Abstract
The influence of social relationships extends beyond emotional well-being to influence long-term physical-health outcomes, including mortality risk. Despite the varied measurement approaches used to examine social relationships within the health literature, the data can be synthesized using social connection as an organizing framework. This review discusses cumulative scientific evidence of links between various aspects of social connection and mortality, as well as supporting evidence for links with morbidity and plausible mechanisms. This evidence fulfills the criteria outlined in the Bradford Hill guidelines for establishing causality. Despite strong evidence currently available, several gaps remain and will need to be addressed if society is to rise to the challenge of developing effective interventions to reduce risk associated with social disconnection. This evidence has important broader implications for medical practice and public health.
Keywords
The public widely recognizes that social relationships have a strong influence on emotional well-being; however, the association with physical health and longevity is largely unrecognized (Haslam et al., 2018). Across several species, including humans, being socially connected to a group serves essential functions vital to survival (e.g., safety and efficiency of effort). Contemporary science supports the notion that social connection is adaptive across the life span, suggesting that humans are “wired” to be social, such that our brains and bodies expect the proximity to others. For example, neuroscientific studies reveal that people faced with a threat use more metabolic resources if they are alone than if they are in the presence of others (Coan & Sbarra, 2015), suggesting that there are biological manifestations of motivational cues to maintain close social ties. However, in recent years, a growing concern about social isolation and loneliness has emerged, based in part on surveys that suggest high prevalence rates (Cigna, 2020). Nonetheless, the prevailing concern remains focused on emotional well-being, and less attention is paid to the potential deleterious physical-health ramifications of social disconnection. Thus, it is important to examine the strength of the scientific evidence establishing the relevance of social relationships to physical health and the implications for public health.
In this article, I summarize my own and other researchers’ work that has helped establish the link between social connection and physical health, with an emphasis on mortality risk. To establish the relevance of social connection to health, I first contextualize how it has been defined and measured in the literature. I then review meta-analytic data concerning the links between components of social connection and mortality, along with supporting evidence on morbidity and plausible mechanisms. This evidence is then contextualized in relation to the Bradford Hill guidelines for the establishment of causality (Hill, 1965). I conclude with a brief discussion of the challenges of developing effective interventions to reduce risk and broader implications for medical practice and public health.
What Is Social Connection?
The scientific evidence on the health implications of social relationships has accrued over several decades, from research spanning multiple disciplines and utilizing a diverse set of conceptualization and measurement approaches. Some measurement approaches have focused on social assets (e.g., social support, social integration), and some on social deficits (e.g., social isolation, loneliness), but most have used a mix of items that capture both connection and disconnection to arrive at a total score that is treated continuously. 1
Three broad categories of measurement approaches have emerged in the epidemiological research on links between social relationships and physical-health outcomes: those focused on structural aspects of social life (e.g., size of network, number of social roles, whether a person lives alone or with others), those focused on functional aspects of social life (e.g., received or perceived availability of social support, resources, and aid), and those focused on the quality of relationships (affective valence; e.g., satisfaction, strain, adjustment). Evidence links each of these conceptualizations to health risk (Holt-Lunstad et al., 2017). Although there is some degree of overlap among these three components of social connection, they are not highly correlated. For example, having a larger social network (social structure) may be associated with more sources of support and aid (functions) and greater satisfaction (quality); however, a large social network (structure) may not provide the level of support needed (function), or, alternatively, may provide reliable social support (function) that comes with conflict, nagging, or intrusiveness (quality). Thus, each component may contribute to risk or protection distinctly.
The term social connection (or, conversely, social disconnection) has been proposed as an umbrella term that encompasses the structure, functions, and quality of social relationships (see Fig. 1; Holt-Lunstad, 2018; Holt-Lunstad et al., 2017). Social connection may be conceptualized as a continuum, such that high levels on the components of social connection are associated with protection, and low levels of social connection (or social disconnection) are associated with risk. Thus, a multifactorial construct of social connection refers to the continuum from risk to protection.
