Screening and Interventions for Social Risk Factors: A Technical Brief to Support the U.S. Preventive Services Task Force

Abstract

Importance:

Evidence-based guidance is limited on how clinicians should screen for social risk factors and which interventions related to these risk factors improve health outcomes.

Objective:

To describe research on screening and interventions for social risk factors to inform U.S. Preventive Services Task Force considerations of the implications for its portfolio of recommendations.

Data Sources:

Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Sociological Abstracts, and Social Services Abstracts (through 2018); Social Interventions Research and Evaluation Network evidence library (January through May 2021); surveillance through May 21, 2021; interviews with 17 key informants.

Study Selection:

Individual-level and healthcare system-level interventions with link to healthcare system that addressed at least one of seven social risk domains: housing instability, food insecurity, transportation difficulties, utility needs, interpersonal safety, education, and financial strain.

Data Extraction and Synthesis:

One investigator abstracted data from studies and a second investigator evaluated data abstractions for completeness and accuracy; key informant interviews were recorded, transcribed, summarized, and integrated with evidence from the literature; narrative synthesis with supporting tables and figures.

Main Outcomes and Measures:

Validity of multidomain social risk screening tools; all outcomes reported for social risk-related interventions; challenges or unintended consequences of screening and interventions.

Results:

Many multidomain social risk screening tools have been developed, but they vary widely in their assessment of social risk and few have been validated. The technical brief identified 107 social risk intervention studies (n = 5997991). Of the interventions studied, 73 (68%; N = 143133) addressed multiple social risk domains. The most frequently addressed domains were food insecurity (68/107 studies,63.5%, n = 145657), financial strain (63/107 studies, 58.9%, n = 5896757), and housing instability (52/107 studies, 48.6%, n = 127497). Food insecurity, housing instability, and transportation difficulties were identified by key informants as the most important social risk factors to identify in healthcare. Thirty-nine studies (36%, n = 5854529) used an observational design with no comparator, and 19 studies (18%, n = 15205) were RCTs. Healthcare utilization measures were the most commonly reported outcomes in the 68 studies with a comparator (38 studies, 56%, n = 126637). The literature and key informants described many perceived or potential challenges to implementation of social risk screening and interventions in healthcare.

Conclusions and Relevance:

Many interventions to address food insecurity, financial strain, and housing instability have been studied, but more RCTs that report health outcomes from social risk screening and intervention are needed to guide widespread implementation in healthcare.

Social conditions and the structural forces that shape them underlie preventable disparities in many health outcomes. In 2018, there were an estimated 38.1 million people living in poverty in the U.S.,¹ and an estimated 1.42 million people experienced sheltered homelessness in 2016.² Identifying and addressing patients’ experiences of socioeconomic adversity is increasingly the focus of many national health system efforts.³ However, evidence-based guidance is limited on how clinicians should screen for social risk factors and whether healthcare-initiated interventions related to these risk factors improve health outcomes.

Social determinants of health are neither inherently positive nor negative, but instead encompass the wide range of social and economic conditions that shape health outcomes for individuals and communities.^4–6 The downstream manifestations of those forces are experienced by individuals as either social assets or social risk factors. To be consistent with prior work that has helped shape this emerging area of health services research, the term social risk factors is used in this article to refer to measurable and intervenable individual-level social and economic conditions that are shaped by broader determinants of health.⁶ Furthermore, since social risk factors captured by screening tools are not always reflective of patient priorities or perceived needs, the term social needs is used to more narrowly refer to instances when patients have indicated interest in assistance related to social risks.⁶

The U.S. Preventive Services Task Force (USPSTF) makes evidence-based recommendations for primary care preventive services. This technical brief aimed to describe the evidence base for social risk screening and interventions and present an overview of implementation challenges in health care to inform USPSTF considerations of the implications for its portfolio of recommendations. It was not intended to systematically review the effectiveness of social risk screening and interventions nor to serve as the basis for a USPSTF clinical practice recommendation.

Methods

Scope of Technical Brief

This technical brief addressed 5 guiding questions (GQs) as shown in Figure 1. Detailed methods, including search strategies, detailed inclusion criteria, and excluded studies, are publicly available in the full technical brief.⁷

Figure 1. Analytic Framework.

