Social connection as a critical factor for mental and physical health: evidence, trends, challenges, and future implications

Depression

There is a strong positive association of social isolation and loneliness with depression from youth to older adulthood. Further, higher social connectedness is protective towards depressive symptoms and disorders¹⁶.

When looking at adults of all ages, 18 years and older, data from the US National Health Interview Survey examined the impact of living alone and the availability of social and emotional support on depression¹⁷. Adults living alone reported significantly higher depression than those living with others, and this difference held across several sociodemographic factors. Adults never or rarely receiving social and emotional support were twice as likely to report depression, but adults living alone were still more likely to report depression even compared to adults living with others who did not receive social and emotional support¹⁷.

Importantly, longitudinal evidence suggests that social isolation and loneliness likely cause or worsen depression over time. For example, a systematic review of 32 longitudinal studies from the general population examined whether subjective feelings of loneliness predicted the onset of a new diagnosis of depression¹⁸. Studies followed participants from six months to 16 years, with an average follow-up of 3.5 years. The odds of developing new depression in adults were more than double among those who reported often feeling lonely compared to those rarely or never feeling lonely. While there were more studies among older adults, the findings were consistent among younger age groups, including university students and new mothers.

Using two large datasets – the Psychiatric Genomics Consortium meta-analysis of major depression (N=142,646)¹⁹, and the Million Veteran Program (N=250,215)²⁰ – to apply a two-sample Mendelian randomization design, loneliness appeared to cause incident major depression and depressive symptoms¹⁴. These analyses were then reversed using loneliness outcome data from the UK biobank. Remarkably, data demonstrated that loneliness causally predicts major depression, but the reverse is also true, with major depression causally predicting loneliness¹⁴. This suggests that loneliness is both a cause and a consequence of major depression; thus, public health strategies to reduce loneliness may potentially be effective in preventing the onset of depression and reducing depressive symptoms, and better treatments for depression are likely to reduce loneliness.

The link between social connection and depression has also been examined among patients in medical settings, suggesting potential spillover effects on other clinical conditions. For example, low social support had a significant positive association with antenatal depression, which contributes significantly to maternal physical health²¹. In a review, 83% of studies found that pregnant women with low social support had greater depressive symptoms¹⁶.

The links between social connection and mental health are also relevant within occupational settings. The strain on employees across sectors, particularly those hit hardest during the COVID-19 pandemic – such as health care providers, educators, and other “essential employees” – has brought greater attention to burnout and other mental health concerns. A meta-analysis of studies in health care workers found that a lack of social support significantly contributed to higher risk for acute stress disorder, burnout, anxiety, depression, and post-traumatic stress disorder²².

Cognitive health

Several meta-analyses consistently show that stronger social connection – including social networks (e.g., number of social contacts, frequency of interaction, marital status, living arrangement) and social engagement (e.g., attending social groups; visiting family, friends and neighbors; engaging in voluntary or paid work, participation in cultural or leisure activities) – is associated with better cognitive function, but the evidence is less consistent for perceptions of loneliness.

For example, a meta-analysis including over 2.3 million participants showed that living alone, having a smaller social network, having a low frequency of social contact, and having poor social support were risk factors for dementia, while loneliness was not²³. However, other meta-analyses did find that greater loneliness was significantly associated with incident dementia^{24, 25}. Conversely, greater social engagement, including a greater number of social memberships, number of social contacts, and more social participation, may be protective, as these were associated with lower dementia risk^{23, 26}.

Physical morbidity

There is a rich and growing body of evidence across a variety of physical health outcomes, including major health indicators such as cardiovascular diseases, stroke and diabetes mellitus.

Cardiovascular diseases are the leading cause of death globally, accounting for roughly one third of all deaths; therefore, factors that increase or decrease this risk can have a major global health impact⁴⁵. Dozens of studies have found that social isolation and loneliness significantly influence the risk of cardiovascular and cerebrovascular morbidities^{15, 29}.

