As discussed in Chapters 2 and 3, the health consequences of social isolation and loneliness are significant across the age spectrum and worsen as people age. Specifically, as discussed in Chapter 2, data from 148 studies with more than 300,000 participants showed that being socially connected was associated with 50 percent increased odds of survival (Holt-Lunstad et al., 2010). Additionally, as discussed in Chapter 3, a meta-analysis of 23 studies using 16 longitudinal datasets found that poor social relationships (e.g., social isolation, loneliness) increased the risk of developing coronary heart disease and stroke, independently of traditional cardiovascular disease risk factors (Valtorta et al., 2016a). Given the significant evidence for the negative impacts of social isolation and loneliness on the health of older adults, this chapter explores the role of the health care delivery system in identifying and addressing social isolation and loneliness.
Recognition of the role of the health care system in addressing social isolation and loneliness is not new. In 1985 Jones et al. drew attention to the role of physicians in addressing loneliness, noting:
General practitioners have unique opportunities to reduce the suffering caused by loneliness. The lonely elderly consult their doctors more often (because of their higher degree of physical disability), and so general practitioners are the professional group most likely to come into contact with the lonely person. By listening to lonely patients and gaining their confidence, the doctor can refer them to appropriate bodies such as the social services, voluntary organizations, neighbourhood schemes and local churches. (Jones et al., 1985, p. 139)
In 2018, Helen Stokes-Lampard, the chair of the Royal College of General Practitioners in the United Kingdom, expressed the following to this committee:
I know as a clinician and as a physician that I cannot solve the problems of my patients’ lives with respect to their social challenges, but if I identify them, if I recognize that those problems are impacting adversely on health, then it is my responsibility to call that out, to signpost people to help.
Given the complexity of the terminology used in relation to social isolation and loneliness, a reminder of key definitions is provided in Box 7-1.
Older adults are particularly high-volume and high-frequency users of the health care system, especially as compared with younger groups. The 2008 Institute of Medicine (IOM) report Retooling for an Aging America: Building the Health Care Workforce found that adults over age 65 use a disproportionate number of health care services, stating
Although older adults make up only about 12 percent of the U.S. population, they account for approximately 26 percent of all physician office visits, 47 percent of all hospital outpatient visits with nurse practitioners, 35 percent of all
hospital stays, 34 percent of all prescriptions, 38 percent of all emergency medical service responses, and 90 percent of all nursing-home use. (IOM, 2008, pp. 3–4)
Research on the impact of social isolation and loneliness on health care utilization and access is limited, and it has had mixed results. While the available research indicates that social isolation and loneliness among adults lead to a heightened focus on utilization, few studies have examined access issues for older adults who are isolated or lonely. Extensive research has been conducted in countries outside the United States, including Australia, Ireland, New Zealand, and the United Kingdom. These studies can be used as a foundation to inform what we know about how social isolation and loneliness affect access and utilization. However, the U.S. health care delivery system is very different from those of other countries, so it is necessary to conduct research within the United States in order to elucidate issues of particular concern to the U.S. health care system. As noted previously in this report, the existing evidence base shows that a variety of indicators are used to evaluate social isolation and loneliness, including the strength of an individual’s relationships and social networks, a person’s living arrangements, and a person’s psychosocial functioning. Similarly, health care utilization is often determined differently in different studies, with readmission rates, length of hospital stay, number of hospitalizations, the use of outpatient services, and primary care visits being examined in various studies. Furthermore, some types of increases in utilization are desirable (e.g., preventive care) while other types of increases in utilization (e.g., hospital readmissions) are not desirable.
Evidence of Increased Utilization
Social relationships are one of many factors that affect health care utilization. Social isolation and loneliness have been found to be associated with an increased rate of hospital readmissions, perhaps as a result of the individual having a smaller social support network to contact when health care issues emerge (Hawker and Romero-Ortuno, 2016; Valtorta et al., 2018b). A similar finding was noted in a study of older American veterans enrolled in a psychogeriatric program (Mistry et al., 2001). However, other evidence suggests a higher number of previous hospitalizations among those with larger family networks (Ha et al., 2019). Social isolation and loneliness have also been associated with increased hospitalizations among older adults and older veterans (Gerst-Emerson and Jayawardhana, 2015; Greysen et al., 2013; Jakobsson et al., 2011), and social isolation has been linked to an increase in Medicare spending, primarily on costs associated with inpatient care and skilled nursing facility care (Flowers et al., 2017; Shaw et al., 2017). However, these spending patterns changed when adjusted for socioeconomic and health status. Despite the increased use of health care, these individuals experiencing social isolation and loneliness are more vulnerable to adverse outcomes and are a greater risk of death (Shaw et al., 2017).
Older adults who have been identified as being at risk for social isolation and loneliness experience longer length of stays when hospitalized (Hawker and Romero-Ortuno, 2016; Mitchinson et al., 2008; Valtorta et al., 2018b), which may be due to having smaller social networks and less social support to provide post-hospitalization care (Mitchinson et al., 2008; Valtorta et al., 2018b). Chronic loneliness is also positively associated with increased physician visits, with the physician–patient relationship providing both social support and medical treatment (Gerst-Emerson and Jayawardhana, 2015). In behavioral health settings, adults (not limited to older adults) living with family make greater use of rehabilitative and social care services, including day care, rehabilitation, socialization, and work-related activities; by contrast, those living alone have a higher number of home visits (Donisi et al., 2013). In one study of veterans enrolled in a psychogeriatric program, the veterans with higher social connections reported increased access to services (Mistry et al., 2001). Living with others and having a stronger social system may serve as a motivator to participate in health care services.