1. What are the available multidomain screening tools to identify social risk and what social risk domains do they identify? How valid are these tools? How does measurement of specific social risk domains vary by screening tool?
2. What social risk-related interventions have been evaluated? What are the characteristics of the studies that have evaluated these interventions and what outcomes do they report?
3. What are the effects of improvements in process outcomes, healthcare utilization outcomes, or social risk outcomes on physiologic and behavioral health outcomes?
4. What are the perceived or potential challenges to implementation of widespread screening and interventions for social risk factors within healthcare? What potential solutions have been proposed to address these challenges?
5. What are the challenges or unintended consequences of screening and interventions for social risk factors to patients and providers? What is the acceptability of screening for and intervening on social risk factors for patients and providers?

Evidence reviews for the US Preventive Services Task Force (USPSTF) use an analytic framework to visually display the questions that the review will address. The questions are depicted by linkages that relate interventions and outcomes. A dashed line indicates a health outcome that immediately follows an intermediate outcome. Additional information available in the USPSTF Procedure Manual.

Data Sources and Searches

For GQ1, a search for available screening tools was not conducted because a 2019 review was identified that addressed this question.⁸ The review included randomized and nonrandomized study designs describing development or empirical use of screening tools assessing 2 or more social risk domains in U.S. populations published since 2000.

For GQ2–5, the Cochrane Database of Systematic Reviews, Cochrane Central Register of Controlled Trials, Ovid MEDLINE, Sociological Abstracts, and Social Services Abstracts were searched from database inception to December 2018. Searches were supplemented by reviewing reference lists of recent reviews and primary studies and the evidence library on the Social Interventions Research and Evaluation Network (SIREN) website (https://sirenetwork.ucsf.edu/) through May 10, 2021. Active surveillance was conducted via targeted journal searches through May 21, 2021. Literature search results were managed using DistillerSR systematic literature review software (Evidence Partners, Ottawa, Canada). A search for conference abstracts and proceedings and other preliminary, unpublished study findings was also conducted.

Semi-structured telephone interviews were conducted with 17 key informants to better understand current clinical context and implementation challenges. Key informants were identified from the SIREN Research Advisory Committee; the National Academies of Sciences, Engineering, and Medicine Committee on Integrating Social Needs Care into the Delivery of Healthcare; and researchers currently conducting studies and actively publishing in the field. These experts represent primary care, policy, research, patient advocacy, social services, public health, federal agency, and payer perspectives, and their work addresses many social risk domains, disadvantaged populations, and healthcare and community settings. Many clinicians who directly provide patient care and health system representatives were recruited to obtain multiple perspectives on practice variations, issues with implementation, and current clinical context. Two sets of interview questions were used – one for researchers that focused on the evidence base and one for implementation experts that focused on implementation of social risk screening and interventions (eMethods).

Study Selection

The abstracts of 17,284 identified articles were reviewed against a priori eligibility criteria (Figure 2). Two investigators then independently evaluated the full text of 546 potentially relevant articles. Studies of patients of all ages conducted in the general population were included. Studies targeting persons with a specific disease were excluded because these studies are typically focused on management of the particular condition and are not applicable to other patients. However, studies that recruited patients with one or more unspecified chronic illness were included. Interventions were included if they addressed at least one of the target social risk domains: housing instability, food insecurity, transportation difficulties, utility needs, aspects of interpersonal safety that are not already addressed by USPSTF recommendations, education, and financial strain. The target social risk domains were aligned to those in the Centers for Medicare & Medicaid Services Accountable Health Communities Model because these are modifiable social risk domains for which there are primary care-referable interventions that are available to most patients.⁹ Interventions at the individual and health care system levels targeting a single or multiple social risk domains were included. Included studies had to have a link to the healthcare system. Randomized and nonrandomized controlled intervention studies; cohort, case-control observational, and pre-post studies; and case series were included for GQ2 and GQ3. For GQ4, all study designs were included except case reports. For GQ5, all study designs except case reports, editorials, and reviews were included. No studies were excluded based on outcomes reported.

Figure 2. Article Flow Diagram.

GQ indicates general question.

^aArticles reviewed for all GQs.

^b Reasons for exclusion: Relevance: Study aim not relevant. No link to healthcare: Study did not have a link to healthcare system. Setting: Study not conducted in a country with a very high Human Development Index (HDI). Intervention: Study did not contain an included intervention type. Population: Study only included patients with specific medical condition(s). Social risk domain: Study did not include at least one target social risk domain. Study design: Study did not use an included design. Abstract only: Full-text publication not available.

