Prevalence of Screening for Food Insecurity, Housing Instability, Utility Needs, Transportation Needs, and Interpersonal Violence by US Physician Practices and Hospitals

IMPORTANCE Social needs, including food, housing, utilities, transportation, and experience with interpersonal violence, are linked to health outcomes. Identifying patients with unmet social needs is a necessary first step to addressing these needs, yet little is known about the prevalence of screening. OBJECTIVE To characterize screening for social needs by physician practices and hospitals. DESIGN, SETTING, AND PARTICIPANTS Cross-sectional survey analyses of responses by physician practices and hospitals to the 2017-2018 National Survey of Healthcare Organizations and Systems. Responses were collected from survey participants from June 16, 2017, to August 17, 2018. EXPOSURES Organizational characteristics, including participation in delivery and payment reform. MAIN OUTCOMES AND MEASURES Self-report of screening patients for food insecurity, housing instability, utility needs, transportation needs, and experience with interpersonal violence. CONCLUSIONS AND RELEVANCE This study’s findings suggest that few US physician practices and hospitals screen patients for all 5 key social needs associated with health outcomes. Practices that serve disadvantaged patients report higher screening rates. The role of physicians and hospitals in meeting patients’ social needs is likely to increase as more take on accountability for cost under payment reform. Physicians and hospitals may need additional resources to screen for or address patients’ social needs.


Introduction
High-quality, coordinated medical care alone cannot ensure that patients achieve optimal health outcomes. 1 Up to 90% of health outcomes are a result of social, behavioral, and economic factors. 2 The association of patients' social needs, such as food insecurity, housing instability, utility needs, transportation needs, and experience with interpersonal violence, with health outcomes and costs is increasingly recognized by the medical community, and an increasing amount of evidence documents that physician-and hospital-led interventions addressing patients' social needs can produce improved health outcomes and less costly medical care. [3][4][5][6] With the goal of improving integration of social needs-informed care into medical care, the American Academy of Family Physicians, the National Association of Community Health Centers, and the American Academy of Pediatrics have launched campaigns to encourage physicians to screen patients for social needs. [7][8][9] Historically, systematically identifying and addressing patients' social needs has not been part of medical practice. 1 Although physicians and hospitals may recognize the association of social needs with patient outcomes, they may be reluctant to assume responsibility for social needs given the complexity of addressing these needs coupled with increasing clinical demands. 10,11 Despite these challenges, the movement toward value-based care has accelerated the focus of physicians and hospitals on population and whole-person health by incentivizing physicians and hospitals to address needs that may be outside of traditional clinical care yet associated with health spending. 3,12 Furthermore, state and federal policymakers as well as private payers are designing programs aimed at integrating social services into clinical care. For example, the Centers for Medicare & Medicaid Services (CMS) created the Accountable Health Community model, and multiple states have recently established waivers that allow Medicaid dollars to pay for services that support patients' social needs. [13][14][15] At the same time, as part of their campaigns to encourage physicians and hospitals to screen patients for social needs, the American Academy of Family Physicians, 7 National Association of Community Health Centers, 9 and the American Academy of Pediatrics 8 have each developed tools aimed at helping physicians and hospitals identify patients with social needs and then offer them support to avail themselves of community resources. Despite support by physician and hospital groups coupled with delivery reform from payers, little is known about the extent to which these initiatives have diffused into care delivery.
Timely, national data are needed to better understand the extent to which physicians and hospitals are embracing recommendations to screen patients for social needs. Previous research has largely focused on specific interventions or in-depth assessments of how some physicians and hospitals address social needs. 3,16,17 Although many health care organization leaders have voiced their commitment to addressing social needs, no national data exist on how often and which types of physicians and hospitals screen patients for social needs. In this study, we used new, nationally representative survey data to assess the prevalence of screening among physician practices and hospitals for social needs, specifically housing instability, food insecurity, utility needs, transportation needs, and the experience with interpersonal violence. We examined how screening efforts vary by organizational characteristics and capabilities, and we identified major barriers to adopting care delivery innovations reported by physicians and hospitals. A better understanding of the current landscape of social needs screening and of the nature of organizations who screen will help stakeholders determine physician and hospital readiness to engage, implement, and scale programs aimed at addressing patients' social needs.