Evidence of No Impact on Utilization
Some studies have found that having fewer social supports has limited to no impact on health care utilization by older adults. In one study, older patients with weaker social relationships did not place greater demands on ambulatory care (as defined by physician visits and community- or home-based services) (Valtorta et al., 2018b). In addition, social isolation (as measured by living alone) was not a predictor of potentially preventable readmissions to hospitals (LaWall et al., 2019). Flowers and colleagues (2017, p. 5) found “no difference in outpatient use or spending for socially isolated Medicare beneficiaries.” Living alone may actually provide some protective health factors, and having such a living arrangement may be an indicator of a person’s level of independence or personal preferences.
Evidence of Decreased Utilization
There is limited evidence indicating a decrease in health care utilization by those older adults who are socially isolated and lonely. In particular, there is some evidence for a lower use of preventive health services (an undesirable decrease in usage), including mammograms, dental visits, immunizations, colonoscopy, general practitioner visits, and an exercise program for joint pain (Vozikaki et al., 2017). Loneliness has been linked to lower Medicare spending when adjusted for health status (Shaw et al., 2017). However, these examples of decreased usage could be a reflection of fewer supports and resources being available to enable such individuals to access outpatient and preventative services.
Concerning prevention, a connection exists between an individual’s level of physical activity and his or her use of health care. Engagement in a physical
activity program was found to be associated with fewer emergency room visits and hospital visits (Jacobs et al., 2013). In that study, participants who became more active were more likely to be male with higher self-reported health and functional independence and reduced rates of loneliness. Engagement in physical activity programs, while beneficial, may be more challenging for older adults who have chronic health conditions and more functional impairments and who are at risk for loneliness. (See Chapter 9 for more on physical activity programs as an intervention for social isolation and loneliness.)
Primary Care and Utilization
As discussed above, social isolation and loneliness affect the quantity and type of health care services used by older adults. Among community-dwelling older adults in the United States aged 60 and older, one study found that chronic loneliness (defined as being lonely at two points of time over 4 years) was a predictor of an increased number of both physician visits and hospital visits; the correlation was independent of sociodemographic variables, subjective and objective health measures, depressive symptoms, insurance status, and financial situation (Gerst-Emerson and Jayawardhana, 2015). Reporting feelings of chronic loneliness and having higher rates of health care utilization were associated with a variety of factors, including depressive symptoms, being married, having difficulty with activities of daily living (e.g., bathing, toileting, dressing), having a higher number of chronic conditions, and having at least a high school or GED-level of education (Gerst-Emerson and Jayawardhana, 2015).
These examples suggest that individuals living with social isolation or loneliness are more likely to use outpatient, emergency department, and inpatient services. However, there is no evidence that these individuals will isolate themselves from the benefits of primary care, and, in fact, loneliness is associated with increased visits to the physician’s office (Gerst-Emerson and Jayawardhana, 2015). Comprehensive and advanced primary care settings serving older adults are well suited to the task of caring for individuals living with social isolation and loneliness and could ultimately help link these individuals to effective interventions.
For example, a collaborative of practice-based networks in Colorado and Virginia assessed the prevalence of loneliness and associated characteristics and behaviors through a survey of registered adult patients presenting for routine primary care (Mullen et al., 2019). The prevalence of loneliness for individuals 65 years and older screened during the study period was 11 percent. As had been found in other studies, loneliness was associated with an increase in health care utilization across outpatient, emergency department, and inpatient settings and also with poorer health status. The study found no evidence that loneliness was associated with individuals isolating themselves from primary care. The authors concluded that “the primary care setting has the potential to identify solutions and implement interventions” (Mullen et al., 2019, p. 113).
Long-Term Care Settings and Utilization
The literature on utilization related to social isolation and loneliness among older adults in long-term care (LTC) settings is scant. One study of data from the National Study of Caregiving and the National Health and Aging Trends Study demonstrated that a sense of community engagement was a significant predictor of likelihood of older adults remaining in the community (Moon et al., 2018). Flowers and colleagues (2017, p. 5) found that “socially isolated individuals were 29 percent more likely to use [skilled nursing facility (SNF)] care and their monthly SNF costs were $75 higher on average.” The authors suggested one explanation, being that these individuals may lack adequate support following a hospital stay, and therefore require higher use of SNFs for rehabilitation. Shaw and colleagues (2017) similarly found that social isolated individuals used more SNF care. They noted that “future study of social isolation in managed care and nursing home populations is warranted” (p. 13).
Data from other countries also suggest that social connection affects utilization in LTC settings. Godin and colleagues (2019) examined “the association between social vulnerability and the odds of [LTC] placement within 30 days of discharge following admission to an acute care facility” (p. 1) among patients with acute respiratory illness admitted to hospitals in Canada. The authors created a social vulnerability index, which included attention to social support, living situation, and social engagement, among other factors. They found that
at younger ages (e.g., 70 years), social vulnerability was associated with lower odds of LTC placement for those who were the frailest, while at older ages (e.g., 90 years), social vulnerability was associated with increased odds of LTC placement in those adults who were non-frail or only mildly frail but did not impact odds of LTC placement among the frailest participants. (Godwin et al., 2019, p. 9)
Another Canadian study of residents of assisted living facilities showed that those with poor social relationships had a significantly increased risk for placement in a nursing home (Maxwell et al., 2013). Data from the English Longitudinal Study of Ageing showed that loneliness predicts LTC admission, independent of functional status (Hanratty et al., 2018).