Data Extraction and Synthesis

For GQ1, the results of the 2019 systematic review on social risk screening tools⁸ were summarized, including the social risk domains addressed. Variation in the tools’ assessment of the social risk domains was summarized by examining the phrasing of questions across tools. The screening tools used in studies included for GQ2 with a screening component were recorded and summarized.

For GQ2–5, data abstraction forms were designed to gather pertinent information from each article meeting inclusion criteria, including participant, intervention, and study characteristics. One investigator abstracted information into the forms, and a second investigator evaluated data abstractions for completeness and accuracy. Disagreements were resolved by consensus. As stated above, the purpose of the technical brief was to describe the characteristics of the evidence base rather than to critically appraise and synthesize the effectiveness of available studies. As such, following technical brief methodology, the risk of study bias was not rated and the strength of the evidence was not graded. The evidence for each GQ was narratively synthesized, with supporting summary tables and figures to characterize the identified evidence. The types of outcomes reported in studies were abstracted. The results (eg, effect sizes) were not abstracted because the effectiveness of screening and interventions was outside the scope of the project. All interviews with Key Informants were recorded and transcribed, and responses were summarized and integrated with evidence from the literature for each GQ.

Results

Social Risk Screening Tools

General Question 1.

What are the available multidomain screening tools to identify social risk and what social risk domains do they identify? How valid are these tools? How does measurement of specific social risk domains vary by screening tool?

The 2019 systematic review⁸ identified available multidomain social risk screening tools, evaluated the degree to which gold standard methods were used in their development, and summarized the available psychometric and pragmatic evidence for the tools. Eighteen tools included in the review are intended for use in primary care settings and address at least 1 of the target social risk domains. The number of questions in these tools ranged from 7 to 118, and administration time ranged from 5 to 25 minutes. The most frequently included social risk domains in the tools were food insecurity, intimate partner violence, housing instability, financial strain, education, and social isolation.

Only 7 tools included reliability and validity testing and in subsequent empirical use, 71% of the tools had been modified from their original form, making it difficult to draw conclusions about their validity. The authors of the review concluded that there were currently no multidomain social risk screening tools with evidence that they can accurately identify social risk, detect changes in social risk, and measure intervention effects.

For the technical brief, the way in which the 18 tools assess target social risk domains was examined. Twelve tools frame 1 or more questions in terms of “concerns”, “worries”, “problems”, and/or “troubles” to detect patient-identified social needs; only 5 tools ask whether patients would like help with needs they have identified. Tools that address food insecurity generally inquire whether patients or families have enough food; 3 tools also ask about intake of fruits and vegetables or healthy food. Legal tools ask whether respondents are eligible for or have previously been denied Special Supplemental Nutrition Program for Women, Infants, and Children (WIC) or Supplemental Nutrition Assistance Program (SNAP) benefits. Questions regarding housing address current housing status, housing quality, and concerns about future homelessness. Five of the nine tools that assess transportation needs do so by asking whether respondents are able to attend medical appointments. Three tools ask about continuing education needs and three focus on education in the context of healthcare (eg, reading pharmacy instructions or medical pamphlets). Seven tools addressing financial strain ask about ability to cover basic necessities (food, housing, medical care, heat) or “make ends meet,” and two tools include items on income and work status.

Key informants reported using a variety of social risk screening tools, many of which were developed by their organization. They reported selecting screening tools because they were clinically validated, had a limited time burden, would result in nationally comparative data, met organizational needs, or incorporated the most important patient needs. Thirteen key informants named specific social risks they consider the most important to identify in healthcare. The most frequently cited social risks were food security (cited by 8 key informants) and housing and transportation (both cited by 6 key informants). Many noted that these social risks are important because they are the most actionable within the healthcare setting or the most critical to well-being. Key informants also recognized that patients should define the social needs that are most important to them and that the most important issues differ by community.

Information about what screening tools have been used in studies comes from the evidence included for GQ2. Forty-nine of the 107 studies included for GQ2 had a screening component, with one or more screening tools used.^10–42 The most frequently used screening tool was the 2-item Hunger Vital Sign tool^43,44 (N = 15), followed by the U.S. Department of Agriculture Household Food Security measure,⁴⁵ (N = 4) Health Leads,⁴⁶ (N = 4) and WE CARE,^11,38 (N = 2). The Homeless Screening Clinical Reminder,³⁵ the Children’s HealthWatch survey,⁴⁷ iScreen,⁴⁸ Cutt’s 3-item Housing Insecurity tool,⁴⁹ PRAPARE,⁵⁰ and the Legal Health Check-Up survey⁵¹ were each used in one study. Some of these tools address only a single domain so were not included in the 2019 review. Twenty-one studies used a study-developed screening tool, some of which were developed de novo while others were modifications of existing tools, supporting the finding in the 2019 review that the majority of tools were modified when used after development.

