Social prescribing and the search for value in health care

Abstract

Health care systems throughout the United States are developing programmes to address patients’ ‘social determinants of health’ – such as housing, food, income and transportation. I investigate the concepts, technologies and infrastructures through which health care systems are turning towards ‘the social’ as an object of clinical knowledge and intervention. Proponents of this movement suggest that developing the clinical capacity to prescribe social resources promises to improve the ‘value’ of care, defined as the amount of health achieved per dollar spent. I argue that initiatives to improve value through social intervention are reconstituting broader arrangements of welfare provisioning.

Keywords:

• health care
• social determinants of health
• value-based care
• insurance
• health policy
• United States

Introduction

In recent years, health care systems throughout the United States have worked to develop the capacity to address the ‘social determinants of health’ (SDOH) in the context of clinical care. According to the World Health Organization (WHO), SDOH are ‘The conditions in which people are born, grow, live, work and age. These circumstances are in turn shaped by a wider set of forces: economics, social policies, and politics’ (World Health Organization, 2013). The movement to address SDOH in clinical contexts has involved experimenting with new practices of identifying patient needs for housing, food, income benefits, transportation, legal support and more, and implementing systems to fulfil these ‘social needs’. Sometimes referred to as ‘social prescribing’ (Alderwick et al., 2018), these health system interventions follow a biomedical logic of diagnosis and targeted treatment. The practice of social prescribing varies but generally involves screening for social needs, referring patients to appropriate social resources, and assisting with accessing and using the resources (Fichtenberg et al., 2019; Gottlieb et al., 2017).¹

Proponents of social prescribing emphasize a dual promise: to improve health outcomes and, at the same time, to decrease health care spending. This promise to impact measures of health and economic cost is consolidated into an important keyword in this movement: value. In this context, the term ‘value’ is derived from the policy agenda known as ‘value-based care’, which aims to maximize ‘the health outcomes achieved per dollar spent’ (M. E. Porter, 2010, p. 2447). Through the framework of value-based care, policymakers have diagnosed the US health care system as suffering from a problem of value, with extremely high costs accompanied by comparatively poor health outcomes. The United States spends more per capita on health care than any other industrialized nation, yet it ranks far below its peers in health outcomes and has some of the worst health inequalities (Squires & Anderson, 2015). Value-based care represents this as a problem of value, of spending too much on services that do not deliver results. Increasing value in health care means decreasing the costs of care while improving outcomes. Addressing patients’ social conditions by way of clinical practice has become a broadly popular strategy to achieve value.

The origins of social prescribing in the United States are often traced back to a physician named Jack Geiger. Working in Mississippi in 1964 and encountering an epidemic of child malnutrition, Geiger began prescribing food:

We gave these prescriptions, these food orders, to the people that needed them, and worked out a system under which they could go to the black grocery stores in any of the 10 towns in our service area, and the grocery store would fill the prescription for food and send the bill to our health center. We paid those bills from the health center’s pharmacy budget. (Geiger, 2005, p. 7)

As Geiger recounts, social prescribing was opposed by state authorities who saw it as an unauthorized expansion of welfare provisioning: ‘The state of Mississippi found out about this and concluded that, clearly, Soviet communism had arrived in the Delta’ (Geiger, 2005, p. 7).

By contrast, the contemporary movement to expand social prescribing represents it as a potential source of value for the capitalist economy. In a 2018 speech, the US Secretary of Health and Human Services and Trump administration official, Alex Azar, described how addressing SDOH can reduce expenditures:

The root cause of so much of our health spending: social determinants of health … [This] is an abstract term, but for millions of Americans, it is a very tangible, frightening challenge: How can someone manage diabetes if they are constantly worrying about how they’re going to afford their meals each week? … One [health system] in Chicago, for instance, began screening high-risk patients for malnutrition, and then supporting them after discharge from the hospital with follow-ups, referrals, and nutrition coupons. The savings were huge: more than $3,800 per patient. (Azar, 2018)

Secretary Azar (2018) went on to say, ‘Social determinants are closely integrated into the priority I’ve laid out to move toward a value-based healthcare system – one that delivers better outcomes at a lower cost’. A value-based approach, according to Azar (2018), preserves ‘what is unique about our American system: its decentralized nature and the key role played by the private sector and civil society’.

How did the practice of prescribing food transform from a communist threat into an effective strategy to increase value? How have renewed investments in social assistance been aligned with reigning rationalities of governance in the United States? In this paper, I examine how social prescribing has been implemented as a solution to problems of value in the health care economy. I find that value-based care increasingly holds health care systems responsible for addressing SDOH and, in doing so, creates new arrangements of responsibility for poverty and social welfare. I argue that the search for value in health care is reshaping ‘the social’ in ways that make it amenable to demands of privatized health care and dominant neoliberal trends in governance.