The culmination of this evidence resulted in a statement published by the American Heart Association in 2022, acknowledging this risk from objective and perceived social isolation²⁸. According to this review of the evidence, there is a clear link of social isolation and loneliness with risk for coronary heart disease and stroke. Among the evidence, a synthesis of data across 16 independent longitudinal studies demonstrates that poor social relationships (social isolation, poor social support, loneliness) were associated with a 29% increase in the risk of incident coronary heart disease and a 32% increase in the risk of stroke²⁹. These findings were consistent across genders.

Low social connection and loneliness have also been associated with a greater risk for hypertension. Indeed, data from the National Social Life, Health and Aging Project suggest that the impact of social isolation on risk for hypertension exceeds that of clinical factors such as diabetes mellitus, pointing to a “causal role of social connections in reducing hypertension” in older age⁴⁶.

Diabetes mellitus is a leading source of disability, lost productivity, mortality, and lower quality of life, affecting nearly half a billion people worldwide, with a significant global economic burden on individuals, health care systems, and countries⁴⁷. Studies have repeatedly shown that social connection (e.g., family support and involvement) can positively influence the management and overall health of individuals with type 1 and 2 diabetes. Large population studies also demonstrate the influence of social connection on the incidence of type 2 diabetes. For example, people with smaller social networks were more likely to have been recently diagnosed with type 2 diabetes, to have previously been diagnosed with this condition, and to have diabetic complications^{48, 49}.

However, gender differences have been found along different indicators of social connection. Low social participation was linked to pre-diabetes and complications among women but not men, while living alone increased the likelihood of previously diagnosed type 2 diabetes and its complications in men but not in women^{48, 49}. These findings were independent of glycemic control, quality of life, and cardiac risk factors.

Diabetic outcomes may be due to better self-care among those who are more socially connected. For example, in a meta-analysis of 28 studies, social support was significantly associated with better self-care, particularly glucose monitoring, and was stronger among those with type 2 than type 1 diabetes⁵⁰. Improving diabetic outcomes via social connection can have cascading public health implications, given that diabetes mellitus often leads to other health outcomes, including heart disease, kidney failure, blindness, amputation and dementia.

There is also evidence to suggest that poor social connection is associated with worse outcomes among those who are already ill. For example, heart failure patients who self-reported high levels of loneliness had a 68% increased risk of hospitalization, a 57% higher risk of emergency hospital visits, and a 26% increased risk of outpatient visits compared with patients reporting low loneliness⁵¹. In a meta-analysis of 13 studies on heart failure patients, poor social connection was associated with a 55% greater risk of hospital readmission⁵². This was consistent across both objective and perceived social isolation, living alone, lack of social support, and poor social network. These data suggest that improving social connection among those who are sick can improve medical outcomes.

Mortality

Several reviews of the evidence, including a NASEM scientific consensus study, have concluded that some of the strongest evidence linking social connection, isolation and loneliness to health-relevant outcomes is that concerning mortality¹⁵. Large population-based epidemiological studies have tracked initially healthy populations over time, for years and often decades, documenting that those who are more socially connected live longer^{35, 38, 41, 42}, while those who experience social deficits (isolation, loneliness, living alone, poor-quality relationships) are more likely to die earlier, regardless of the cause of death^{33, 36, 37, 39, 40, 43}. Although social isolation has been implicated as a risk factor for death by suicide⁵³, most meta-analyses on mortality exclude suicide as a cause of death.

Based on meta-analytic data, one estimate suggests that the association between social connection and survival may be as high as 50%⁴², while isolation is associated with 32% and loneliness with 14% increased risk for earlier death³³. While estimates vary to some extent, they may be conservative, given that many reviews and meta-analyses often exclude studies that focus specifically on deaths due to unnatural causes such as unintended injuries, violence or suicide. While there are more studies and stronger effects on cardiovascular-related deaths (e.g., myocardial infarction, stroke) and cancer-related deaths (e.g., leukemia, lymphomas, breast cancer)⁴¹, more research is still needed on these, in addition to other disease-related causes of death.

Over the years, the number of studies, the rigor of methodology, and the size of samples have all increased substantially, replicating the finding that social connection decreases the risk of premature mortality and providing stronger confidence in this evidence. For example, longitudinal data from the UK Biobank regard nearly half a million people, reducing the likelihood of random error⁵⁴. These data demonstrate that social isolation significantly increases risk for earlier all-cause mortality, overall and consistently across subgroups (i.e., males and females, young and older, health and unhealthy, various ethnicities), even after adjusting for a robust set of lifestyle, socioeconomic, biological, and health risk factors⁵⁵.