Social Risk Interventions

General Question 2.

What social risk-related interventions have been evaluated? What are the characteristics of the studies that have evaluated these interventions and what outcomes do they report?

One hundred seven studies (n = 5997991)^{10–42,52–125} met inclusion criteria (Figure 2), including 19 RCTs, 15 cohort studies, 34 pre-post studies, and 39 observational studies without a comparator. Participant, intervention, and study characteristics for included studies, categorized by social risk domain targeted, are presented in eTables 1 – 3 in the Supplement.

Participant Characteristics

Twenty-eight studies^{10,11,13–15,17–19,22,24,28,30,32,33,38,83,85,91,98,100,109,112,115–117,120,122,124} (n = 63523) enrolled only children and adolescents (and their caregivers), and 68 studies^{12,20,21,23,25–27,29,31,34–36,40,42,52–60,62,63,66–72,74–76,78–82,84,86–89,92–97,99,101–106,108,110,111,113,114,118,119,121,123,125} (n = 5928936) enrolled only adults, including older adults (ie, age 18 and older). Two studies^88,114 enrolled only adults age 65 years or older (n = 1943), 3 studies^61,64,77 enrolled participants of all ages (n = 1280), and 1 study¹⁰⁷ enrolled children/adolescents and older adults (n not reported). Twenty-five studies^{10,15,19,22,32,34,42,60,96,97,100,107–113,115–117,121,123–125} (n = 73998) recruited a general, nontargeted patient population; all other studies (n = 5923993) targeted patients with particular demographic, medical, or social risk characteristics. Participants were most frequently selected based on age (49 studies, n = 64821)^{11–14,16–18,23–25,27,28,30,31,33,36–39,56,58,59,62,63,65,66,68–73,76–78,80,82,84,86–90,93,94,101,102,114,120} and specific social risk(s) (eg, homeless, low income) (22 studies, n = 41372).

Intervention Characteristics

The majority of studies targeted patients, caregivers, or both (95 studies, n = 5990128),^{10–19,22–25,27–31,33–37,39,41,42,54,56,58–74,76–104,106–125} and 12 studies^{20,21,26,32,38,40,52,53,55,57,75,105} (n = 7963) targeted physicians or other clinicians. Sixty-eight percent of interventions (N = 73, n = 143133) addressed multiple social risk domains (range, 2–14; mode, 8).^{10–12,16,18,19,21–26,29,30,34,37–42,52,54,55,57–59,61–66,68,69,72–74,76,77,79–82,84–93,95,98–106,109,111,114,115,117–119,121,122} Many of the social risk domains addressed in multidomain interventions were non-target domains, such as healthcare and medication access/affordability, substance use, and employment. Looking at single domain interventions and the individual target domains included in multidomain interventions, 68 (63.5%, n = 145657) address food insecurity,^{10,11,13,14,16,17,19–28,30–34,36–38,40–42,52,53,55,57,58,63,72,74,75,77,78,80,81,85,89,93–95,99–103,105,107,109,111–125} 63 (58.9%, n = 5896757) address housing instability,^{10–12,16,18,19,21–26,29,30,35,37–42,52,55,57–59,61,64,68,72–74,76,77,79,81,84–90,92,93,95,98,99,101–106,109,111,114,115,117–119,121,122} 52 (48.6%, n = 127497) address financial strain,^{10,12,16,19,21,22,24–26,29,37,52,54,56–59,61–69,72,73,76,77,79,81–83,85,87,88,91,95,96,98,99,102,103,108,109,111,114,115,119,121,122} 35 (32.7%, n = 101409) address transportation needs,^{10,12,19,23,34,40,42,55,58,65,66,70–72,74,77,81,82,84,88,93,97,99,102,103,105,106,109–111,114,115,117,119,122} 27 (25.2%, n = 90495) address education,^{11,12,19,21,22,24,25,29,37–42,60,64,65,72,76,85,90,99,101,103,109,111,122} 22 (20.5%, n = 75851) address utility needs,^{10,11,15,16,19,22,24,26,40,42,62,72,74,101,103,105,109,111,115,117–119} and 8 (7.5%, n = 9650) address interpersonal violence.^{12,25,37,39,90,111,115,117}

Twenty-four studies^{23,29,63–66,68,69,72,73,76,77,81,82,84,87–89,93,95,104,106,114,118} (n = 59057) evaluated interventions that addressed 1 or more social risks but also included 1 or more other components related to medical management. In these studies, it is not clear whether outcomes are due to the effects of the social risk component(s) or these other elements. The three most frequently included nonsocial components were case management or care coordination, in-home healthcare, and health education.