Method

This paper is informed by more than five years of ethnographic research in California, including a team-based study of primary care-based social needs interventions from 2015 to 2017 (Fleming et al., 2017, 2019, 2021) and ongoing fieldwork (2018-) in hospital emergency departments, community settings and government health departments where social needs interventions are being discussed, designed, and implemented. This research has examined how health care providers address SDOH in their everyday work and how they navigate ethical and economic imperatives in providing care for the poor. The questions that I ask in this paper – How has social prescribing been implemented as a solution to problems of value? How is ‘the social’ in the SDOH made concordant with reigning neoliberal rationalities? – emerged from these projects. To answer these specific questions, I analysed an additional 21 interviews (2019–2021) with experts in the field of health system social interventions (including health care providers, administrators, data analysts, policy specialists and researchers), which included a focus on the role of value-based policy incentives in projects to integrate health care and social services. I also analysed written policies, white papers from governments and non-governmental organizations, conference proceedings, and published literatures. By using fieldwork data from multiple sites and areas of expertise along with a variety of written documents, I approached social prescribing as an assemblage articulated across an array of policies, concepts and research programmes, and administrative and clinical practices (Hogle, 2019; Wahlberg, 2022). Within this assemblage, I analysed value-based care as a policy agenda shaping the epistemic, normative, technological, and political economic elements of social prescribing (Shore et al., 2011). I organized findings broadly by the rationale (the why) and the practice (the how) of social prescribing then delineated sub-themes in accordance with the sections presented below.

Examining ‘the social’ in social prescribing

Drawing from genealogical studies of ‘the social’ (Castel, 2017; Donzelot, 1979; Foucault, 2003) and science and technology studies (STS) (Latour, 1993, 2005; Shapin & Schaffer, 2011), I approach the social as a domain that is constructed historically through practices of knowing and intervening into specific problems of governance. This approach situates the social as a ‘problematic unity’ (Dean, 2010), a precariously accomplished collectivity or domain of shared responsibility and interest, and an assembled set of bonds and associations that are always threatening to dissolve (Latour, 1993, 2005). Histories of the social demonstrate that, rather than being a universal domain of forces and processes that was discovered by modern social sciences, the social gets constructed in particular times and places as a solution to distinctive problems of managing populations. Modern states developed techniques of forging the social in response to threats of unemployment and worker political unrest, and subsequently rearticulated social governance in relationship to a proliferating set of problems – e.g. pauperism, orphaned children, single mothers, the elderly, the insane, the criminal, the homeless (Castel, 2017; Dean, 2010; Deleuze, 1979; Donzelot, 1979).

To investigate how practices of social prescribing are constructing the social in particular ways, I take up the strategy of STS to attend to knowledges, technologies, and infrastructures. STS scholarship has demonstrated how formations of the social are mediated by material practices of counting, measuring, and data aggregation as much as by policies or economic structures (Jasanoff, 2004). Rather than taking recourse to social categories as self-evident and given forms to explain health system practices, I ask, following Latour (2005), ‘What are the new institutions, procedures, and concepts able to collect and to reconnect the social’ (p. 11)? I therefore track how, in the search for value, health care systems are deploying diverse epistemic and technical elements to re-assemble the social.

Over the nineteenth and twentieth centuries, iterations of the social were variously assembled and scaffolded through techniques of statistical calculation and risk thinking (Hacking, 1990; T. M. Porter, 1995), concepts of population (Foucault, 2009), and technologies of insurance and risk pooling (Ericson et al., 2000; Ewald, 1991). Social insurance has been a technical means of assembling individuals into collectivities with shared interests through the socialization of risk (Collier et al., 2021; Ewald, 1991). Insurance technologies also worked to undergird the moral orders binding the social, especially the value of solidarity amongst citizens (Castel, 2017; Dean, 2010).

The social has undergone multiple transformations in Euro-American contexts from its origins in eighteenth century Europe, its crystallizations in twentieth century welfarism, and its dismantling under neoliberalism. The shift towards neoliberal governance in the late twentieth century saw a range of projects to destroy the social ‘conceptually, normatively, and practically’ in the name of freedom for markets and individuals (Brown, 2019, p. 28). Neoliberalism is associated with the privatization and ‘unpooling’ of risk in insurance markets (Ericson et al., 2000), the individualization/de-socialization of responsibility, and the dismantling of social welfare institutions (Wacquant, 2009). This includes state withdrawal from governing the population as a social aggregation and a turn towards addressing the population as a collection of self-responsible individuals (Brown, 2019; Schram, 2000). Health knowledges and governing technologies increasingly focused on disciplining and managing the behaviour of autonomous individuals, leading scholars to consider whether we have witnessed a ‘death of the social’ (Rose, 1996).

During the late twentieth and early twenty-first century, health scientists and policymakers reasserted a conception of the social as a ‘determinant’ of health. The SDOH framework has undergirded a robust research programme demonstrating linkages between social and economic conditions and health outcomes, often emphasizing differences between rich and poor populations (Friel & Marmot, 2011). Drawing from a long history of social medicine, the SDOH framework reframes disease inequities in explicitly moral terms as injustices to be corrected through social and economy policy (Marmot et al., 2008). In understanding social and economic conditions as direct contributors to health outcomes, SDOH was intended to remedy an overemphasis on individual behaviour and responsibility. In this sense, SDOH reiterates modern conceptions of the social that recognize individual responsibility as ‘always traversed by external determinations’ including family background, social class and a range of social and economic forces (Rose, 1996, p. 333). As I will suggest, while social prescribing deploys the SDOH framework, it also re-individualizes social conditions as diagnosable ‘social needs’ in terms that are broadly concordant with neoliberal and biomedical logics.

Accounts that understand market-based governing principles to be associated with a waning of the social – a withdraw from governing the social and an investment in disciplining individuals – overlook the ways that the social has been reproduced via new knowledges and technologies. The current expansion of social prescribing compels us to consider how the social may be (re)assembled today within conditions thoroughly organized by the market-based governing rationalities of neoliberalism. Implementing social prescribing involves a range of knowledges and technologies, including professional expertise about the SDOH, social needs screening tools, clinical systems and workflows, digital infrastructures and data analytics, as well as financing and governance structures. In the search for value in health care, this array of elements is coalescing around the social as a principal site of experimentation and innovation.