Methods
This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline. 18 We used novel data from the National Survey of Healthcare Organizations and Systems (NSHOS), 19 supplemented by additional data from the OneKey database (IQVIA Inc), 20 the American Hospital Association's (AHA) Annual Survey, 21 and the US Census 22 to characterize US physician practices and hospitals that screen patients for social needs. The NSHOS, which actively accepted survey responses from June 16, 2017, to August 17, 2018, is a suite of nationally representative surveys, including a survey of physician practices with 3 or more primary care physicians and a survey of nonspecialty, acute care hospitals. The NSHOS collected data on the structure, ownership, leadership, and care delivery capabilities of these organizations. 19 This study was approved Dartmouth College's institutional review board, which waived informed consent for NSHOS because risk to respondents was low and the survey asked them to respond in their professional capacity.

Sampling and Data Collection
The NSHOS used the OneKey database, 20 produced by IQVIA, to identify and sample physician practices and hospitals. OneKey relies on proprietary data collection efforts, the American Medical Association's Physician Masterfile, and publicly available sources. OneKey provides information on the relationships among physicians, practices, hospitals, and health care systems. The NSHOS used a stratified-cluster sampling design that drew samples of physician practices and hospitals with different ownership and composition structures, including samples of both system-owned and independent physician practices and hospitals. The NSHOS includes a single response from each of 2190 physician practices and 739 acute care or critical access hospitals.
In the physician practice survey, the NSHOS surveyed practices with at least 3 primary care physicians, with primary care defined as family medicine, geriatrics, internal medicine, or preventive medicine specialties. Practices that also included specialty physicians were part of the sample. The target respondent was a medical director, physician, or practice manager. The NSHOS sampled 4976 physician practices and obtained a response from 2333 practices, for a response rate of 46.9%. We removed 143 responses because of ineligibility, which left 2190 responses to analyze.
In the NSHOS hospital survey, hospitals included short-term acute care and critical access hospitals; specialty hospitals were excluded. The NSHOS sampled 1628 hospitals and obtained a response from 757 hospitals, for a response rate of 46.5%. We removed 18 responses because of ineligibility, which left 739 responses to analyze. Target hospital respondents were leaders, such as the chief medical officer, chief executive officer, or other C-suite leaders.

Main Outcome Measures
To assess screening for social needs among physician practices and hospitals, the NSHOS asked "Does your practice have a system in place to screen patients for food insecurity (yes/no), housing instability (yes/no), utility needs (yes/no), transportation needs (yes/no), or interpersonal violence (yes/no)?" Hospitals were asked a similar question. We selected these 5 domains because these are part of the CMS's Accountable Health Communities model. We characterized physician practices and hospitals that report screening patients for all 5 social needs.
Physician practices were characterized using data from the NSHOS, OneKey, and the US Census.
Characteristics of practices included practice ownership, physician composition, type of practice, share of Medicaid revenue in the practice (by tertile), region, rural (those with Յ49 999 residents) or metropolitan area, and participation in payment and delivery reform initiatives, including bundled payments, primary care improvement models (eg, medical homes, comprehensive primary care), and accountable care organizations (ACOs).
We characterized hospitals using NSHOS, OneKey, the AHA Annual Survey, and US Census data. [19][20][21][22][23] Measures from the AHA included hospital ownership, share of inpatient admissions that were covered by Medicaid, and participation in an ACO contract. The NSHOS measures included academic vs nonacademic hospital, critical access status, region, and revenue received from episodebased payments. We used US Census data to obtain the area-level share of residents whose income was below the federal poverty level, which we divided into tertiles, and metropolitan vs rural areas.
We compared barriers to care delivery innovations that physician practices and hospitals reported by screening status. Respondents reported whether potential barriers were not a barrier, were a minor barrier, or were a major barrier to their use of care-delivery innovations. Innovations were not specific to screening for social needs. Barriers focused on processes to identify or disseminate innovations and lack of time, financial resources, knowledge, or incentives.