Study Characteristics

The majority of studies were conducted in primary care (58 studies, n = 83119),^{11,13–25,27–30,32–34,36–38,40–42,52–57,59,61,62,66,67,69–71,73,77,80,81,83,85,94,96,100,102,108,109,111,114,117,121,123} followed by multiple settings (15 studies, n = 27650),^{10,64,72,79,84,91–93,98,101,106,120,122,124,125} emergency departments (10 studies, n = 4004),^{12,39,60,68,75,76,87,112,116,119} inpatient hospitals (8 studies, n = 9059),^{31,58,82,86,89,90,95,106,113} homes (6 studies, n = 6344),^{63,65,74,78,88,99} outpatient clinics (6 studies, n = 5792469)^{26,35,97,105,107,110}, and transitional housing (1 study, n = 11).¹⁰⁴

Thirty-six percent of studies (39 studies, n = 5854529)^{13,14,18,19,22,23,25,27,30–33,35,37,40,42,54,57,61,63,66,67,72,77,79,83,91,96,109,111–113,116–119,121,124,125} used an observational design with no comparator. Many of these were descriptions of feasibility testing with small cohorts. The most common study design with a comparator was pre-post (34 studies, n = 46707),^{15,20,21,24,26,34,39,41,52,53,55,56,59,60,68,69,74,75,80,81,84,85,92,95,97,99,104,106–108,110,120,122,123} followed by RCTs (19 studies, n = 15205),^{10,11,16,28,38,58,62,65,73,76,82,86–88,93,98,101,105,115} and cohort studies (15 studies, n = 81550),^{12,17,29,36,64,70,71,78,89,90,94,100,102,103,114}. Table 1 shows the number of studies addressing the social risk domains by study design. Fifteen of the 19 RCTs addressed housing instability (n = 12893),^{10,11,16,38,58,73,76,86–88,93,98,101,105,115} 12 addressed financial strain (n = 7677),^{10,16,58,62,65,73,76,82,87,88,98,115} 10 addressed food insecurity (n = 10317),^{10,11,16,28,38,58,93,101,105,115} 8 addressed transportation needs (n = 10888),^{10,58,65,82,88,93,105,115} 7 addressed utilities help (n = 8540),^{10,11,16,62,101,105,115} 5 addressed education (n = 2781),^{11,38,65,76,101} and 1 addressed interpersonal violence (n = 611).¹¹⁵

Table 1.

Social Risk Domains Addressed by Study Design, No. of studies (%) (No. of Participants)

Study Design	Food Insecurity	Housing Instability	Transportation Needs	Utility Needs	Interpersonal Violence	Education	Financial Strain
RCT	10/19 (52.6%) (n=10317)	15/19 (78.9%) (n=12893)	8/19 (42.1%) (n=10888)	7/19 (36.8%) (n=8540)	1/19 (5.3%) (n=611)	5/19 (26.3%) (n=2781)	12/19 (63.2%) (n=7677)
Cohort study	9/15 (60.0%) (n=49321)	8/15 (53.3%) (n=71206)	6/15 (40.0%) (n=41795)	1/15 (6.7%) (n=34225)	2/15 (13.3%) (n=571)	5/15 (33.3%) (n=65162)	6/15 (40.0%) (n=70869)
Pre-post	20/34 (58.8%) (n=11244)	19/34 (55.9%) (n=12051)	10/34 (29.4%) (n=23165)	4/34 (11.8%) (n=8188)	1/34 (2.9%) (n=466)	8/34 (23.5%) (n=3761)	13/34 (38.2%) (n=22244)
Observational without comparator	29/39 (74.4%) (n=74775)	21/39 (53.8%) (n=5800607)	11/39 (28.2%) (n=25561)	10/39 (25.6%) (n=24898)	4/39 (10.3%) (n=8002)	9/39 (23.1%) (n=18791)	20/39 (51.3%) (n=26519)