Schematic illustrating social connection as a continuous multifactorial risk factor with components related to structure, function, and quality. Low levels of these components are associated with risk and high levels are associated with protection.
Health Consequences
Mortality
Social disconnection, including social isolation and loneliness, is associated with serious long-term consequences, including earlier death. Some of the earliest evidence comes from the sociologist Emil Durkheim (1897/1951), who found that people who were more socially isolated were at increased risk for death from suicide. Although the link to mortality may be particularly apparent for suicide, there is robust evidence that social disconnection is associated with an increased risk of death from any cause.
Indeed, the strongest evidence linking social connections to health-relevant outcomes comes from the literature that has examined mortality as an outcome. Large-scale prospective epidemiological research has followed individuals over years, often decades, to determine whether social-connection indicators predict who lives or dies, or time to death. In one of the first large-scale syntheses of this evidence, data from 148 studies were analyzed to examine the protective effect of being socially connected (assessed in terms of structure, function, and quality). Results revealed that social connection increases odds of survival by 50% (Holt-Lunstad et al., 2010). A subsequent meta-analysis focused specifically on indicators of social disconnection among more than 3.4 million participants followed for an average of 7 years. After accounting for multiple covariates, analyses revealed that loneliness was associated with a 26% increased risk for premature death, social isolation was associated with a 29% increased risk, and living alone was associated with a 32% increased risk (Holt-Lunstad et al., 2015). Both subjective (loneliness) and objective (social isolation and living alone) indicators of social disconnection were significant predictors of mortality risk. In both meta-analyses, the findings were consistent across gender, country of origin, and cause of death. These findings were also independent of age and initial health status, which rules out reverse causality.
There are now several additional meta-analyses that have documented the association between some component of social connection and mortality (Pinquart & Duberstein, 2010; Rico-Uribe et al., 2018; Roelfs et al., 2011; Sbarra et al., 2011; Shor & Roelfs, 2015; Shor, Roelfs, Bugyi, & Schwartz, 2012; Shor, Roelfs, Curreli, et al., 2012). Although there is variability in the strength of the effect across measurement types, the magnitude is comparable with and in some cases exceeds the magnitude of risk for other risk factors for mortality (e.g., obesity, physical inactivity, air pollution; Holt-Lunstad et al., 2017). Thus, the cumulative data provide compelling evidence that being socially connected is a protective factor and being socially disconnected is a risk factor for premature death.
Morbidity
The growing body of research has found not only that social connection extends life span, but also that social disconnection is associated with several physical-health outcomes, including Type 2 diabetes (Lukaschek et al., 2017), susceptibility to cold viruses and upper respiratory illnesses (Cohen, 2021), and pulmonary function (Cheng et al., 2014), as well as with health-related lifestyle behaviors (Marquez et al., 2014). The strongest evidence comes from those studies examining cardiovascular outcomes. A meta-analysis of 16 longitudinal data sets found that having poor social relationships was associated with a 29% increase in the risk of incident coronary heart disease and a 32% increase in the risk of stroke (Valtorta et al., 2016). Consistent with these findings, an analysis of data from nearly half a million people in the UK Biobank found that isolated and lonely persons are at increased risk of acute myocardial infarction and stroke (Hakulinen et al., 2018). The evidence linking social connection to cardiovascular outcomes is the strongest, and more research on social connection’s association with other physical-health conditions and behaviors is needed (National Academies of Sciences, Engineering, and Medicine [NASEM], 2020).
