffusion of responsibility” for improving health status across agencies with disparate missions, such as education and employment. Efforts may also require a long time horizon and the practice of redistributive politics, neither of which predicts broad support or a smooth policy-making process. Determinants of policy for health care, however, are significantly different, resulting in the distribution of resources documented by Bradley and Taylor (2013). Especially after decades of neoliberal social policy, it is unsurprising that proponents of population health are turning to the health care system for financial resources.

Shortell, for example, proposes a “risk adjusted community-wide population health budget” provided by CMS and private health insurers to a “community-wide accountable entity for achieving predetermined quality and health status targets for, initially, a defined set of conditions” (emphases mine) (2013: 1). This is described as a “bold” strategy for “advancing population health” and envisions a level of social reorganization that Shortell admits may not be achievable, especially in the near future. Magnan et al. (2012) suggest a number of more modest strategies. These include diverting a portion of nonprofit hospitals’ community benefit or performance improvement funds; allocating a small share of health insurance premiums; and holding providers accountable for improvements in community health.

Perhaps the availability of resources accounts for the fact that although these proposals and others acknowledge the social contributors to population ill health, they nevertheless place health care entities at the center of its remediation. Even Casalino et al., who conclude that ACOs will concern themselves with patient but not geographic population health—and note that an average of 5 percent of hospitals’ community benefits currently go to community health improvement—go on to suggest that these provider organizations could “lead by example” and help “catalyze and cooperate with broader efforts” (2015: 827). One wonders what the other participants think of “integrator coalitions” or “accountable communities.” Surely sectors such as housing, education, and income maintenance have their own (contested) ideas about solutions to the problems their clients face and whether, given their tight budgets, they ought to care about cost containment in the bloated health care system. Whereas social programs might welcome an infusion of health care dollars, they have their own policies, politics, and logics of provision. These are not referenced in the many discussions of the triple aim.

In fact, many provider efforts to address population health consist of linking patients to existing social services or coordinating these with medical care. Washington Adventist Hospital, discussed above, organizes paid staff and volunteers to do safety checks and half-day clinics in identified “hot spots” where residents place frequent 911 calls. Beyond this, the hospital coordinates a patient's care through a network of community partners, such as churches and community gardens (Butler, Grabinsky, and Masi 2015), but does not seem to increase the capacity of these partners to meet the needs of the local population. The designer of the ACO, Elliott Fisher, and his colleagues imagine that an ACO serving a homeless population will use some of its shared savings to partner with local housing agencies to “help patients get into stable housing” (Lewis et al. 2012: n.p.), but it seems equally likely that an ACO will seek to “leverage” existing community resources to reach its own goals. Linking patients to a threadbare safety net may achieve patient population health but also edge out other community residents with an equal claim on services but without a powerful provider organization pushing them to the head of the line.

Some hospitals, health systems, and ACOs may distinguish themselves with efforts to address population health (e.g., Casalino et al. 2015), but they are new to this fight. Perhaps they should line up behind the weary poverty warriors to learn from them, even if hospitals are unlikely partners in social reform. Ironically, this may be a political strength of the triple aim. It undertakes to improve population health without signaling real social change. The triple aim may direct some resources to population health, but this will depend on what providers and payers can gain from the arrangement. Real population health requires social justice. The triple aim may prove salutary for some patients, but it is not a substitute for social justice and cannot be.

References