The outcomes reported in studies were grouped into 6 categories adapted from a recent review of social risks interventions:¹²⁶ process, social risk, physiologic and behavioral health, healthcare utilization, cost, and clinician outcomes. Some of the outcomes in the physiologic and behavioral health outcomes category (eg, changes in substance use or dietary intake) do not fit the standard USPSTF definition of a health outcome— outcomes that are experienced or felt by patients and affect patients’ length or quality of life. This is an important consideration to the USPSTF since evidence for an intervention’s effect on health outcomes is the basis for USPSTF preventive service recommendations.¹²⁷

In the 68 studies with a comparator, 38 studies (n = 126637)^{12,16,17,29,39,56,58,60,64,65,68–71,73,78,81,82,84–90,92–95,97,98,102–106,110,114} reported healthcare utilization outcomes (eg, emergency department visits and inpatient admissions), followed by physiologic and behavioral outcomes (eg, mental health status and changes in substance use) in 32 studies (n = 49593),^{10,12,16,17,29,34,36,41,56,58,59,62,64,65,68,73,76,80,85–87,89,90,92,93,98–101,115,120,122}. Twenty-seven studies (n = 42790)^{10–12,15,16,28,29,39,41,56,62,64,65,68,75,80,85,87,90,92,95,98,99,108,115,120,123} reported social risk outcomes (eg, resolution of food insecurity), 22 studies^{10–12,17,20,21,24,26,28,34,36,38,39,41,52,53,56,75,85,88,101,107} (n = 15292) reported process outcomes (eg, referrals or resources provided), 15 studies^{56,64,65,68,69,74,78,84,85,87,92,94,95,104,106} (n = 22985) reported cost outcomes (eg, return on investment), and 6 studies^{21,39,52,53,55,105} reported clinician outcomes (eg, confidence in social risks knowledge and screening). Six RCTs^{10,11,28,38,88,101} reported process outcomes (n = 4182), 9 RCTs^{10,11,16,28,62,65,87,98,115} reported social risk outcomes (n = 5639), 13 RCTs^{10,16,58,62,65,73,76,86,87,93,98,101,115} reported physiologic and behavioral health outcomes (n = 7626), 11 RCTs reported healthcare use outcomes (n = 10859),^{16,58,65,73,82,86–88,93,98,105} 2 RCTs^65,87 reported cost outcomes (n = 1791), and one RCT¹⁰⁵ reported clinician outcomes (n = 4917).

To investigate whether social risk interventions that focus on children and their families differ from those targeting adults, a comparison of the social risk domains addressed and outcomes reported for pediatric and adult studies was conducted and is presented in Table 2. Food insecurity and housing instability were the most frequently addressed domains in both pediatric and adult studies. Healthcare utilization outcomes were reported in 30/68 (44%) adult studies but only in 3/28 (11%) pediatric studies, while physiologic/behavioral health outcomes were reported in a similar percentage of adult and pediatric studies (21/68 (31%) and 8/28 (29%), respectively).

Table 2.

Social Risk Domains and Outcomes in Pediatric and Adult Studies, No. of Studies (%) (No. of Participants)

	Social Risk Domains
Population	Food Insecurity	Housing Instability	Transportation Needs	Utility Needs	Interpersonal Violence	Education	Financial Strain
Pediatric	23/28 (82.1%) (n=54407)	14/28 (50.0%) (n=25255)	6/28 (21.4%) (n=21717)	9/28 (32.1%) (n=27087)	2/28 (7.1%) (n=7021)	8/28 (28.6%) (n=16198)	11/28 (39.3%) (n=24564)
Adult	40/68 (58.8%) (n=89704)	40/68 (58.8%) (n=5867509)	27/68 (39.7%) (n=77418)	12/68 (17.6%) (n=48434)	3/68 (4.4%) (n=1650)	13/68 (19.1%) (n=71582)	34/68 (50.0%) (n=98059)
	Outcomes
Population	Process	Social Risk	Physiologic/Behavioral Health	Healthcare Utilization	Cost	Clinician
Pediatric	20/28 (71.4%) (n=58895)	12/28 (42.9%) (n=15505)	8/28 (28.6%) (n=8959)	3/28 (10.7%) (n=5251)	2/28 (7.1%) (n=164)	2/28 (7.1%) (n=7796)
Adult	26/68 (38.2%) (n=5807458)	22/68 (32.4%) (n=5808196)	21/68 (30.9%) (n=37641)	30/68 (44.1%) (n=117935)	14/68 (20.6%) (n=21781)	7/68 (10.3%) (n=7700)

Figure 1 shows the number of studies that addressed each target social risk domain and the type of outcomes reported in the 68 studies with a comparator. The largest number of studies addressed housing instability and financial strain with healthcare utilization and physiologic/behavioral health outcomes reported, followed by food insecurity with process and physiologic/behavioral health outcomes reported and transportation needs with healthcare utilization outcomes reported.