The value proposition of social prescribing

The proliferation of social prescribing in the United States is driven by the proposition that addressing social conditions for some patients will improve the value of health care. Research articles and reports in the field typically begin by describing the deep dysfunction of the political economy of health care: At nearly 18 per cent of gross domestic product GDP in 2018, US health care spending accounts for a greater proportion of the economy than any other industrialized nation and is projected to continue growing faster than the general economy (Centers for Medicare and Medicaid Services (CMS), 2020a). Meanwhile, health outcomes are much worse than peer nations, with increasing numbers of poor people experiencing a heavy burden of illness. In addition to growing concerns about health inequalities, political fears revolve around the concern that such a large slice of GDP devoted to one industry may be harmful to the economy and ‘reduce the resources available for other worthy government programs, erode wages, and undermine the competitiveness of US industry’ (Berwick & Hackbarth, 2012, p. 1513). Because employers in the United States cover the cost of health insurance for employees, rising health care costs are thought to have caused the wage stagnation US workers experienced over the last several decades (Lee, 2020). Further, a significant proportion (as much as half) of the health care spending is considered wasteful, spent on overpriced and unneeded services (Cutler, 2018).² At the core of each of these problems is the concern that health care is misvalued, costing far more than the benefit the services offer.

According to proponents of social prescribing, poverty plays a pivotal role in causing the overspending on health care for an important subset of the population. This is due to a small minority of people living in poverty who become severely ill from preventable causes and subsequently consume large amounts of health care. In the United States, about 5 per cent of the population account for 50 per cent of medical spending (Mitchell, 2020). A substantial proportion of these high-cost patients experience poor social conditions, including low or no income, homelessness, food insecurity, transportation barriers, interpersonal violence and social isolation. These social conditions produce heavy burdens of chronic illness and disallow proper self-management, resulting in advanced forms of illness that lead to repeated acute care visits – i.e. emergency room visits and hospitalizations. In the United States, decades of disinvestment in social services have resulted in a group of impoverished people becoming an extremely high cost-burden to the state and insurers via their repeated acute medical crises. Health systems thus seek to intervene into these patients’ social conditions to help prevent and manage disease and reduce health care spending. While the social prescribing literature largely focuses on ‘unnecessary’ use and over-consumption of care among patients with unmet social needs, analysts have long recognized a multitude of drivers of health care spending in the United States, especially high prices (Anderson et al., 2003).

The concentration of illness and health care spending in zones of poverty was problematized for social governance in the first decade of the twenty-first century with the development of new data analytics and surveillance techniques (Fleming et al., 2021). Health systems began experimenting with forms of surveillance that integrate data on costs, health, and geo-spatial location and discovered that small segments of the population – sometimes referred to as ‘super-utilizers’ – who often reside within the same geographical zones, or ‘hot spots’, accounted for large proportions of health care spending (Finkelstein et al., 2020).

The rationale for investments in social prescribing also relies on the growing scientific evidence that the SDOH play a significant role in shaping health outcomes. A report from the National Academies of Sciences, Engineering, and Medicine (NASEM) (2019b) titled Investing in interventions that address non-medical, health-related social needs, is representative of the narrative put forth by my interviewees and the literature. The authors begin the report by stating the urgent need to ‘keep health expenditures from consuming an ever-greater portion of US economic output’, and suggest that the ‘growing recognition that social determinants of health are important influences on population health’ may offer a solution to this problem. The authors then cite estimates that social and environmental factors account for nearly 50 per cent of determinants of health, while medical care (access and quality) determine only 10–20 per cent of health outcomes.

Given that social determinants affect health outcomes more than clinical care, proponents of social prescribing suggest that health care resources should be reallocated to address SDOH. Researchers have argued that people who live in countries where governments spend more on social services compared to medical care have better health and live longer. On average, European countries spend $2 on social services for every $1 spent on health care, while the United States spends less than 60 cents on social services per dollar spent on health care (Bradley & Taylor, 2013). This broad-scale ineffectiveness of medical spending to produce better health for the nation is represented as a problem of ‘value’, of not achieving adequate health per dollar spent. The nation’s investments in health care – by states, public and private insurers, and individuals – are not translating to good health, and stakeholders must consider how to invest differently to achieve better health. As a health system manager told me, ‘We are spending in the wrong places on the wrong things’.

A wide range of new policies have included incentives for health systems to take up social prescribing as an avenue to increased value. Starting in 2017, the Centers for Medicare and Medicaid Innovation devoted $157 million in grants to support health care-based social interventions, which supported programmes across 21 states. In California, where my fieldwork is based, Medicaid (public insurance for low-income Americans) authorized up to $1.5 billion in spending (matched by another $1.5 billion of state funding) between 2015 and 2020 for health systems to better coordinate health care and social services and to address social determinants in health care settings. While California devoted significant state funds to expand the programme, 41 other states participated in this federal waiver programme allowing them to repurpose Medicaid funds for innovative care activities, many of which involve social prescribing (Kaiser Family Foundation (KFF), 2020). Social prescribing is also growing in the United Kingdom, Australia, Canada, the EU and Asia, and often includes a focus on cost reduction despite the vastly different health insurance and social care systems across these regions (Morse et al., 2022). However, the problem of ‘value’ is uniquely articulated in the United States, as the result of its market-based health care system, the significant role of the private sector in health care, and much higher spending on care.