Factors Associated with Access
A limited number of studies have examined access to health care and its relationship with social isolation and loneliness. A lack of transportation resources limits one’s ability to get to medical appointments, for instance, and living in remote areas limits an individual’s social networks and other resources, including rural health services (Hadley Strout et al., 2016; King and Dabelko-Schoeny, 2009). Moreover, for lesbian, gay, and bisexual older adults, living in
rural communities has been found to reduce social networks (King and Dabelko-Schoeny, 2009). As people experience functional declines and adverse health events, access to resources is further compromised by increased isolation (King and Dabelko-Schoeny, 2009).
Having better and more robust social networks makes it more likely that individuals will make greater use of health and social services, as demonstrated in the Village model, a consumer-driven housing model for aging in place in which the coordination of needed health and social services is provided by the residents and delivered within these communities (Graham et al., 2014). Greater access to services, including to health care, was associated with the use of companion resources (e.g., friendly visiting, check-in calls), volunteer involvement, and attending social activities. However, the benefits were lessened for those who had worse self-reported health. As noted earlier, Ha and colleagues (2019) found that those with larger family networks are likely to have had a higher number of previous hospitalizations. In these cases, family served as the primary source of support during times of medical need and provided more people to call on to access care.
Summary of the Evidence on Access and Utilization
The results of studies of the impact of social isolation and loneliness on health care access and utilization are mixed, with the evidence suggesting an association between loneliness and an increased use of inpatient care, more doctor’s visits, increased re-hospitalizations, and longer length of stays. Persons with larger social networks tend to rely more on outpatient services (as opposed to inpatient stays) than those with smaller networks. Physical functioning and health status are linked with both social isolation and loneliness. Older adults who are higher functioning and have higher perceived health status have more options to be socially connected. The oldest of the older adults appear to have fewer options for social connection, thus placing them at greater risk. Furthermore, most of the research examines utilization at the systems level, not at the level of the individual, and therefore individual characteristics (such as the impact of comorbidities like depression) are not the focus of the analysis. Issues of access such as transportation, geographical location, and socioeconomic status all contribute to an individual’s risk for social isolation and loneliness.
Targeting the major social and behavioral risk factors for health offers a way to improve population health and even reduce health disparities. Healthy People,1 a program of the U.S. Department of Health and Human Services,
“provides science-based, 10-year national objectives for improving the health of all Americans” (HHS, 2019a). The program establishes benchmarks and monitors progress in order to:
- Encourage collaborations across communities and sectors,
- Empower individuals toward making informed health decisions, and
- Measure the impact of prevention activities (HHS, 2019a).
Addressing the social determinants of health is a priority in the Healthy People 2020 agenda (HHS, 2019b). Social cohesion is noted as a key issue within the determinant area of social and community context and social support.
On a macro level, public health and managed care organizations affect large populations of older adults through policy and programs (e.g., the fully integrated Medicare and Medicaid special needs plans). These health plans are structured to take a holistic and comprehensive approach to addressing the social determinants of health. For example, one of the social risk factors observed most frequently by those in the Care Wisconsin program was limited social supports (Fouad et al., 2017). Gottlieb et al. (2016) assessed the efforts of 25 geographically dispersed Medicaid managed care organizations (MMCOs) that designed programs to address the social needs of beneficiaries. The authors suggest that one way to address the non-medical factors related to health is to design programs that are integrated into clinical settings. However, the authors found that MMCOs are “not yet systematically engaged in comprehensive [social determinants of health] strategies to improve health or change health care utilization patterns” (Gottlieb et al., 2016, p. 374).
On an individual level, older adults in the United States will ideally experience first-contact care that is comprehensive, continuous, and coordinated through the primary care experience. Comprehensive and advanced primary care delivers significantly more high-value care and better health care access and experience than typical primary care without significantly altering the overall volume of outpatient, emergency department, or inpatient visits (Levine et al., 2019). Newer models of primary care have built on this success. A 3-year primary care medical home intervention, which included a shared savings initiative that created incentives for specific structural transformation, resulted in statistically significant improvements in performance on selected quality measures in all-cause and ambulatory care–sensitive emergency department visits as well as a reduced use of specialty care and higher rates of ambulatory primary care visits (Friedberg et al., 2015). Comprehensive primary care delivered to Medicare beneficiaries aged 65 years or older has been shown to be associated with fewer emergency department and hospital admissions as well as lower Medicare expenditures per beneficiary per month (Bazemore et al., 2015; O’Malley et al., 2019). Assessing the social determinants of health (including social isolation and loneliness) is key to comprehensive primary care.
Primary Care and Assessment of Social Determinants of Health
Many health care delivery systems are exploring the feasibility and impact of practice-based strategies to identify and address such social determinants of health as social isolation and loneliness. The 2019 National Academies consensus study report Integrating Social Care into the Delivery of Health Care noted that “patients visiting health care organizations are increasingly being asked to answer social risk screening questions in the context of their care and care planning” (NASEM, 2019, p. 38). However, the study also notes that collecting data on social determinants of health in the health care setting “may be affected by unconscious or implicit biases held by program leaders and practitioners” (NASEM, 2019, p. 38).