Several meta-analyses and systematic reviews have documented similar findings across different ways of examining the issue, including social relationships broadly, social networks, social contact frequency, marital/partnership status, marriage dissolution, social isolation, loneliness, and living alone^{32, 43}. While the magnitude of the effect varies to some extent across studies and depending on which aspect of social connection is being examined, the evidence points to the same general conclusion: indicators of greater social connection are associated with reduced risk, while indicators of social deficits are associated with greater risk for premature mortality.

Trends in society

Societal trends over the past several decades indicate that, as a population, we have become less socially connected and more isolated, and that a high proportion of the population is lonely.

Based on the available data, loneliness has generally shown little improvement over the last few decades, and may be getting worse. For example, a massive synthesis of 345 studies on emerging adults (ages 18-29), who completed the UCLA Loneliness Scale between 1976 and 2019, found that average loneliness levels linearly increased annually across the 43 years⁸⁴. Furthermore, a meta-analysis of data from 113 countries concluded that a substantial proportion of the population in many countries experiences problematic levels of loneliness⁸⁵.

According to the Gallup Global State of Connection survey, nearly a quarter (24%) of the global population reports feeling “very lonely” or “fairly lonely”, although there was variability across countries⁸⁶. Of the 29 countries where at least one third of the population felt lonely, 22 were in Africa, four were in the Middle East, and three in South Asia. This also demonstrates that loneliness is not just a wealthy Western country issue, and may even be more severe in other areas of the world. However, inconsistent measurement tools and scoring methods have led to vastly different prevalence estimates. Notably, prevalence rates often favor one indicator (e.g., loneliness) over others, yet indicators may interact in meaningful ways. Thus, the prevalence of those who lack social connection in one or more ways may be far larger than any estimate of a single indicator.

Loneliness trends provide an incomplete picture of the state of social connection, and we must look at the other ways in which individuals and communities may lack connection. For example, data from the American Time Use Survey, regarding how Americans spend their day, demonstrate that, over the past two decades, Americans have spent more time in isolation and less time with household and non-household family members, friends, community engagement, and companionship⁸⁷. Although the COVID-19 pandemic exacerbated these trends, social isolation was increasing, and engagement with family, friends and others (co-workers, neighbors, acquaintances) was declining for years prior to the pandemic. This is consistent with other trends, such as those documenting a decline in social capital and participation in religion^{88, 89}, and changes in family structure (e.g., decline of extended families, rise of single-occupancy households)⁹⁰ – many of which are seen globally.

Contemporary society in much of the world is evolving rapidly, likely contributing to our current trends and having important implications for the direction of the trends going into the future. Rapid shifts that may be relevant to social and population health include the increasing aging population, widespread adoption of remote working, increased automation, economic strain and inequity⁹¹, migration and mobility, mental health crisis among youth, rise in xenophobia, civil and political unrest, and environmental crises, all of which may potentially exacerbate trends concerning social connection.

These trends of declining social connection, combined with the evidence on the bidirectional associations with mental and physical morbidities, point to an urgent need to take action. Because multiple factors have been contributing to these trends, building over decades, simply returning to pre-pandemic levels of connection or reducing time on social media may only bring limited benefits.

Trends in scholarly attention

There are also striking trends in the scientific study of the topic. The surging interest in social isolation and loneliness is reflected in research, as demonstrated by the substantial increase of studies on this topic over recent years, potentially providing greater understanding and justification for action. Thus, understanding how loneliness and isolation have been studied over time may provide additional insight.

To examine publication trends, we first used the PubMed by Year search tool. Because of the diverse literature on social, mental and physical health outcomes, the search was limited to two social variables (loneliness and social isolation) and two health outcomes (depression and mortality). We further scanned additional scientific databases (including PsycINFO for depression) using the same social and health variables. The searches were limited to articles published in peer-reviewed academic journals between 1972 and 2023. The mortality search terms included “social isolation” OR “loneliness” AND “death and dying” or “mortality” or “mortality rate” or “mortality risk”. The depression search terms included “social isolation” OR “loneliness” AND “major depression”. PsychINFO also allows narrowing search by methodology: thus, we further used the search parameters “empirical”, “quantitative”, “longitudinal”, “prospective”, “retrospective”, and “clinical trial”.