Statistical Analysis
We characterized physician practices and hospitals screening for all 5 social needs by using means and proportions with 95% CIs. We used χ 2 tests to calculate significance. Significance was set at 2-sided P = .05. As a sensitivity analysis, we compared physician practices with hospitals that screened for food insecurity, housing instability, and utility needs; that screened for food insecurity, housing instability, utility needs, and transportation needs; and that did not screen for any needs. We excluded survey respondents with missing data for a given variable.
We used the OneKey database to weight responses from physician practices to all practices with 3 or more primary care physicians and from hospitals to all acute care, nonspecialty hospitals that met our sample frame definition. Data were weighted to adjust for nonresponse bias. The OneKey data allowed us to adjust responses based on practices' and hospitals' ownership, structure, and composition. Statistical calculations were performed using Stata Statistical Software, Release 13 (StataCorp).

Characteristics of Hospitals Screening for Social Needs
We observed few significant differences among hospitals based on whether they screen for all needs:   We did not observe any differences in characteristics between organizations screening for some needs vs those screening for all 5 needs. When comparing organizations that screened for 0 needs, we observed some changes in significance. For physician practices, those that were not in an academic medical center were more likely to screen for 0 needs (34.5%; 95% CI, 31.0%-38.0% vs 27.6%, 95% CI, 23.3%-31.9%; P = .02); there were no significant differences by region, and there were no differences for commercial ACO contracts. For hospitals, nonacademic medical centers were shared savings (11.0%; 95% CI, 6.7%-15.3% vs 4.4%; 95% CI, 2.4%-6.5%; P = .001), and hospitals without revenue from episode-based payments (10.9%; 95% CI, 6.7%-15.1% vs 4.4%; 95% CI, 1.9%-7.0%; P = .008) were all more likely to report screening for 0 needs.

Barriers to Care Delivery Innovations
When asked about barriers to general dissemination of innovations, practices and hospitals that did not screen for social needs were significantly more likely to report a lack of financial resources, time, and incentives as major barriers ( Table 3). For example, 51.0% (95% CI, 47.7%-54.4%) of practices that do not screen for social needs reported the lack of financial resources as a major barrier, whereas 37.5% (95% CI, 29.7%-45.9%) of practices that screen for all 5 needs reported this barrier (P = .001).
Similarly, 60.0% (95% CI, 52.9%-66.8%) of hospitals that do not screen for social needs reported the lack of financial resources as a major barrier as opposed to 36.2% (95% CI, 28.6%-44.5%) of hospitals that screen for all 5 social needs (P < .001).