When Tong and colleagues (2018) evaluated the experience of clinicians conducting assessments for social needs (e.g., transportation, food access, housing, social connections), they found three themes that were associated with positive outcomes:
- knowing the patient better,
- understanding the patient’s social circumstances, and
- addressing self-management through such steps as exercise and dietary counseling, addressing financial barriers to medications, and helping with transportation.
However, the process of screening was labor intensive, and the yield varied by how well the clinician knew the patient over time and the willingness of patients to discuss their social needs. The individuals in a practice who may need such assistance most may be the least likely to come in for assessment. Furthermore, if needs are found, connecting patients to resources in the community is difficult.
Tong and colleagues (2018) took a targeted approach to the clinician screening of a subset of registered patients who resided in a geographic region likely to predispose those living there to having social needs. In this study, 57 percent of the targeted cohort visited the practice during the study period and agreed to screening. Of these, more than 70 percent reported at least one social need, yet only 3 percent of those individuals accepted assistance with meeting that need. The authors suggest that the limited number of individuals who were willing to receive help may represent a manageable first step for primary care clinicians who may be otherwise overwhelmed by the volume and prevalence of social needs within the population of patients served by their practice. However, because the approach only addresses those who are willing to participate in an intervention, additional consideration is needed concerning how to address those who are in most need of intervention but who may not initially be willing to participate.
Adding assessments of the social determinants of health to busy health care practices may be considered a burden. In the United Kingdom, Walters et al. (2017) studied the feasibility of embedding a health and social risk appraisal tool into the electronic health records (EHRs) of five English National Health Service primary care practices. The Well-Being Interventions for Social and Health study assessed 454 community-dwelling people aged 65 and older. The fact that the “already burdened” practices were interested in implementing the tool and willing to implement it universally was a positive finding. However, Walters and colleagues (2017) expressed concern that the subset of patients who completed the case-finding tool may not be representative of the needs of the entire population served. They concluded that a practice-based case-finding approach may limit access to services for the high-risk populations who need them, such as the poor, severely ill, and homebound. Furthermore, in a study of strategies for collecting data on the social determinants of health for the EHR, “clinicians did not want to collect [social determinants of health] data themselves, preferring to transfer that responsibility to another team member” (Gold et al., 2017, p. 6). Concerns were also raised about the impact on workflow.
Researchers are considering how informatics might be used to make social determinants of health data collected in the EHR accessible, the use of implementation science to address program development and deployment, and natural language processing to identify information related to the social determinants of health, such as an individual patient experiencing social isolation, in clinical notes (Bazemore et al., 2018; Hripcsak et al., 2015; Zhu et al., 2019). (See later in this chapter for more on the EHR.)
The following sections highlight the general opportunities and challenges related to the clinical assessment of social isolation and loneliness.
The evidence strongly indicates that social isolation and loneliness affect health. Because of this, the health care sector has a role to play in identifying individuals at risk for, or already experiencing, social isolation and loneliness in order to mitigate the health consequences. However, clarity is needed about whether the best approach is a formal screening process or identification of these issues within the patient population. The differences between screening and identification and why the committee chose to highlight these differences are discussed below.
General Principles of Screening
The National Institutes of Health suggests that screening tests can help detect conditions or illnesses early in an illness course or before symptoms are apparent (NIH, 2017). The purpose of such screening is thus to decrease the risks of
certain illnesses, their complications, and their related mortality. Other definitions of screening are:
Screening is the process of identifying healthy people who may have an increased chance of a disease or condition. The screening provider then offers information, further tests, and treatment. This is to reduce associated risks or complications. (Public Health England, 2019)
Screening refers to the use of simple tests across a healthy population in order to identify individuals who have disease, but do not yet have symptoms. (WHO, 2019a)
However, all tests have associated risks and benefits, and the determination of when a screening test is warranted is a source of much debate. The U.S. Preventive Services Task Force reviews the evidence and makes recommendations about whether a particular screening test has sufficient support to be widely adopted into clinical practice.2 The task force focuses on primary prevention (i.e., when there are no symptoms or signs of the illness or behavior). To date, the majority of screening recommendations by the task force have focused on disease-specific or medication-specific concerns. There are few, if any, categories of screening that relate to the social determinants of health in general or certainly for social isolation and loneliness specifically. Thus, there are currently no recommendations for screening for social isolation and loneliness at a national level.
Ultimately, the basic concept underlying screening is that the early detection of risk factors or of early disease will result in better clinical or public health outcomes. See Box 7-2 for criteria commonly used to determine if screening is warranted.