All studies using those search terms were bracketed into time periods to determine how many articles populated by our search terms were published within each period. Studies published in the past two decades (2004-2023) were demarcated into 2-year periods (2004-2005, 2006-2007, 2008-2009, etc.), while studies published in 1972-2003 were grouped (31 years). Figure 2 graphically presents the number of studies on isolation and loneliness over the years.

Data suggest an exponential increase in the scientific study of social isolation and loneliness. Over the past two decades, the number of relevant articles has grown, with significant increase since 2020. For example, the number of papers published in each subsequent two years since 2020 exceeds the number of studies from 1972 to 2003 combined. However, it is unclear whether scientific interest in other indicators of lacking social connection is similarly surging.

Trends in awareness

Several factors may presumably contribute to greater awareness of the importance of social connection and related aspects of lacking connection (i.e., social isolation and loneliness). These include scientific advancement, social media, government initiatives, the COVID-19 pandemic, and advocacy.

Significant advances in scientific research over the past few decades, especially in the last 5-7 years, may have shed light on the scale of the problem and provided greater confidence in scientific findings. Advancements in social technologies and the widespread use of social media platforms may have played a dual role in awareness. Increased experience of feelings of loneliness associated with that use, and the facilitation of awareness campaigns, discussions and support networks related to health and well-being may occur simultaneously⁹².

Government initiatives may have also played a role in greater awareness. Countries have recognized the urgency of the issue and appointed Ministers, formulated policies, and developed strategies to address loneliness and isolation, and highlight social connection as a priority. Awareness efforts have also been undertaken by national and international civil society organizations, coalitions, and networks that have emerged as powerful advocates⁷. These include the UK Campaign to End Loneliness, the Canadian Genwell Project, Australia's Ending Loneliness Together, and the annual Global Loneliness Awareness Week. These collective efforts aim to raise awareness, promote community engagement, and foster a culture of connection.

Unfortunately, trends in public awareness appear to be limited to only certain outcomes. A large survey of US and UK adults published in 2018 found that, when the public was asked to rank various factors contributing to a longer life (e.g., not smoking, exercising, limiting alcohol, maintaining a healthy weight), social connection was amid these factors, but it was rated among the lowest in importance, significantly underestimating its impact relative to effect sizes reported in the scientific literature⁹³.

Due to a variety of factors occurring since that survey was published – i.e., the COVID-19 pandemic, national awareness campaigns, and increased prevalence within the population – public perception of the health relevance of social connection was expected to increase. However, 2023 data from the UK and a nationally representative sample of US adults demonstrate that there has been essentially no change⁹⁴. Despite increases in public discourse on social isolation and loneliness, the importance of these and other aspects of social connection for health and survival are still underestimated among the public.

Implications from these trends

Overall, these trends point to a large and potentially increasing scale of those lacking social connection, and parallel trends suggest increasing attention within scholarship on the consequent impact on health. Yet, the public perception of risk does not align with either the increasing scale or evidence of the magnitude (effect sizes) for health. This suggests that increased education and awareness of the health relevance is needed.

Discrepancies between the scientific evidence and public perception may have significant implications. First, public perception may significantly influence how resources are allocated and prioritization of various issues within public health agendas⁹⁵. If the public does not perceive social connection and markers of its deficit (e.g., loneliness and social isolation) as relevant to health⁹³, funding and efforts may not be directed towards addressing them adequately, despite their demonstrated impact on health outcomes^{4, 15}. Second, public perception influences individual behaviors and societal norms. If social connection is not widely recognized as a protective factor, and loneliness and isolation as serious health risks, individuals may be less likely to change their own behavior or support others experiencing loneliness or isolation⁹⁶. This may perpetuate social disconnection and exacerbate the problem.