In sum, while modern social governance initially took the industrial worker as its primary object of assistance and reform, twenty-first century social investments are consolidating around the chronically ill consumer of health care services. In the discourse of social prescribing proponents, poor social conditions are driving high rates of chronic illness, causing an extreme over-consumption of care among some patients, and leading to ever-greater spending on health care. Investments in social assistance may therefore protect the national economy from the threat posed by the costs of over-consumption and overspending on care.

Social prescribing for value

Implementing social prescribing involves developing new clinical practices and adapting existing organizational processes to address patient needs for housing, food, income benefits, transportation, legal support and a range of other resources. In this section I identify three activities of social prescribing – targeting, screening and coordinating social resources – that each demonstrate how the social is rearticulated within a biomedical logic and in ways intended to improve value.

Targeting

Targeting describes how sub-populations are prioritized and managed according to their possibility for generating return on investment (ROI) within a value-based system. ‘We need to target resources. The evidence on ROI is that you have to target resources’, a health policy expert told me during an interview. The purpose of targeting is to direct resources to individuals for whom access to social resources could influence their health and consumption of health care services. That is, the first step in designing a social prescribing programme is to identity sub-populations who are ‘impactable’ in ways that can generate value. Impactable populations, most often, are composed of people with chronic illness who have the potential to improve with better illness management, in contrast to people with end stage or congenital conditions.

Targeting in social prescribing is thoroughly technological enterprise, integrating new data surveillance systems, predictive algorithms and econometric analyses. Targeting relies on the technological ‘ability to make visible individuals who may be harbingers of risk and reassembling them into unique categories’ (Hogle, 2019, p. 557). That is, health systems manage and define populations by collecting vast amounts of data and defining new categories of prioritized individuals based on their likelihood of generating value.³

So-called ‘super-utilizers’ of acute care services are one of the main ‘target populations’ of social prescribing programmes in the United States, as their reductions in acute care utilization promise the largest ROIs for health systems (Fleming et al., 2019). Yet, interventions continue to expand to other groups whose health and costs may be impacted by access to social resources. A health policy expert told me, ‘You need to first build the value from that initial target population and then titrate out to other populations’. That is, health systems should begin by targeting programmes to individuals whose improvements are most likely to build value and then expand to populations that may produce smaller ROIs.

The target populations are typically defined by some combination of health conditions (e.g. heart disease, diabetes, asthma), social characteristics (e.g. veterans, recently incarcerated, low-income), type of social need (e.g. housing, food assistance, transportation), and utilization history (e.g. frequent emergency department visits or hospitalizations) (Gottlieb et al., 2017). Example target populations include children from low-income families visiting the emergency department, homeless adults with alcohol-related emergency visits, recently incarcerated people with chronic illness, or socially isolated older adults with barriers to transportation. Across the field of social prescribing, the most frequently targeted needs include housing, food, and transportation (Alley et al., 2016; Fichtenberg et al., 2019). Health systems have also developed interventions to target needs for income and economic security, education and employment, social isolation, legal services, childcare, and services for intimate partner violence and community violence, with most programmes designed to identify and address multiple social needs (Gottlieb et al., 2017).

In one social intervention programme where I conducted fieldwork, patients were targeted if they had three or more hospitalizations for chronic illness in 12 months and documented social needs, with the rationale that targeting this population would result in the largest reductions in health care spending. While this programme targeted patients based on previous utilization, another programme I studied targeted patients with risk of future utilization, but who had not yet become the highest-cost patients. As a programme administrator explained in an interview, ‘We identify patients based on what we call ‘rising risk’. We use something called the predictive risk model, where we have more than a 100 different data elements that pull into this risk model’, including data on past use of health services, medical and mental health diagnoses, and a range of other factors including history of homelessness or involvement with criminal justice or foster care systems. The risk model predicts who is, as she said, ‘most likely to have an avoidable emergency room visit or avoidable inpatient admission’ and these individuals are prioritized for social care.

The practice of using risk models and predictive algorithms to target social resources diverges from the reigning logics of distribution governing means- and morality-tested social welfare programmes.⁴ As health systems integrate social services into care delivery, the contrasting rationalities of risk-based targeting in biomedicine and morality and needs-based eligibility criteria in social services increasingly collide. Interviewees described a ‘clash of cultures’ between these two service sectors. The comparatively greater political and economic power of the health care sector in the United States may mean that social services organizations will be compelled to remake themselves to serve the goal of achieving value in the health care sector (Taylor & Byhoff, 2021). This includes taking up targeting for ROI as a dominant logic of distribution. An administrator directing efforts to integrate a large public health care system with a social services agency told me, ‘Targeting will be important for health care and social services. I think both’. Participants expressed concern that sending medically complex patients to the front of the line for social services may violate the norms of fairness within the social services sector.

Screening and documenting

A second process whereby social prescribing is reformulating the social is in the integration of social diagnostics and information into health care systems. Researchers and health systems are working to develop screening tools to identify social risks in the context of clinical encounters and to integrate the resulting social diagnoses into medical data infrastructures. The movement towards social prescribing is working to refashion social conditions as medical diagnoses, concordant with systems of disease classification and biomedical infrastructures. This involves representing social risks such as housing and food insecurity as ‘diagnoses’ for which patients can ‘screen positive’, triggering a social prescription.