Screening Versus Assessment for Social Isolation and Loneliness
Loneliness and social isolation have high prevalence rates in adults over age 60 (see Chapter 1) and have been linked to significant health consequences and increased mortality risk (see Chapters 2 and 3), which indicates that it might be valuable to have a national standard or recommendation for screening for social isolation or loneliness. Ultimately, however, the value of such a standard or recommendation will depend on several factors, not just the prevalence and health consequences of social isolation and loneliness, but also whether there is a potential treatment or way to mitigate risks in such situations as well as the potential risks of screening and its possible unintended consequences (Garg et al., 2016). Several elements of the Wilson and Jungner criteria (see Box 7-2) support screening for social isolation or loneliness. However, at present there is a lack
2 For more information, see https://www.uspreventiveservicestaskforce.org/Page/Name/about-the-uspstf (accessed December 16, 2019).
of the sorts of policies and data needed to guide clinicians in making decisions about specific treatments or interventions (see Chapter 9). Furthermore, a review of the existing literature finds no high-quality studies demonstrating that social isolation and loneliness can be prevented through primary prevention, although there are some promising secondary and tertiary prevention areas of research (see Chapter 9). Because of the paucity of literature on successful interventions for specific populations, it is difficult to conclude that formal screening protocols for social isolation and loneliness could reduce prevalence rates or negative health consequences. Yet, because of the high prevalence rates and broad-reaching health effects, the committee concludes that the health care system is well poised to develop methods and processes for identifying social isolation and loneliness in health care settings, even if the methods do not rely on a formal screening protocol. As such, the committee concludes that it is more appropriate to talk about the identification of loneliness and isolation as risk factors for health consequences rather than about how to screen for social isolation and loneliness.
Specific Concerns for Clinical Assessments
Similar to the concerns about clinician burden that were raised earlier in this chapter, some people have voiced concerns that charging health care providers with identifying social isolation and loneliness or their risk factors among patients
may create a burden by asking providers to identify social problems that cannot be readily fixed. However, a recent study examining clinician burnout demonstrated that when clinicians felt they had the needed support to address problems such as social isolation, burnout rates were actually lower (De Marchis et al., 2019a).
As the health care delivery system mobilizes to incorporate assessments and, as the research develops, to integrate interventions for the social determinants of health, some researchers are cautioning that not all patients may view primary care interventions of this type as positive. Kharicha and colleagues (2017) found that individuals identified as lonely often did not perceive primary care and community interventions as desirable or helpful and that they perceived a stigma in being labeled as lonely. In particular, many participants did not see loneliness as an illness and therefore did not see a role for primary care physicians, whom they perceived as not being able to help with non-physical problems. “For many, loneliness was a complex and private matter that they wished to manage without external support” (Kharicha et al., 2017, p. 1733). However, another study found that a strong majority of adults reported screening for (or, in the committee’s preferred language, identification of) risk to be appropriate; as such, a fear of stigmatization should not necessarily be considered as a barrier to implementation (De Marchis, 2019b).
Chapter 6 describes many of the tools used in research settings to measure social isolation or loneliness. Unfortunately, few if any implementation studies examine how to use these tools in clinical settings or define which are the most favorable tools to use in specific settings or populations. The committee suggests that a logical approach would be to consider how and why the tool is being used. For example, in health care encounters if clinicians are seeking to determine how to improve health care outcomes or how to reduce the risk of negative health care outcomes, it may be necessary to identify both social isolation and loneliness. In order to compare certain groups (e.g., by age or high-risk attribute), defining the target population (i.e., those who are isolated versus those who are lonely) will help determine what demographic factors underlie the context for their social isolation or loneliness. Another aspect of choosing the best tool is having a framework or theory of change. For example, is the intervention intended to decrease loneliness, to decrease social isolation, or both? Answering this question will help determine whether to use questions related to social isolation, loneliness, or other composite measures.
Implementing Assessment Tools
Despite the limited data on implementation, when health care providers are adapting or selecting research tools for clinical settings they can consider various
general factors that are important to the successful implementation of an identification tool. These include the amount of training required to administer the tool; whether a tool can be administered by clinicians and, potentially, ancillary staff and other clinical team members; the time required to administer the tool; and the tool’s availability and validation in other languages. Using this framework and working with the available evidence, the committee concluded that the existing tools likely to have the greatest success in clinical settings are the Berkman–Syme Social Network Index (for measuring social isolation) and the three-item UCLA Loneliness Scale (for measuring loneliness). (See Chapter 6 for descriptions of these measures.) No composite measurement tools have been designed specifically to measure the overarching rubric of social connection (although some may approximate this). The three-item UCLA Loneliness Scale alone does not get into the details of the quantification of isolation; similarly, the Berkman–Syme does not delve into the subjectivity of loneliness.
As noted in Chapter 6, concerns exist regarding the relevance of current tools, and particularly as to whether the measures developed years or decades ago can fully capture the expectations and values of older adults today. This is likely to be especially relevant for measures of social isolation as modes of interaction and social preferences have changed significantly in recent years and decades. Some have suggested using “living alone” as a proxy measure for social isolation and loneliness. However, as living alone may be a distinct or even positive experience, particularly for those who choose to live alone, and as it does not necessarily capture the distress of loneliness, this question by itself may not be sufficient to fully capture the health risks or the entire context of those experiencing social isolation or loneliness. Similarly, the Berkman–Syme measure, for example, has a single question that focuses specifically on religiosity and participation in religious group activities, which may create a bias against those who do not participate in religious groups but do participate in other social activities with equal benefit. Furthermore, the Berkman–Syme measure also only asks about “telephone use,” which may not account for other modes of communication in today’s society such as texting and video calling. Another drawback of many of these tools is their limited testing and availability in other languages.