Discussion
The link between social needs and health care outcomes has long been understood as important to achieving optimal health, yet this link has only recently been thrust into the spotlight as physicians and hospitals begin to take on accountability for population health and spending outcomes. 12,[26][27][28][29] These new data suggest that most US physician practices and hospitals are screening patients for at least 1 social need (most often, experience with interpersonal violence), and most are not screening patients for the 5 social needs that CMS has prioritized: food insecurity, housing instability, utility needs, transportation needs, and experience with interpersonal violence. 30 Attention to the association of social needs with medical outcomes is widespread, but dissemination of care delivery innovations in physician practices and hospitals is notoriously slow 31 and the barriers to linking medical and social care are high because funding streams and supportive organizations tend to be siloed. 32 At the same time, despite the increasing recognition, we do not yet know if screening for patients' social needs will translate into better care outcomes or lower spending because current evidence is limited. 33 Given the lack of a clear evidence base coupled with steep barriers to addressing social needs within clinical care, it may be that physician practices and hospitals that are screening patients for social needs are early champions of screening.
Rates of screening vary by organizational characteristics, with hospitals and practices that serve more disadvantaged patients reporting higher screening rates. Nearly one-third of federally qualified health centers screen patients for all 5 needs. Practices with exposure to delivery and payment reform, including primary care improvement models, bundled payments, and commercial ACO contracts, are more likely to screen. Among hospitals, academic medical centers are more likely than nonacademic medical centers to screen patients for all social needs.
Hospitals and physician practices have different capabilities, needs, and motivations, which may influence the differential uptake of screening by hospitals. Physician practices may be motivated to screen patients for social needs to help provide more coordinated, comprehensive care in lowercost settings (particularly if they participate in payment reform models and believe that addressing social needs will reduce health care spending), but they may lack the financial or staffing resources to routinely screen in the course of clinical care. 32 Hospitals may have more resources, including staffing, financial, and technological, as well as more processes, protocols, and standardization in care delivery, which may explain higher screening rates among hospitals compared with physician practices.
Hospitals may also be more likely to screen patients for transportation and housing needs as part of their discharge processes because they are subject to federal regulations on patient safety as part of their certification from CMS. 34 For example, given the patient's medical condition, Medicare requires that hospitals assess whether a patient has adequate transportation from the hospital as well as a safe place to be discharged to. 35 High-profile cases, such as the Maryland woman discharged from the emergency department on a street corner in the cold wearing a hospital gown, have garnered outcry from the public and the media. 36 In response to such cases, some states have put in place stricter discharge planning requirements, especially in the emergency department given that patients with greater social needs and barriers to care are more likely to use the emergency department. 37 To implement screening protocols and begin addressing patients' needs under a fee-for-service model, physicians and hospitals will need financial support. Payers could allow physicians and hospitals to bill for evidence-based programs, such as FoodRx, that have been shown effective at addressing needs and improving outcomes. 4 The CMS could consider expanding care management billing to include managing care for patients who are both at risk or have clinically complex conditions in addition to social needs.

Limitations
Our results should be interpreted in light of several limitations. This research is by nature descriptive, and our estimates are based on self-reported survey data from physician practices and hospitals.
Senior leaders in hospitals and physician practices, who completed the survey, may not always know whether their organization screens patients for social needs. Furthermore, although our survey asks whether the organization has a "system in place" to screen patients for social needs, we do not know how systematically the organization is screening patients (eg, all patients vs those with specific payers vs those who have clinically complex conditions). Nor do we know how health care organizations are using patients' screening results such as through offering referrals to community resources. Although weighted to represent the national landscape of practices and hospitals in our sample, there could be social desirability bias in our estimates, or leaders may not know about informal screening that takes place. Nevertheless, there is a paucity of systematic evidence regarding screening for and addressing social needs in mainstream clinical practice; we believe this study begins to fill this gap. 45 Given the current focus on social needs from state and federal policymakers, payers, and physicians and hospitals, it seems likely that pressure on physicians and hospitals to identify and begin addressing patients' social needs will continue. States are increasingly creating incentives for medical physicians, hospitals, and managed care organizations to integrate and address patients' social needs. 29 For example, Oregon's Coordinated Care Organizations program will require participants to devote part of their budget to addressing social needs in 2020. 46 The CMS has invested heavily in supporting social needs via their Accountable Health Communities project, which evaluates screening and referrals for social needs. 13 Private payers are also active in this area: Blue Cross Blue Shield has food delivery programs, and UnitedHealthcare is piloting a centralized hub model for addressing social needs in key markets. 47,48 Understanding which types of physician practices and hospitals are already screening patients for social needs can help policymakers know where to target their efforts. For example, nonacademic hospitals and physician practices that are not Federally Qualified Health Centers may need further incentives to screen patients.