Finally, accurate awareness of the health implications among the public may facilitate destigmatizing the issue and promoting help-seeking behavior⁹⁷. When people view loneliness and isolation as a personal rather than a health issue, they may be less inclined to seek support and resources to address these challenges. Aligning public perception with the evidence on the importance of social connection is essential to shaping effective policies, nurturing more connected and supportive communities, and promoting health.

A common language

Given the array of terms used in the scientific literature, one potential barrier to prioritization within health settings is lack of precision in terminology. It is clear that we need a common language, but the term “loneliness” may fall short. Loneliness is often used as a catch-all term outside academic scientific contexts, but it is defined and measured more narrowly within the scientific literature. While definitions of loneliness vary somewhat, there is broad consensus that it is distinct from social isolation^{7, 99}.

Loneliness is a subjective, unpleasant feeling based on the discrepancy between one's desired and actual level of social connection¹⁰⁰. It is most often distinguished from social isolation as a separate but related construct⁷. While isolation and loneliness can coexist, they differ in meaningful ways. Social isolation is objectively being alone, having few relationships or infrequent social contact. Thus, social isolation is objective, while loneliness is subjective. Although both social isolation and loneliness can be involuntary, isolation may be chosen¹⁰¹. Both are indicators of lacking social connection, but there are many indicators of social connection and, thus, many indicators of social connection deficits⁸. Social disconnection and loneliness are not equivalent⁴³, and this has implications for measurement and assessment, intervention, policy, and more.

Across scientific disciplines, several constructs have emerged as relevant. Table 2, although not comprehensive, highlights some of the most widely used terms represented in the research and identified in the US Surgeon General Advisory⁴. Pinning down definitions is challenging, given that the same term has been used to refer to different things, while different terms are used to describe the same thing among studies. Some terms, such as social capital, lack a clear consensus on definition^{102, 103}.

Why is this important? These terms refer to related but distinct constructs. Reviews of this evidence find that these measures are not highly correlated empirically^{8, 104}. Thus, when we only measure one of these, we cannot assume that we are capturing the full scope of how social factors influence health.

We need a common language. “Social connection” has been offered as an umbrella term to encompass these distinct but related terms^{4, 8, 15, 105}. From this perspective, the myriad of diverse concepts in the scientific literature can be organized into three key themes or components: structure, function and quality. The first component, structure, represents the human need to have others in our life and is often measured by the size and variability of relationships within a network, being part of groups, and regular social interactions. It is the foundation upon which the other components of social connection are built. The second component, function, recognizes that these connections serve essential functions or purposes. Namely, connections can be relied upon for support to meet various needs and goals. Functions are often measured by the interchange of support that is received or perceived to be available, which can be emotional, informational or tangible, and can help us navigate life's challenges. The bulk of the studies within the scientific literature have primarily examined indicators of these structural or functional components. However, a growing body of research is assessing and recognizing the importance of the quality of social relationships, networks and interactions. Thus, the third component, quality, refers to our connections' positive and negative aspects. High quality is often measured by the level of satisfaction or intimacy, whereas low quality includes social negativity such as conflict, strain or ambivalence.

While Figure 3 is helpful in identifying these core conceptual themes, individual measurement approaches may overlap to some degree between social connection components. Furthermore, specific assessment tools may appear to align clearly within one component, but contain items that overlap with other components¹⁰⁴. Generally, high levels of each of these components have been linked to better health and lower levels of poorer health. To more comprehensively understand underlying causes, predict outcomes, and measure risk, we need to consider the distinct contributions of the structure, function and quality of social connection.

Data across multiple scientific disciplines have linked various social connection indicators to health outcomes⁸. Strong structure, function and quality of social connection may be considered optimal for health. On the contrary, when all three are low, this would be associated with high to severe risk. However, there may be unevenness in the extent to which any individual experiences the three components of social connection. The descriptions in Table 3 help illustrate the disaggregation across these components and their relation to various risk profiles. Nonetheless, there is likely further complexity of risk, given that many indicators within each component of social connection are on a continuum and may have synergistic effects. For example, longitudinal data from nearly half a million people, followed for an average of 12.6 years, demonstrated that low levels on both structural and functional indicators of social connection resulted in a significantly higher risk for cardiovascular disease mortality (hazard ratio, HR=1.63), compared to low levels on structural (HR=1.27) or functional (HR=1.17) components alone⁵⁴.