Advances in health technologies, including the digitization of medical records, has resulted in vast amounts of data from both “formal” sources, such as clinical tests and imaging, and “informal” sources, such as wearable consumer devices and health-tracking applications in smartphones. This explosion of population-level data, coupled with an emphasis on evidence-based medicine, has led to increasing investment in predictive analytics in health care. These technologies include machine learning and statistical risk scoring and have been widely used
Data for predictive analytics can be gathered from a variety of sources, including various types of information captured in the EHR, such as visit patterns, medications, and patient portal messages; retail activity such as prescriptions filled or over-the-counter medications purchased; social media and internet usage, including search history data;3 and physical activity monitors such as wearable sensors and other consumer health devices. Other sources of data that may be of use for predictive analytics are information captured in files maintained in social service agencies and, in the future, outcomes data from the widespread implementation of assessment tools for social isolation and loneliness and for potentially precipitating life events (Weissman et al., 2020).
Additional Implementation Needs for Clinical Assessment
In addition to selecting the right tools for the valid clinical assessment of social isolation and loneliness, it is also necessary to determine:
- who should receive the assessment (i.e., everyone or just those most at risk),
- who should conduct the assessment,
- the ideal frequency of assessment for different subpopulations, and
- the appropriate interventions, referrals, and follow-up care.
When social isolation or loneliness are identified, it may be appropriate to assess for other potential co-existing conditions (e.g., depression, safety concerns, cognitive impairment) and to engage in advanced care planning, particularly if the individual has no friends. That is, advanced care planning may be needed in situations in which the individual is not capable of making his or her own medical decisions but has no surrogates to make those choices. In addition, follow-up will be needed to determine the severity and the individual’s response to any potential interventions or resources provided. This may be done at the next clinical encounter. No literature currently delves into the appropriate frequency of measurement, but following more established models (e.g., depression screening) may be a reasonable comparison. Also, the initiation of an intervention may depend heavily on the individual’s willingness to participate in such an intervention (as is true for many issues of health care and adherence to treatment). Finally, it is critical for health care systems to maintain these assessments in easily identifiable locations
3 A recent study of the Google search histories of those admitted into the emergency department found in the week leading up to admission, more than 50 percent of those patients searched for information about their symptoms or nearby hospitals (Asch et al., 2019).
in the EHR. (See the next section for more on documentation of social isolation and loneliness in the EHR.)
The IOM Committee on the Recommended Social and Behavioral Domains and Measures for Electronic Health Records was commissioned “to identify domains and measures that capture the social determinants of health to inform the development of recommendations for Stage 3 meaningful use of electronic health records” (IOM, 2014, p. 5). EHRs facilitate the systematic implementation of evidence-based interventions in clinical practice. Large-scale EHR products include modules for tracking the social determinants of health. Features can be customized within the modules, which also include functions for facilitating follow-up and linking to community agencies.
The IOM committee issued its reports in two phases: first, an identification of domains and criteria for inclusion and, second, specific measures in each domain along with issues and opportunities related to the implementation of the measures. The criteria for domain selection included
- strength of the evidence associating the domain with health;
- usefulness of the domain for decision making, monitoring, and research;
- availability of standardized measures;
- feasibility of using the measures in a clinical setting;
- sensitivity of personal information; and
- accessibility of data from other sources.
The domains selected include (1) sociodemographic, (2) psychological, (3) behavioral, (4) social relationships, and (5) neighborhoods and communities. The domain “social connections and social isolation” was described as an item not routinely collected in clinical settings but nonetheless a crucial domain for inclusion, with the evidence supporting its inclusion equivalent to the evidence supporting the inclusion of race, education, physical activity, tobacco use, and neighborhood characteristics.
The measures recommended in the Phase 2 report (IOM, 2014) are included in Table 7-1. Importantly, the committee concluded that the Berkman–Syme Social Network Index could be adopted into EHRs (IOM, 2014). The inclusion of this information in the EHR may vary according to clinical setting (e.g., primary care, inpatient, emergency department), and decisions about its use in EHRs will need to take into account the purpose of the information, how it is used, how to track it over time, and how to ensure it is easily viewable and extractable. Some EHRs have modules for the social determinants of health that enable structured documentation and presentation of the data. Additional options for
TABLE 7-1 Recommended Domains and Measures from the Institute of Medicine Committee on the Recommended Social and Behavioral Domains and Measures for Electronic Health Records
|Alcohol use||3 questions||Screen and follow-up|
|Race and ethnicity||2 questions||At entry|
|Residential address||1 question (geocoded)||Verify every visit|
|Tobacco use||2 questions||Screen and follow-up|
|Census tract-median income||1 question (geocoded)||Update on address change|
|Depression||2 questions||Screen and follow-up|
|Education||2 questions||At entry|
|Financial resource strain||1 question||Screen and follow-up|
|Intimate partner violence||4 questions||Screen and follow-up|
|Physical activity||2 questions||Screen and follow-up|
|Social connections and social isolation||4 questions||Screen and follow-up|
|Stress||1 question||Screen and follow-up|
NOTE: Domains/measures are listed in alphabetical order; domains/measures in the shaded area are currently frequently collected in clinical settings; domains/measures not in the shaded area are additional items not routinely collected in clinical settings.
SOURCE: Adapted from IOM, 2014.
locations to capture these data in the EHR include the history review, the review of systems, or “flowsheets” or “questionnaires” that are already seamlessly incorporated into intake workflows. In order to highlight the clinical significance of loneliness and social isolation in the EHR, a best practice includes adding loneliness or social isolation to problem lists with their corresponding International Classification of Diseases, Tenth Revision (ICD-10) codes.