Mechanisms
Growing evidence indicates that there are multiple mechanisms that plausibly explain how social connection can impact physical-health outcomes (see Fig. 2). Social connection may influence morbidity and mortality via psychological factors, such as perceived stress (Hostinar, 2015) and depression (Cacioppo et al., 2010); behavioral factors, such as sleep (Kent de Grey et al., 2018), physical activity, and smoking (Shankar et al., 2011); and biological factors, such as inflammation (Uchino et al., 2018). These pathways are not entirely independent and are interconnected. Several reviews have documented plausible biological mechanisms (e.g., Cacioppo et al., 2015; Hostinar, 2015; NASEM, 2020; Uchino, 2006) that explain how social connection may “get under the skin” to influence long-term health outcomes. For example, perceptions of social threats, such as loneliness, may elevate activity of the sympathetic nervous system (SNS) and hypothalamus-pituitary-adrenal (HPA) axis. These alterations in HPA activity (glucocorticoid resistance) and the SNS (innervation of immune-system organs regulating leukocyte development) interact to promote a pro-inflammatory response. Social experiences can alter human genome-wide activity by impairing glucocorticoid response genes and increasing pro-inflammatory transcription (Cole et al., 2007). Notably, chronic inflammation has been implicated in physical (e.g., cardiovascular atherosclerosis, cancer metastasis, pulmonary functioning), mental (e.g., depression), and cognitive (e.g., Alzheimer’s disease) health, which suggests that it may be a common pathway by which social disconnection may influence multiple health outcomes. Despite documentation of these pathways, increased understanding of the molecular mechanisms that promote the behavioral and physiological effects of social disconnection on health is needed.
Simplified model of possible direct and indirect pathways by which social connections influence morbidity and mortality. Solid lines indicate directional influences of social connection on morbidity and mortality, and dotted lines indicate the influence of morbidity on social connection.
Directionality and Causality
Directionality
How can we be sure that social connection leads to better health? Perhaps individuals who are unwell are simply less able than healthy people to be socially engaged. As depicted in Figure 2 (see the dotted line between morbidity and risk factors), physical-health problems may increase risk for social isolation and loneliness; however, despite evidence of bidirectional associations, the cumulative prospective epidemiological evidence establishes a directional effect of social connection, or lack thereof, on mortality risk. Regardless of initial health status, people who are more socially connected have greater odds of survival (Holt-Lunstad et al., 2010, 2015).
Causality
Cumulative evidence indicates that greater social connection is associated with greater odds of survival, but is this association causal? Establishing causality can be challenging given that most epidemiological research is observational, not experimental. Other public-health issues, most notably smoking, have faced similar challenges as risk is often identified via epidemiological evidence. This challenge to establishing causality has been addressed by the Bradford Hill guidelines (Hill, 1965), a frequently cited framework for drawing causal inference in epidemiological research. Moving from association to causation is an important step for taking preventive action against environmental causes of disease. The Bradford Hill guidelines provide nine criteria (see Table 1) by which to evaluate a potential risk factor. None of these criteria other than temporality (causes must precede effects) are regarded as necessary by themselves (Lucas & McMichael, 2005). Furthermore, satisfying any single factor is not sufficient to establish causation; rather, when all or most of the criteria have been met, this can suggest the likelihood of a causal relationship rather than simply an association. These guidelines place particular emphasis on the temporality of the relationship, its strength, the presence of a plausible dose-response relationship, the consistency of findings in diverse studies, and coherence with other disciplinary findings and biomedical theory (Lucas & McMichael, 2005).
Bradford Hill Guidelines and Summary of Potential Supporting Evidence for Effects of Social Connection on Physical Health
As noted in the text, all criteria are not necessary to establish causality. Although the criterion of specificity is not met, this is true of many lifestyle and behavioral risk factors, including smoking, which also gives rise to multiple health outcomes (e.g., cancer, heart disease, stroke, diabetes).