Effects of Improvements in Intermediate Outcomes on Health Outcomes

General Question 3.

What are the effects of improvements in process outcomes, healthcare utilization outcomes, or social risk outcomes on physiologic and behavioral health outcomes?

Since evidence for intervention effects on health outcomes is often lacking, the USPSTF considers the relationship between changes in intermediate outcomes and changes in health outcomes when evaluating the effectiveness of an intervention.¹²⁷ Although most studies that reported physiologic and behavioral health outcomes also reported other outcomes, only 4 studies reported on the effects of changes in intermediate outcomes (eg, process, social risk, or healthcare utilization outcomes) on physiologic and behavioral health outcomes. Two studies^56,67 (n = 981) found improvements in health outcomes (i.e., psychosocial aspects of quality of life and well-being scores) in patients whose intermediate outcomes improved after receiving welfare benefits advice in primary care but not in patients without improvement in intermediate outcomes. The other 2 studies^10,128 (n = 1957) — 1 examining provision of targeted information related to community, hospital, or government resources addressing social risks and 1 on supportive housing—found no associations between intermediate and health outcomes.

Challenges of Social Risk Screening and Interventions

General Question 4.

What are the perceived or potential challenges to implementation of widespread screening and interventions for social risk factors within healthcare? What potential solutions have been proposed to address these challenges?

Information on perceived or potential challenges to social risk screening and interventions, and proposed solutions to these challenges, was gathered from key informants and a scan of reviews, case studies, other descriptive research, and opinion articles identified in the literature searches. Table 3 shows the most commonly cited patient-, clinician-, health system-, and community-level factors that may present challenges to implementation of social risk screening and interventions in healthcare settings, and proposed strategies to overcome these challenges.

Table 3.

Most Commonly Cited Potential Implementation Challenges and Solutions^*

Level	Screening		Intervention
	Perceived Challenge	Proposed Solutions	Perceived Challenge	Proposed Solutions
Patient	Stigma and privacy concerns149,154	Use of patient-centered care models151 Developing trusting relationship with patients33 Identification of patient strengths and assets when screening for social risk factors155	Logistical barriers to referral follow through (eg, transportation issues)143,156	Explore alternative delivery models, such as co-located services (eg, food pantries or WIC services offered in healthcare setting)157
Clinician	Concern about lack of referral resources158–161	Increasing clinician incentives to screen151 Facilitating clinician access to referral and support services157 Partnering with organizations that maintain referral lists162,163 Use of SSRL vendors (key informant suggestion) Frequent updating of resource lists or databases162	Lack of clinician enthusiasm to sustain intervention after conclusion of research-funded interventions159	Sharing outcomes data with clinicians151 Identification of clinical champions151
Health System	Concerns about social risk data collection and management by healthcare organization and partnering organizations162,164,165	Developing digital infrastructure that is interoperable between healthcare and social care organizations151,165,166 Integrating social risk data into EMR systems151,165,166 Partnering with data analytic vendors (key informant suggestion)	Sustainability of funding164,167,168	Financing integration of healthcare and social care151 Payment reform (eg, expanding Medicare coverage for social needs services)165 Exploring novel funding opportunities (eg, public-private partnerships) (key informant suggestion)
Community	None cited	----	Limited capacity of social resources155	Supporting community partners with financial and infrastructure needs151,166 Warm handoffs† to community partners to ensure resources are available for referrals (key informant suggestion)

^*Cited in the literature and/or by key informants

^†In-person with patient

Abbreviations: EMR = electronic medical record; SSRL = social service resource locator; WIC = women, infants, children

Acceptability and Unintended Consequences of Social Risk Screening and Interventions

General Question 5.

What are the challenges or unintended consequences of screening and interventions for social risk factors to patients and clinicians? What is the acceptability of screening for and intervening on social risk factors for patients and clinicians?

Fifty-two studies^{12,14,21,24,25,32,33,37,38,52,55,58,60,62,68,82,83,85,91,96,99–101,105,108,111–114,123,129–150} provided data on patient- or clinician-reported satisfaction or challenges after implementation of social risk screening or interventions (eTable 4 in the Supplement).