The IOM committee’s report outlined several potential implementation issues related to the general capture, storage, and use of data that are self-reported or externally sourced (e.g., from community agencies or national surveys). The concerns discussed in the report included privacy, data quality, and the burden imposed on clinicians by incorporating additional data collection into the clinical workflow. The report had a pragmatic emphasis on choices that enable action in clinical settings, but it stopped short of making recommendations related to the communication and coordination infrastructure (e.g., between medical and community-based service providers) that would be needed to implement interventions to address issues related to social determinants for individual patients or groups.
This infrastructure issue was subsequently taken up by the Robert Wood Johnson Foundation’s (RWJF’s) report Using Social Determinants of Health Data
to Improve Health Care and Health: A Learning Report (DeMilto and Nakashian, 2016). This report identified four key barriers to the widespread use of social determinants of health data for improving health:
- lack of knowledge and consensus on measures,
- resource and power differences between social services and health care organizations,
- lack of effective multi-sector collaboration, and
- rigid technology systems.
The RWJF report argues for sharing data across sectors and suggests that EHRs may not be the right tool for this purpose. Research has shown that effective coordination among community agencies, including health care organizations, requires not only an information architecture, but also a process for maintaining a social connection among the agencies (DeMilto and Nakashian, 2016).
While information systems and other digital tools provide a platform to facilitate the assessment and documentation of social isolation and loneliness and, ultimately, to inform action at an organizational level, there are many well-documented challenges related to the use of information technology infrastructure. A number of these challengers were identified by the National Academies consensus study report Integrating Social Care into the Delivery of Health Care, including varying degrees of access to digital infrastructure, a lack of data standards and interoperability, and privacy and security considerations (NASEM, 2019).
The key challenge in implementing effective interventions is linking the information to actions such as outreach, intervention, and follow-up (see Figure 7-1). This requires the following:
- Incorporating assessment data into clinical workflow.
- Establishing evidence-based clinical decision-support protocols for action that are based on assessment findings.
- At the local level, establishing an infrastructure for connections with community resources.
- Using the infrastructure for communication with other clinical providers to support patients transitioning among clinical settings, including secure electronic messaging that protects patient privacy when available.
- Communicating with patients via patient portals and personal health records when available.
Between 2016 and 2018 Kaiser Permanente Northwest (KPNW) used a combination of EHR capabilities and patient navigators to screen more than 11,000 patients for the social determinants of health. The aim of this KPNW initiative was to better understand the connections needed to address patients’ non-clinical needs and to understand the impact on health outcomes of meeting those needs
(Friedman and Banegas, 2018). In a parallel process, OCHIN, Inc., a nonprofit health care innovation center in Oregon, reported on the process that was used and the lessons that were learned through the implementation of a social determinants of health screening tool across a network of more than 400 federally qualified health centers in the KPNW region, taking advantage of the KPNW tool’s social isolation domain and questions (Gold et al., 2017). Researchers identified several considerations for using the tool, including striking a balance between standardized data collection and the need to adapt to local context, identifying patients who do not want assistance, determining a method for updating lists of local resources and referrals, and accommodating different staffing structures.
- Evidence strongly indicates that social isolation and loneliness have an adverse impact on health. Therefore, the health care sector should play a role in identifying individuals at risk for, or already experiencing, social isolation and loneliness in order to mitigate their health consequences.
- Evidence suggests an association between loneliness and increased use of inpatient care, more health care provider visits, increased re-hospitalizations, and longer length of stays.
- Access to services is influenced by factors such as transportation, geographical location, and socioeconomic status, all of which may be associated with social isolation and loneliness.
- The health care system is well poised to develop and evaluate methods and processes to identify social isolation and loneliness in health care settings.
- It is more appropriate to consider and plan for the identification of loneliness and social isolation as risk factors for health consequences than to consider how to screen for social isolation and loneliness.
- The current identification tools most likely to be successfully implemented in clinical settings are the Berkman–Syme (for measuring social isolation) and the three-item UCLA Loneliness Scale (for measuring loneliness).
- More effort is needed to update existing measures as well as to develop better instruments for assessing social isolation and loneliness in clinical settings that can fully capture the experience of today’s older adults.
- A key aspect of selecting a tool for use in clinical settings is standardization within a specific organization so that everyone within the organization uses the same tool or set of tools rather than resorting to different tools, everyone uses validated tools, and everyone refrains from using only parts of existing tools or creating a new, unvalidated tool.
- Advances in health technologies, increases in data collection, and an emphasis on the use of evidence-based medicine have led to the application of predictive analytics to various health concerns. Predictive analytics may also be of value for identifying individuals at risk for social isolation and loneliness.
Chapters 2 and 3 show the strength of the evidence for the mortality and morbidity impacts of social isolation and loneliness on older adults. As such, the committee concludes that the health care system has an important role to play in the identification in clinical settings of social isolation and loneliness in older adults. In fact, a single interaction with the health care system may represent the only touchpoint for the most isolated and lonely older adults. For example, a home health worker may provide the only face-to-face interaction for an older adult who is housebound, has no family, and does not belong to a religious institution or social group.