There are multiple screening tools being developed and evaluated, with two of the most widely used being the Protocol for Responding to and Assessing Patients’ Assets, Risks, and Experiences and the Accountable Health Communities Health-Related Social Needs Screening Tool (CMS, 2019). There is no standard practice for conducting social screenings and programmes continue to experiment with screening patients in locations such as waiting rooms, in offices during clinical intake, bedside in the hospital, by phone, or using tablets.

Increasingly, health care providers understand social screening to be an essential component of their professional expertise.⁵ Most major health care professional associations now advocate for universal social screening and better training for providers to conduct assessments and provide referrals for social resources. For instance, in 2018, the American Academy of Family Physicians (AAFP) launched the EveryONE Project Toolkit, intended to equip physicians with the skills and capacities to address SDOH in their practices. The organization released a Social Determinants of Health Policy, which states, ‘The AAFP supports the assertion that physicians need to know how to identify and address social determinants of health in order to be successful in promoting good health outcomes for individuals and populations’ (AAFP, 2019).

Health care systems across the United States are implementing social needs screening, in many cases without systems in place to adequately meet the identified needs. Some researchers argue that eliciting information about social conditions such as homelessness or food insecurity without subsequently offering services to address these issues can lead to stigma, frustration, and distrust, and therefore is potentially harmful to patients (Garg et al., 2016). Those promoting this perspective have pointed to WHO guidelines asserting that a central tenet of ethical medical screening is the availability of appropriate referral services (Wilson & Jungner, 1968). Others contend that even in the absence of adequate social resources, assessments of social conditions can have a direct benefit for the patient. In this view, social needs screening may improve care and health outcomes by enabling individual tailoring of clinical care (including practices of drug prescribing, scheduling or communication) and create deeper understanding and trust between patient and provider (Ford-Gilboe et al., 2018; NASEM, 2019a).

To accommodate social diagnoses, standardized medical classifications systems are being revised and expanded. For instance, the International Classification of Disease (ICD), the most widely used medical classification system, has recently added diagnostic codes for a range of social conditions. The ICD classification system, for instance, uses the codes Z59.0 for homelessness, Z59.41 for food insecurity, and Z59.5 for extreme poverty. While existing diagnostic classification systems contain an array of diagnostic codes for social conditions, researchers have found they are largely inadequate for use in social prescribing programmes (NASEM, 2019a). They often do not correspond to the social domains encompassed by screening tools or available resources for intervention. One study found that a widely-recognized diagnostic coding system has 14 codes for utilities (e.g. lack of cooling, lack of heat, no electricity) but no codes of ability to pay utility bills, which would correspond more closely to the resources provided by social prescribing programmes (Arons et al., 2019).

This emphasis on using the proper diagnostic codes to characterize patients’ social conditions reflects the biomedical logic of linking accurate diagnosis and targeted treatment. Representing social needs as diagnoses that can be integrated into biomedical infrastructures may be working to individualize social conditions, emphasizing ‘prescription’ as a solution to poor social conditions rather than structural or policy change.

Coordinating social resources

Social prescribing is refiguring the relationship between health care and social services sectors. In most social prescribing programmes, after a patient ‘screens positive’ for a social need, a staff member assists with accessing and using the needed social resource.⁶ Generally, the social resource is administered by an organization external to the health care system. Delivering social resources via biomedical institutions therefore requires ‘cross-sector collaboration’ in which health care and social service organizations integrate and align workflows, goals, funding and governance structures. Such collaborations enable staff to work across the boundaries of these traditionally siloed service sectors. At the same time, as health care systems interface with social service organizations, social services are increasingly infiltrated and dominated by medical languages, metrics and goals (Taylor & Byhoff, 2021).

To improve value, health care systems must prioritize metrics and operational goals that demonstrate the impacts of social prescribing activities on health and the costs of care. Some interviewees explicitly raised the concern that this demand might ‘medicalize’ social conditions and force social services into formalized system of diagnosis, referral and treatment. One policy expert told me: ‘Medicalization of social services would be a bad outcome. I want to think much more in terms of partnership … We are really only at the beginning of integration of health care and social services. When we get better at this, it’s more about really finding out what CBOs [community-based organizations] care about, what their metrics are, their reporting metrics, their funding structures, and integrating those, rather than imposing the metrics that we want’. Other interviewees were less optimistic about the prospects of equitable power sharing in cross-sector collaborations, given that much of the incentive for investing in these programmes is driven by health care financers including public and private health insurers and states.

While health systems are rapidly expanding social intervention capacities, comparable investments are not being made in the social service sector, creating what has been called the ‘road to nowhere problem’ (Alderwick et al., 2018): the systems of screening and coordinating for social needs within the health care system lead to a dead end when state- or community-based resources are unavailable. As health care systems ramp up social intervention activities, ‘proactive and systematic screening will likely increase the number of people who seek assistance from the [social service] agencies’ (Kreuter et al., 2020, p. 679).

Proponents of social prescribing have therefore begun systematic efforts to assess and ‘optimize community capacity to address health-related social needs’ (CMS, 2020b). In the process, it was discovered that ‘There have been no published data that describe social services system capacity – broadly defined as the amount of assistance a community can produce to address social needs’ (Kreuter et al., 2020, p. 679). The geographic zones in which health systems operate each contain their own diverse set of community-based and governmental resources including, for instance, faith-based food pantries, municipal shelter systems and county cash assistance programmes. As health care systems seek to leverage social resources to impact value, health researchers are now providing the most thorough mapping and quantification of social resources to date. While modern social governance was born in close conjunction with states’ capacity to aggregate data on land, population and economic goods within their national boundaries (Hacking, 1990), social governance today may be reimagined via a medical mapping, quantification and redeployment of resources.