Using the Bradford Hill guidelines to evaluate the case for social connection as a causal risk factor for health and mortality indicates that there is indeed some evidence to support this idea (see Table 1; NASEM, 2020; see also the more detailed analysis in Howick et al., 2019). The meta-analytic data on mortality fulfill multiple criteria, including strength, consistency, temporality, and analogy. In addition to establishing the overall strength of effect, the prospective studies that have followed people over years or decades establish temporality, and the consistent findings across more than 148 studies using a variety of measurement approaches establish consistency and analogy. Going beyond the mortality data, several reviews have documented plausible biological mechanisms (e.g., Cacioppo et al., 2015; Hostinar, 2015; Uchino, 2006). Data from four nationally representative samples document a dose-response effect of the structural, functional, and quality components of social connection on multiple health-relevant biomarkers across the life span (Yang et al., 2016), supporting a biological gradient. Although obtaining experimental evidence in humans is challenging, there is experimental evidence from studies in which nonhuman animals have been randomly assigned to be housed alone or housed socially. Nonhuman-animal models also provide molecular, cellular, immunologic, and behavioral validation of potential effects of human social disconnection and an experimental method of charting the causal interactions among social perception, neural activity, immunologic function, and health (Cole et al., 2015). Further, acute lab-based and intervention research in which people have been randomly assigned to different social situations has yielded health-relevant outcomes (Uchino, 2006). In short, according to the Bradford Hill guidelines, the evidence supports the likelihood of a causal link between stronger social connections and better health and longer life (Howick et al., 2019); nonetheless, the number of studies is larger and the quality of evidence is stronger for some criteria than others, and additional research will help strengthen confidence in the conclusion that there is a causal relationship.
Gaps and Challenges
Despite strong evidence overall, there are important limitations and gaps in current understanding of the implications of social connection for physical health. In this section, I discuss the key gaps identified in NASEM’s (2020) consensus report focused on social isolation and loneliness in older adults; however, these gaps and challenges apply more broadly to the literature on social connection and its implications for people of all ages.
Measurement
Despite the evidence that social connection is a multifactorial risk factor, and despite the Institute of Medicine’s recommendation that social-connection assessment be included in all electronic health records (Matthews et al., 2016), no single assessment tool is comprehensive. For this reason, few studies assess multiple components of social connection within the same sample or examine their synergistic effects, despite evidence that predictive value is greater when complex measures are utilized (Holt-Lunstad et al., 2010). Furthermore, although presently, aggregate data do not support a threshold effect and there is evidence of a dose-response effect suggesting a continuum from risk to protection, more research is needed to establish standardized clinical cutoff scores and/or a risk classification system, which may be critical to application in clinical settings and public-health implementation.
Underrepresented samples
Mortality data are primarily based on middle-age and older adult samples, given that mortality and morbidity are more likely to occur in these age groups than in younger individuals; however, more studies of younger samples are needed for researchers to fully understand the developmental trajectories of social connection and health across the life span. Meta-analytic data have been drawn from countries worldwide; however, there have been few studies from developing countries and few examining potential cultural, ethnic, or economic differences. Thus, more data are needed from groups that have been underrepresented in prior research.
Intervention to reduce risk
Despite strong evidence of the health consequences of social disconnection, there is less compelling evidence on how to intervene to reduce risk. Notably, most of the epidemiological data are based on existing (naturally occurring) social connection, and it is less clear what strategies may adequately fulfill social needs when social relations are absent or insufficient. Reviews of intervention studies show mixed effects and generally use less rigorous methodology relative to the epidemiological studies establishing the health effects. There are several issues that may contribute to the mixed outcomes among the intervention studies. These include inadequate attention to a guiding theoretical framework, specification of measurement, length of follow-up, and underlying mechanisms (NASEM, 2020). Furthermore, most of the intervention evidence comes from tertiary efforts aimed at individuals most severely affected. Among public-health interventions, tertiary efforts are the costliest, require the most resources, and are the least effective; therefore, greater attention and research should be aimed at preventative efforts across the risk trajectory, including primary, secondary, and tertiary prevention.
Social technologies
Over the past decade, significant changes in the means by which people connect socially have occurred. Much of the cumulative epidemiological data predates widespread use of smartphones, social media, and other social technologies (e.g., video chat, social robots, voice-enabled assistive devices). Although there is a growing body of research examining social media, most of the data are cross-sectional; thus, prospective longitudinal studies are needed to understand directionality and long-term effects. Because social technologies are increasingly being explored as intervention tools and being widely implemented during the COVID-19 pandemic, often without sufficient evidence of efficacy, a better understanding of their strengths, their limitations, and related ethical considerations is needed (NASEM, 2020).