However, no single clinical indicator (or measure) serves as a marker for the presence or risk of social isolation or loneliness. Because of the scarcity of literature on effective interventions (see Chapter 9), it is premature to conclude that formal screening for social isolation and loneliness could reduce prevalence
rates or negative health consequences. Yet, because of the high prevalence rates (see Chapter 1) and extensive health effects (see Chapters 2 and 3) of social isolation and loneliness, the committee concludes that the health care system is well poised to begin the process of developing methods to identify social isolation and loneliness in health care settings, even if providers are not using a traditional screening approach. By first identifying those at highest risk, and potentially whether their social isolation or loneliness is acute or chronic, clinicians and health care researchers may be able to use these findings to target appropriate clinical and public health interventions to individual patients as well as to target high-need regions and populations served by a practice or health care system. Furthermore, this will support a stepwise approach to care that includes the identification of individuals at risk, the provision of education, and, ultimately, intervention. Finally, for many older adults who are socially isolated or lonely, health care providers may be able to identify underlying causes for the social isolation and loneliness that may be addressed through established evidence-based practices. For example, as discussed in Chapter 4, hearing loss is associated with social isolation and loneliness. In this case, a practitioner would be able to make appropriate referrals to a hearing health specialist. While some question the value of identifying individuals at risk for social isolation and loneliness when in many cases specific, effective interventions have not been developed, the committee recognizes that many health care providers and professionals are already implementing programs for social isolation and loneliness, and so program developers need to understand best practices for identifying at-risk individuals to engage in these programs. Finally, within this context, the committee emphasizes that the preservation of an individual’s own decisions regarding his or her life is essential as a guiding principle for all interventions, including assessment. Therefore, the committee identifies the following goal and recommendations:
GOAL: Translate research into health care practices in order to reduce the negative health impacts of social isolation and loneliness.
RECOMMENDATION 7-1: Health care providers and practices should periodically perform an assessment using one or more validated tools to identify older adults experiencing social isolation and loneliness and to initiate potential preventive interventions after having identified individuals at elevated risk due to life events (e.g., loss of someone significant, geographic move, relevant health conditions).
- In the case of older adults who are currently socially isolated or lonely (or at an elevated risk for social isolation or loneliness), health care providers should discuss the adverse health outcomes associated with social isolation and loneliness with these older adults and their legally
appointed representatives. Providers should make appropriate efforts to connect isolated or lonely older adults with needed social care.
- For older adults who are currently socially isolated or lonely, health care providers should attempt to determine the underlying causes and use evidence-based practices tailored to address those causes (e.g., hearing loss, mobility limitations).
As discussed in Chapter 6, there are a variety of established tools to measure social isolation and loneliness, each with different strengths and weaknesses. Despite the limits of the evidence base concerning how to best implement these tools in clinical settings, the committee concluded that an important aspect of selecting a tool for use in clinical settings is standardization. This means that within a specific health care system or organization, all clinicians would use the same tool or set of tools rather than resorting to different tools; they should also use only validated tools and refrain from using only parts of existing tools or creating new, unvalidated tools. While the committee recognizes that some variation in choice of appropriate tools may be necessary for assessing certain specific populations or health conditions, it emphasizes that the chosen measurement tool needs to match the research question. (That is, if assessing for loneliness, for instance, the tool needs to be validated specifically for the measurement of loneliness, as opposed to other indicators of social connection.) Furthermore, the committee notes that the thresholds for identifying socially isolated or lonely older adults and their risk of health impacts will vary with the tool used and the health profile of the person being assessed. While there are limitations to current tools, the committee asserts that the use of existing validated tools is necessary in order to address social isolation and loneliness more fully in clinical settings. However, the committee recognizes that more effort is needed to update existing tools and develop better tools that can fully capture the experience of social isolation and loneliness among today’s older adults.
The committee also notes that more research related to assessment is needed to evaluate the ethical implications and unintended consequences of assessments as well as to determine specific implementation parameters, including
- who should receive the assessment,
- who should conduct the assessment,
- the ideal frequency of assessment for different subpopulations, and
- the appropriate interventions, referrals, and follow-up care.
A variety of mechanisms for performing these assessments may need to be explored, including the Medicare annual wellness visit; hospital discharge planning; pre-admission, quarterly, or other assessments for long-term care settings;
or other opportunities in which assessment for social isolation and loneliness may be incorporated.
Linking those who are implementing new interventions in clinical settings with formally trained researchers early on can help ensure robust research design. Therefore, in order to improve the evidence around the use of specific tools in clinical settings, the committee makes the following recommendation:
RECOMMENDATION 7-2: Health care systems should create opportunities for clinicians to partner with researchers to evaluate the application of currently available evidence-based tools for assessing social isolation and loneliness in clinical settings, including testing and applications for specific populations.
Finally, the committee concludes that assessment data should be included in clear locations in the EHR. Therefore, the committee makes the following recommendation:
RECOMMENDATION 7-3: The committee endorses the recommendation of previous National Academies reports that social isolation should be included in the electronic health record or medical record.
As noted in both this chapter and Chapter 6, the committee recognizes limitations of current measures of social isolation (e.g., Berkman–Syme) in capturing current modes of interaction. However, as stated previously the committee asserts that the use of existing validated tools is necessary in order to move forward. The measures used and captured in the EHR need to be updated as better measures are developed. The committee further notes that research will be needed to determine