Conceptually, loneliness may represent the signal or symptom of unmet social needs. However, loneliness does not represent low levels across all three social connection components. Comparisons demonstrate these distinctions. For example, meta-analyses that establish the effect size for the aggregate measures of social connection on mortality were significantly larger than the effect size for loneliness^{43, 105}. Thus, loneliness is not the same as lacking social connection.

Measurement

Another challenge related to the need for a common language is the lack of consensus on measurement¹⁰⁴. The most widely used measurement tools have helped to build a robust evidence base, but may have limitations when applied to other settings. For instance, most measurement tools were developed in Western countries, prior to the widespread shift to digital and remote means of socializing. Measurement not only needs to be predictive of the outcomes of interest, but must also be feasible to use. Notably, what is feasible may differ in different contexts, such as research, clinical settings, population surveillance, and evaluation of the effectiveness of interventions. Currently there is no measure that is multi-factorial, validated, and feasible or adapted to become feasible across settings.

Given the multi-factorial conceptualization of social connection^{7, 80}, a considerable challenge is developing a feasible multi-factorial measure. Not all social connection components are typically assessed, because this would take too much time. Due to time and space constraints, assessments in medical settings and population surveillance may only assess one indicator; however, this approach will likely result in risk assessment errors. For example, if an individual is assessed on an indicator of the structural component of social connection (e.g., frequency of social contact) and found to have high levels, we may assume that this person is at low risk. However, this assessment may miss low levels on the other two components (e.g., low social support, poor quality relationships), which may put the individual at risk. Similarly, we might assume that this person is at low risk if scores are low on an assessment of loneliness, yet the person may have little or no social contact with others. Furthermore, if an individual is low on one component, we may be missing potential protection associated with high levels on the other components. Thus, one's overall social risk profile may be incomplete because of the limited scope of assessments.

Effective intervention and prevention strategies

The next major challenge is reducing risk through effective intervention and prevention strategies. Social connection is complex, with various factors contributing to its increase or decrease, directly and indirectly⁸. Generally, social connection occurs naturally among individuals and within communities. However, when it does not, intervention becomes necessary to reduce risk. Direct actions, programs or initiatives can be implemented to increase social connection or decrease forms of social disconnection intentionally.

Key challenges include: a) the capacity to develop and evaluate intervention strategies; b) the difficulties to understand what works best for whom in what context; and c) the limited scope of existing strategies, and the need to ensure the full scope of social connection across the socio-ecological model, sectors of society, and life course.

Targeted vs. broad approaches

Another major challenge is whether to focus solutions on people most severely affected or broadly on the population. When social connection needs are not met, the mental and physical health consequences are broadly found across age and other demographics. However, isolation and loneliness are unequally distributed across the population. Groups that experience marginalization – i.e., lesbian, gay, bisexual, transgender, queer and/or questioning (LGBTQ+) people, racial minorities, migrants, those with disabilities⁹¹ – and life circumstances that may or may not co-occur with aging (e.g., functional or sensory impairments) are disproportionally affected. Thus, a significant challenge is whether to focus efforts more broadly or these groups.

One perspective is that prioritization should be given to the most vulnerable populations and addressing their needs. By focusing on specific populations, such as marginalized or underserved communities, efforts can be directed toward reducing health disparities and ensuring that resources reach those who need them the most¹⁴⁵. Many sources recommend tailoring interventions to address specific needs, barriers and enablers within these groups, in order to increase the effectiveness of these interventions^{15, 146}. This approach may be a more efficient use of limited resources, funding and personnel. Moreover, it is easier to measure the impact of targeted approaches, as they are narrowly defined and address a smaller population. However, identifying and targeting people “at-risk” may potentially pathologize and stigmatize such groups further and place the burden of change on the individual. Thus, it has been argued that we should focus on the factors that put people at risk instead of group membership¹⁴⁷.