Implications for Medical Practice and Public Health
Decades of cumulative evidence point to potential long-term health risks if steps are not taken to mitigate the effects of social disconnection. Given the significant prevalence of social disconnection within the population, social connection should be prioritized in public health (Holt-Lunstad et al., 2017). Recently, NASEM (2020) published an expert consensus report on the health and medical dimensions of social isolation and loneliness, making recommendations for achieving each of the following five goals: develop a more robust evidence base, translate current research into health-care practices, improve awareness, strengthen ongoing education and training, and strengthen ties between the health-care system and community-based networks and resources.
One of the specific recommendations is that “the Department of Health and Human Services should establish and fund a national resource center for evidence, resources, training and best practices on social isolation and loneliness” (NASEM, 2020, p. 217). Stakeholders within and outside of the health-care system are implementing new interventions for social isolation and loneliness using approaches for which there are varying degrees of evidence and that have varying degrees of effectiveness; thus, such a resource is needed. The Administration for Community Living, a division within the Department of Health and Human Services, is currently establishing a clearinghouse with scientific oversight for this purpose.
As stakeholders and researchers design interventions, a systems approach may help move social connection beyond clinical approaches and into the realm of public health (Holt-Lunstad, 2018). Individuals exist within larger social structures, such as families, communities, societies, and cultures; thus, scientific inquiry, as well as prevention and intervention efforts, must expand beyond the focus on the individual. By considering the interaction among individual, relationship, community, and societal factors (i.e., the social-ecological model), researchers can begin to better understand potential causal factors that may originate within these larger social structures to subsequently inform potential interventions at each level. Whereas clinical interventions may focus on the individual (e.g., maladaptive cognition or social skills) or relationship (e.g., peer support, emotion-focused couples therapy), community-level approaches consider characteristics of neighborhoods, schools, and workplace settings that may facilitate or hinder social connection. For example, concerns about neighborhood safety and walkability have been linked to isolation and loneliness (Bergefurt et al., 2019); thus, designing physical spaces, policy, and practice to provide greater social opportunities within community settings may help prevent social isolation.
The COVID-19 pandemic restrictions aimed at reducing spread of the virus via limiting social contact have powerfully highlighted the social relevance of nearly every sector of society. Substantial changes in any sector, via changes in the built environment or policy, can potentially increase or decrease social connection at a population level. The World Health Organization’s (2014) Health in All Policies framework can be used to as a starting point to systematically evaluate existing policy and consider new policies in the health, transportation, housing, employment, education, and environmental sectors that can contribute to social connection and ultimately public health (Holt-Lunstad, 2020). These broad societal factors create a normative climate that either enhances or diminishes the prioritization of social connection.
Conclusion
Strong and compelling evidence shows that social connection is a significant protective factor and social disconnection is a risk factor for morbidity and mortality. The challenge now is what to do about it. Collaborative interdisciplinary science has contributed to the understanding of the complexity of social connection and is further needed to address existing gaps in the evidence to more precisely identify underlying causal mechanisms that may be leveraged in prevention, intervention, and policy efforts. Establishing and evaluating evidence-based practices will be key in prioritizing social connection within medical practice and public health.
Recommended Reading
Holt-Lunstad, J. (2018). (See References). A review placing the construct of social connection within a systems framework to facilitate better understanding of the contributing causal factors that can be targeted for interventions aimed at improving physical-health outcomes.
Holt-Lunstad, J, Robles, T. F., & Sbarra, D. A. (2017). (See References). A review of the evidence supporting the need for social connection to be a public-health priority.
Holt-Lunstad, J, Smith, T. B., Baker, M., Harris, T., & Stephenson, D. (2015). (See References). A meta-analysis of data from 70 prospective studies examining the magnitude of the effect of social isolation, loneliness, and living alone on risk for premature mortality.
Holt-Lunstad, J., Smith, T. B., & Layton, J. B. (2010). (See References). A meta-analysis of data from 148 prospective studies examining the magnitude of the effect of social connection on mortality.
National Academies of Sciences, Engineering, and Medicine. (2020). (See References). An expert consensus report that summarizes the evidence for the relevance of social isolation and loneliness within medical and health-care settings and makes consensus recommendations.