In the face of often-limited social services, according to a National Academies (2019a) report, ‘The health sector must decide whether to build capacity to directly assist patients’ non-medical needs or, alternatively, to partner with community-based organizations [and state agencies] that have experience in addressing social needs and the necessary capacity to address those needs’ (p. 97). Increasingly, health care systems and insurance organizations are opting for the former strategy and investing directly in social resources. Horwitz, Chang, Arcilla and Knickman (2020) reviewed the activities of 78 health systems and found, over two years, at least $2.5 billion of health care dollars being redirected to investments in social services, with $1.6 billion committed to housing resources. The programmes where I conducted fieldwork generally took a hybrid approach, relying mostly on external social service organizations, but devoting some health care dollars directly to services such as onetime cash payments for rental assistance or utility bills, cell phones and transportation services.

Some health systems take a more expansive view, redefining themselves as ‘anchor institutions’, or institutions that strategically wield their political economic power to reshape social conditions. Health systems are comparatively well-financed and therefore can ‘harness their economic power to inclusively and sustainably benefit the long-term health and well-being of the communities they serve’ (Healthcare Anchor Network (HAN), 2020). For instance, the University of California, San Francisco’s Anchor Institution Initiative states that this health system ‘is committed to working with the community to leverage its $7 billion operating budget to strengthen its ability to improve the health of the city’s underserved and under-resourced communities and to promote health equity’ (UCSF Anchor Institution Initiative, 2020). Anchor institutions seek to influence local economies through strategies such as employing people from underserved communities, supply chain and purchasing policies, and business development loans. As health systems are often one of the largest employers within a geographic area, they have begun to reconceive their training and employment practices as interventions into the SDOH. This anchor mission, then, represents economic activity (labour markets and commodity chains) as a domain of health system intervention appropriate for influencing social conditions which in turn promote health.

Value-based care as a technology of social governance

Social prescribing literatures represent the overconsumption of care among people experiencing poor social conditions as a key factor driving high levels of health care spending. As I have suggested, proponents of social prescribing argue that new investments in the SDOH can mitigate this overspending and its threat to the economy. In this section, I consider how value-based care – a policy agenda to maximize health outcomes achieved per dollar spent – is deployed as a technology of governance intended to spur investments in social welfare, and to do so in ways that are aligned with reigning logics of neoliberal governance.

Risk sharing

Value-based care finds its expression through new financial and analytic techniques of ‘risk sharing’ through which health care providers and payers (typically insurers and states) are encouraged to align their interests. Some authors use ‘value-based care’ and ‘shared-risk payment models’ interchangeably to refer to contracts between payers and providers in which financing is contingent on the success of health systems in improving care and decreasing costs. Value-based care contrasts with longstanding ‘fee-for-service’ payment arrangements in which providers are reimbursed for each service they provide. Under fee-for-service, providers are financially rewarded for increasing volume of services. While fee-for-service incentivizes providers to maximize services, payers, on the other hand, are incentivized to limit spending. Payers have experimented with implementing strict clinical guidelines to reduce unnecessary treatments and tests, which have been criticized for decreasing the quality of care and deepening health disparities (Rylko-Bauer & Farmer, 2002). In the view of policymakers, fee-for-service positions providers and payers as adversaries with competing interests, with providers seeking to increase volume of services and payers seeking to limit volume, and with neither actor having significant incentives to improve the quality of care.

While the past several decades of neoliberal management has worked to ‘unpool risk’ in health insurance markets and society (Ericson et al., 2000), value-based care is working to forge new domains of shared risk. Risk-sharing, as Hogle (2019) observes, ‘puts providers in the position of arbitrating financial and health risks, blurring their role with that of insurers’ (p. 557). The assumption of value-based care is that, by spreading risk from insurers to health care providers, providers will act more like businesses or entrepreneurs within a health care market, competing to provide the highest value services to ‘consumers’.⁷

By ‘increasingly tying payment to value’ (Burwell, 2015, p. 897), value-based care merges the interests of insurers and providers and extends the influence of market rationalities within health care. Importantly, to meet cost and quality objectives, health care providers are allowed more flexibility in how they use resources and organize care delivery. Under a fee-for-service system, each billable service is rigidly defined by the payer. Value-based care, by contrast, typically allows more freedom for experimentation, including by allowing health care dollars to be spent on social prescribing activities. Within the logic of value-based care, exposing health care providers to greater risk while simultaneously allowing more flexibility promotes innovation and leads to the creation of higher-value services. This policy agenda asserts that innovation goes hand-in-hand with exposure to risk and competition, reflecting the productive nature of risk within market rationalities (Brown, 2019).

Policymakers and researchers understand these new value-based arrangements to be driving the expansion of social prescribing. When health care organizations ‘are at risk for health care costs, they have a financial incentive to make investments in social care’ (NASEM, 2019a, p. 114).⁸ Value-based care ‘can increase attention to and investment in the social rather than medical determinants of health by creating new, attractive avenues for private investments in programmes and services that both improve population health outcomes and allow governmental entities to achieve greater value and efficiency in the allocation of public resources’ (Lantz et al., 2016, p. 2054).