Another perspective is that we should focus efforts across the population to have larger shifts, rather than just targeting a small portion of the population⁷⁷. Broad approaches can lead to systemic changes in policy, environment, and societal norms, laying the foundation for long-term health improvements. Implementing broad interventions might also benefit from economies of scale, reducing the cost per individual reached compared to targeted interventions.

Both targeted and broad approaches are necessary, starting with broad measures to address general issues, while using targeted interventions to address specific needs within the population. However, targeted approaches should be focused on the factors associated with risk (e.g., marginalization) rather than group membership, to avoid further stigmatization. Universal approaches may help prevent social disconnection, whereas more targeted approaches may be needed for those who are already isolated, lonely, or socially disconnected in other ways for prolonged periods or at severe levels. A hybrid strategy can leverage the strengths of both approaches to maximize public health outcomes.

Limited scope of existing approaches

Despite the growing body of research focused on interventions, the scope of solutions is limited in several ways. The Systemic approach Of Cross-sector Integration and Action across the Lifespan (SOCIAL) framework points to gaps and opportunities in solutions across the socio-ecological model, sectors of society, the life course, and prevention⁸⁰.

Evidence points to underlying root causes across the socio-ecological model (e.g., individual, interpersonal, community, institution, society)⁸, yet most interventions are being deployed at the individual level¹⁴⁸. A scoping review of interventions for older adults, including evidence from 30 countries, found that the majority of interventions only measured loneliness, and only three societal-level interventions were found¹⁴⁹.

The health care sector, including both clinical and community health settings, is most often the target of interventions and programs. However, we need to expand our approaches across sectors to engage the whole of society. No one sector of society is likely to be able to address this issue. The “Social in All Policy” approach¹³⁸ recognizes the health and social implications across sectors and “systematically takes into account the health implications of decisions, seeks synergies, and avoids harmful health impacts in order to improve population health and health equity”¹⁵⁰.

Social connection is vital at every stage of life, yet most solutions are focused later in life^{111, 151}. The evidence of a dose-response effect of social connection on biomarkers of health across stages of life⁴⁶, and the importance of early social environment⁵, highlights the need for efforts to address this issue across life. Social isolation during childhood, for instance, is associated with increased cardiovascular risk factors in adulthood, such as increased blood glucose levels, high blood pressure, and obesity¹⁵².

As is the case with most health issues, primary, secondary and tertiary prevention approaches are needed to address social isolation and loneliness. Unfortunately, few interventions focus on helping the society become more socially connected. Prevention efforts may have many longer-term benefits, such as avoiding costly interventions later, reducing disease burden, and improving quality of life¹⁵³.

Efforts to gather and synthesize data, and to identify evidence gaps, are underway. These and similar efforts aim to help create centralized resources to single out evidence-based interventions effective in reducing social isolation and loneliness, or increasing social connectedness. However, without sustained funding, there will be difficulties to evaluate the evidence supporting their effectiveness.

Long-term implications of the COVID-19 pandemic

During the COVID-19 pandemic, social developmental processes were significantly disrupted, with potentially critical long-term health implications. Socialization during early life plays a crucial role in shaping a child's development and long-term health¹⁵⁴. Early socialization provides the foundation for healthy relationships (e.g., learning to communicate, cooperate, negotiate, share), emotional well-being (e.g., learning empathy, expressing emotions), cognitive development (e.g., information processing, critical thinking, problem-solving), self-identity (e.g., self-confidence, self-esteem), ability to navigate cultural norms and values, and overall physical health¹⁵⁵.

Early life experiences with caregivers, peers, schools and communities are all key contributors to the early social environment that ensures longer-term well-being and survival^156-160. However, the pandemic severely limited socialization for roughly three years, during this critical developmental period, for infants, young children and adolescents. Since it is well documented that early social experiences significantly predict later social, mental and physical health^{136, 161, 162}, the long-term health implications of the pandemic may be disproportionately borne for an entire generation.

There may also be longer-term consequences among adults, due to widespread behavioral adaptation that may be sustained over time. The massive adoption and implementation of tools to cope with reduced social contact (e.g., remote work, contactless delivery services, streaming entertainment services, telehealth, automation) came with significant advantages, including increased flexibility, autonomy, convenience, safety, and in some cases cost-effectiveness^163-167.