Patients

Thirty-one articles^{12,37,52,58,60,62,82,100,101,112,114,123,129,131–134,136,137,139–141,143,24,55,85,96,108,145,148} included positive patient reports of satisfaction with and acceptability of screening and interventions and referred to improvements in the patient-clinician relationship and high comfort levels. Eleven articles^{33,83,85,99,105,111,113,133,135,139,146} reported on challenges or unintended consequences of screening or intervention for patients, including discomfort (eg, shame about social risks) and confidentiality issues (eg, fear of legal repercussions such as being reported for child maltreatment due to food insecurity). One study found paradoxical effects of improvement in social risks; families who participated in SNAP and increased their earned income had their SNAP benefits reduced or cut off and subsequently faced economic strain that diminished their ability to pay for housing, utilities, healthcare, or food.¹³⁵ Two articles reported that there were no adverse effects from the intervention.^99,105

Clinicians

Seventeen of the eighteen articles^{14,32,38,52,55,83,91,96,123,137–139,142,144,147–150} that reported on clinician satisfaction with screening and interventions were positive, with clinicians stating that screening was not overly time-consuming and led to improvements in the patient-clinician relationship, patient care, and clinician knowledge and competence. The one negative report was related to difficulty in incorporating the intervention into clinician schedules.⁸³ Fifteen articles^{14,21,25,32,33,52,68,105,130,139,142,144,147,149,150} reported on challenges or unintended consequences of social risk screening or interventions for clinicians, including lack of time to conduct screening or follow up on positive results and inability to track the success of referrals.

Discussion

This technical brief identified and described the evidence on social risk screening and interventions. Many multidomain social risk screening tools are available, but few have undergone reliability and validity testing. Food insecurity, housing instability, and transportation difficulties were identified by key informants as the most important social risk factors to identify in healthcare, and these are three of the most frequently addressed social risk domains in the 107 intervention studies identified, along with financial security. Thirty-six percent of studies used an observational design with no comparator, and only 18% of studies were RCTs. Healthcare utilization measures were the most commonly reported outcomes in studies with a comparator. The literature and key informants described many challenges to implementation of social risk screening and interventions in healthcare.

In keeping with the USPSTF focus on recommendations for primary care clinicians about preventive services for asymptomatic people, this technical brief focused on population-based screening in primary care to detect unrecognized social risk factors and interventions to address them. Some experts have argued that screening should only be done when there is the capacity to address identified social risks. Though there is ongoing debate about the merits of screening without social risk-targeted interventions,¹⁵¹ it is nonetheless relevant to note that few existing screening tools assess patients’ interest in assistance for identified social needs.¹⁵²

This technical brief was prepared to inform USPSTF efforts to incorporate social risks into its recommendation process. The USPSTF considers services that are provided in or referable from primary care. While screening for social risk factors can be done in primary care clinical settings, many subsequent activities to intervene on social risks involve a referral from the healthcare team to a non-healthcare setting, such as public health, social service, and community-based organizations. This requires effective partnerships with these resources, adding a layer of complexity to implementation of social care in clinical settings. This technical brief identified many perceived or potential challenges to the implementation of social risk screening and intervention programs in healthcare. However, actual unintended consequences from social risk screening and interventions were rare in the studies that reported these outcomes. More data on the challenges encountered during implementation of social risk screening and interventions in healthcare settings and ways that these challenges have been addressed successfully would clarify what barriers and solutions need to be considered before scaling implementation efforts.

Limitations

This technical brief has several limitations. First, searches and inclusion criteria were limited to studies with the most relevance to the USPSTF scope and purpose. As such, studies in the general population were focused on and studies conducted in patients with a specific disease were excluded. Social risk screening and interventions may have different effects in patients with specific chronic conditions requiring complex management, such as diabetes. Second, studies conducted in countries that are not rated “very high” on the HDI were also excluded, which may have left out a considerable amount of research. Third, other limitations stem from the methods used given the focus of the technical brief on identifying and describing existing research rather than systematically reviewing the effectiveness of screening and interventions for social risk factors. Critical appraisal was not conducted, and some of the included studies may be of poor quality and would not meet criteria for a USPSTF review and recommendation. Outcomes data were not abstracted and results were not evaluated to determine the effect of interventions on outcomes.

Conclusions

Many interventions to address food insecurity, financial strain, and housing instability have been studied, but more RCTs that report health outcomes from social risk screening and intervention are needed to guide widespread implementation in healthcare.

Figure 3.

Number of Studies by Social Risk Domain and Outcome Category (N = 68)