The logic of value-based care assumes that, through exposure to risk, health care providers will make investment decisions from the perspective of value production and that, by doing so, they will shift resources to social interventions. In this way, value-based care may function as a technology of the social (like risk pooling in social insurance), offering a justification for investments in social assistance and realigning interests among states, insurers, service providers, and people living in poverty.

Markets for social investment

Value-based payment arrangements also bring capital and markets into these new structures of aligned interests, primarily by re-coding the potential savings accrued through more efficient systems (often publicly funded programmes serving people living in poverty) as promissory capital unleashed to investors. Hogle (2016, 2019) notes that, ‘Business analysts optimistically estimate that more than $300 billion a year in new value could be created in healthcare by 2021, with two-thirds of that coming from reductions to national healthcare expenses’ (p. 380, emphasis added).

Under a value-based system, health care providers are encouraged to conceptualize the distribution of health care resources in terms of ‘investment’, with the aim of maximizing their ‘ROI’. Through the implementation of market-like incentives in value-based payment arrangements, health care organizations can directly benefit from the ROI generated by investments that successfully increase the value of care. ROI accrues to the organization through a combination of direct payouts for meeting metrics of quality and cost efficiency, as well as through ‘shared savings’ arrangements in which they keep a portion of the money saved through decreased spending. Health systems that do not meet certain benchmarks may be subject to penalties.⁹

Value-based care is also a site of technological innovation requiring intensive capital investment. This includes technologies for surveillance and data analytics, as described above in the context of ‘targeting’. Further, for health care systems to ‘operate in a value-based environment’, as interviewees often described it, they also need to collect and report data for auditing purposes. To receive payments under a value-based system, health systems need to invest in technologies and infrastructures that enable them to demonstrate the impacts of social prescribing activities on health and costs. These audit needs have opened significant new markets for private technology firms to sell tools for data collection, analysis and reporting.

The work of delivering integrated social and medical services is also facilitated by redesigned infrastructures that incorporate social risk screening tools, display social information for providers, and enable referrals and ‘interoperability’ with social service providers (Gold et al., 2017). The three health systems in California where I conducted fieldwork each contracted with private sector vendors to create electronic medical records in which service providers could access integrated documentation on health care, social services and a range of community-based services including behavioural health treatment. The examples cited here are just a few of the elements of an ‘explosion of new business-oriented tools, products, and consulting services designed to assist providers in managing the health care use, costs, and outcomes of the populations for which they are financially responsible’ (Lantz, 2019, pp. 36–37).¹⁰

As industries rapidly create new technologies and services to facilitate improvements of value in health care, health systems are confronted with the problem of how to capture the ROI from investments in social prescribing. The benefits of providing social services and resources such as housing, food, income, transit, or other resources tend to be spread out broadly, accruing to numerous entities beyond the health care organization that makes the investment. In this sense, the character of the social as a sphere of collective living presents significant challenges to reconceiving of it as a site of capital investment. Social prescribing programmes may result in a reduced burden to public education systems, social services agencies, criminal justice systems, and beyond. ‘Such investments are in effect public goods, and thus benefits cannot be efficiently limited to those who pay for them – which makes it more difficult to capture return on investment’ (Nichols & Taylor, 2018, p. 1223).

Practitioners I spoke with referred to this as the ‘wrong pocket’ problem, defined in recent National Academies of Science conference proceedings as follows: ‘Where organization A might invest the money, but organization B reaps the benefits at some point in the future … [which is] a significant barrier to investing in successful [social prescribing] programs’ (NASEM, 2019b, p. 54). When the benefits accrue across multiple organizations, there is no mechanism for capturing payments or tax from those entities and directing it back to the investor. Further, in a competitive marketplace of insurance products, consumers change health plans, and when they do, a competing organization may reap the benefit and capture the ROI from another’s investment in their social welfare.

Several interviewees noted that it is easier for health systems to capture ROI when interventions are further ‘downstream’ – focused on individual-level needs of poor people who already face significant, expensive illness – rather than ‘upstream’, or preventative. Downstream interventions that lift people out of extant social and health crisis offer demonstrable improvements in health outcomes and costs relative to baseline. It is more challenging for organizations to capture ROI of investments in upstream interventions, which may prevent people from sliding into health crisis and becoming expensive patients in the first place. The diffuse, socially shared nature of upstream interventions – as well as the long timeframe for the benefits to accrue – are limitations for the business model forged by value-based care.

Conclusion

The question is no longer whether there is an appropriate role for the US health care system in addressing the social determinants of health, but what that role is, how to create the right policy context for innovation and how health care can partner more effectively with providers of social services to meet patients’ most pressing needs given the fragmented, typically under resourced nature of the social sector (Solomon & Kanter, 2018, p. 48).

In this paper, I interpreted the movement towards social prescribing as part of a shifting, historical construction of social governance. Genealogical accounts of social governance suggest that since, at least, the nineteenth century, the rationale for state action at the level of the social represented poverty (and especially the economic precarity of workers) as a threat to the economy and to liberal-democratic political regimes (Castel, 2017; Dean, 2010; Deleuze, 1979; Donzelot, 1979, 1988; Foucault, 2003). Poverty and inequality posed a threat to the legitimacy of the liberal-capitalist order, as these conditions set the stage for revolutionary worker organizations and political demands for a socialist state. The ‘social question’, as it became known through the nineteenth and twentieth centuries, asked, what is the role of the state in addressing social conditions to maintain political stability and protect the national economy? Social insurance, population sciences and statistics emerged as technologies of governance in response to the social question, offering new strategies to create collective interest and industrial harmony (Hacking, 1990; Rabinow, 1995).