These advantages have led to preferences that may limit our social contact – particularly with co-workers and weak-ties. Reductions in social contact with both may be critical factors for future health, given the significance of workplace relationships^{168, 169} and the evidence on the importance of weak-ties^{170, 171}. While these preferences are obviously not universal, a large portion of the population values such conveniences. Even if they are not preferred, they are often incentivized by lower costs¹⁶⁶. For example, roughly half of patients preferred in-person visits and half preferred a video visit, but 23.5% switched to a video visit if the cost was lower¹⁷². What was once initiated or scaled to help us cope with isolation, is now what may be reinforcing isolation, with potential long-term implications for exacerbating existing levels of social disconnection and corresponding health consequences.

The widespread behavioral adaptation to spending more time alone, or not leaving the house, may contribute to a societal shift that normalizes social isolation. This is increasingly being depicted in cultural narratives of a “social-battery” that is drained by socializing, and the benefits of “self-care”, “me-time”, and solitude. While there is evidence of some benefits of solitude¹⁷³, the evidence is quite heterogeneous, and benefits appear limited to short-term bouts of solitude, not chronic time alone^{174, 175}. There is instead – as we have seen – robust empirical evidence of the harmful effects of social isolation on mental and physical health outcomes, and increased risk for premature mortality. If time spent alone is praised and encouraged, while the risks of social isolation are diminished within public discourse, the consequences to health are likely to be magnified.

The pervasive experience of social isolation and loneliness during the pandemic also fueled self-proclaimed “experts” and influencers who pushed common-sense approaches to wellness, and in some cases misinformation^{176, 177}. Coupled with a growing distrust in institutions, including science, this may lead to confusion on what is credible. When local community organizations and the general public are skeptical or distrust science, government, and each other, this may increase social disconnection and stifle the development and acceptance of effective interventions and programs to reduce isolation and loneliness.

Technological advancements

The rapid evolution of digital technologies has already demonstrated co-occurring shifts in socializing. Much has been written about the mental and physical health implications of the use of social media^{92, 178-183}. Similar rapid developments and widespread adoption are occurring with artificial intelligence (AI) and large language models (LLM) tools, that have the potential to similarly result in both benefits and harms, but to an exponentially larger extent.

The long-term consequences of AI are yet unknown, but predictions often fall into either a utopian or dystopian outlook for the future, both of which have implications for social health and policy¹⁸⁴. From a utopian perspective, AI will be the cure to the “loneliness epidemic”, with 24/7 access to emotional support for all, and increased automation will free up more time for leisure. From a dystopian perspective, AI will replace human interaction, and diminish trust in others due to blurred lines between fact and fantasy. Moreover, several jobs will no longer require humans, resulting in a lack of meaning and purpose in life, and perhaps even the risk of a downfall of humanity altogether¹⁸⁵.

Some of the potential short-term benefits that already have some limited evidence involve AI-powered virtual companions or chatbots that can engage in conversation and provide immediate emotional support¹⁸⁶, and 24/7 access to mental health support as well as increased accessibility among those experiencing barriers (e.g., language, privacy concerns, social anxiety) to in-person therapy. However, increased isolation may occur if there is over-reliance on AI interaction and emotional support and forming attachments to AI companions^{187, 188}. Generative AI may also magnify our own biases, leading to information echo chambers that further isolate us from others^{189, 190}.

Generative AI is neither inherently good or bad for health and humanity. Our current decisions and actions will starkly impact the trajectory of our future, extending across all sectors of society¹⁸⁴.

Recommendations to reverse trends

Several countries are beginning to take steps to promote social connection, and the global COVID-19 pandemic crystalized and accelerated the urgency to act and to coordinate efforts. Table 4 provides a set of recommendations for national strategies to foster social connection and address social isolation and loneliness.

These recommendations align with those made by the National Academy of Sciences, Engineering, and Medicine^{15, 27, 92}, expert consensus documents and reports, the US Surgeon General Advisory⁴, the WHO¹⁹¹, the US Centers for Disease Control and Prevention¹⁹², the American Heart Association²⁸, and national organ-izations or groups of states (e.g., European Union, UK, Australia, Japan)^193-195.