My analysis of social prescribing found that, in the United States, we are witnessing an expansion of social governance which has been concerned with addressing poor social conditions (represented as SDOH) as a means of protecting the economy from overspending on health care. Value-based care deploys metrics, technologies, policies and modes of accountability intended, in part, to spur investments in social assistance as a means of increasing value. I suggested that value-based care works as a technology of social governance by strategically spreading risk in ways that produce new arrangements of interest among insurers (public and private sector), health care providers and technology firms in improving the conditions of people living in poverty.

While value-based care produces new alignments of interest, it has done so within a context thoroughly defined by a neoliberal rationality. Neoliberalism has typically been associated with the active dismantling of social assistance programmes. By contrast, the movement towards health system social intervention demonstrates how social governance may be reconstructed within assemblages concordant with market rationalities. The rationale for social assistance retains, at least rhetorically, the objective to bolster the economy, but has shifted focus from the dangers posed by disaffected workers to the dangers of over-consumers of health care services.

The proliferating discourse on the social within health care is diverse, offering a multitude of often-conflicting visions of collective responsibility. As illustrated by the ‘anchor institution’ concept, in some cases, the health care sector is taking an ever-more expansive view of its role in affecting social conditions. However, the search for value in health care generally involves the use of data analytics to direct social resources in increasingly narrow and targeted ways with the aim of maximizing impact on health care utilization among select sub-populations. Increasingly, people exposed to poverty may become worthy of social resources when their health care consumption becomes excessive, or they are predicted to pose a future risk to the health care system. This indicates a significant shift away from needs- or rights-based claims to social welfare, and raises further questions about how biomedical, algorithmic and economic rationalities are converging to remake social governance today.

Ethical approval statement

This research was approved by the Institutional Review Board at the University of California, Berkeley protocol number 2018-07-11225. All participants provided informed consent prior to data collection.

Acknowledgements

The author is grateful to Amanda Brewster, Jerry Zee, Michael D’arcy and Craig Willse for their generous feedback. I also thank the anonymous reviewers for their excellent suggestions.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This project was supported by grant number K01HS027648 from the Agency for Healthcare Research and Quality. The content is solely the responsibility of the author and does not necessarily represent the official views of the Agency for Healthcare Research and Quality.

Notes on contributors

Mark D. Fleming

Mark D. Fleming is an anthropologist and Assistant Professor in the School of Public Health at the University of California, Berkeley. His work focuses on the political economy of chronic illness, and the contemporary interchange of medical, social and carceral forms of governance.

Notes

1 ‘Social prescribing’ is a contested term, with some suggesting that it invokes an overly-medicalized image of the practice of coordinating social resources in health care contexts. Another common label is ‘social needs intervention’ or ‘health care sector strategies to improve social conditions’.

2 There is disagreement around each of these concerns, including whether it is bad for a country to spend so much on health care. ‘There is no economic law that governs how much money should be spent on any industry’ (Cutler, 2018, p. 493).

3 These practices are an extension of evidence-based, preventative medicine, a modality of care that arose in the twentieth century that makes treatment decisions based on ‘risk factors’. Social scientists have examined how risk-based thinking enabled the expansion of medicine’s boundaries and the created new objects of intervention including the not-yet-sick bearer of risk factors (Dumit, 2012).

4 The use of risk models and algorithms to determine eligibility has expanded, particularly in the fields of homelessness services, where risk calculations are used to determine access to housing and supportive services (Eubanks, 2018).

5 The emerging literature and statements by professional associations very rarely reference ‘social medicine’ as a longstanding precursor to these practices. As Adams, Behague, Caduff, Löwy and Ortega (2019) argue, multiple genealogies of social medicine are embedded in contemporary health care, public health, and global health projects concerned with ‘social determinants of health’ in largely unacknowledged ways.

6 The professional titles of the staff accomplishing this work varies widely across programmes, but often involves nurses, social workers, community health workers, patient navigators, or physicians. Further, the intensity of programmes varies. At the ‘light touch’ end of the spectrum, interventions consist of screening and providing patients with information about or ‘referrals’ to community resources. Patients then pursue the resources on their own, with little or no follow-up from the social intervention staff. ‘High-touch’ programmes, by contrast, provide wraparound services with extensive, in-person support and follow-up in the community. I conducted ethnography of a high-touch programme embedded in the primary care clinics of two public hospitals (Fleming et al., 2017, 2019, 2021). The staff of these programs were organized into dyads of nurse and community health worker, who shared a caseload of about 50 patients. The dyads were part of a larger team that included social workers and a physician medical director.

7 In one of my field sites, health care providers made a point of referring to service recipients as ‘consumers’ rather than patients, which was intended to emphasize their agency and choice.

8 The question of whether these financial incentives are sufficient to drive the uptake of social intervention or if other mechanisms are involved remains an active area of research (Brewster et al., 2020).

9 The degree of penalties differs across contracts. Most hospitals, for instance, are subject to penalties for 30-day readmissions. In the public sector, metrics used in most value-based payment arrangements are set by the federal government and include measures of both health care quality and cost efficiency.

10 Hogle (2019) has characterized value-based care as an ‘accountability assemblage’ composed of ‘big data analytics, changing health information technology (HIT) infrastructures, novel cost accounting techniques, historical and political policy contexts, and intensified public health focus on social and behavioral influences on health as much as genomics’ (p. 558). I am arguing that this assemblage is being mobilized to reorient health systems toward new social intervention activities.