Blind spots in health perception: the underestimated role of social connection for health outcomes

Abstract

Background

Robust evidence indicates that having few or poor-quality social connections is associated with poorer physical health outcomes and risk for earlier death (Snyder-Mackler N, Science 368, 2020; Vila J, Front Psychol 12:717164, 2021).

Aim

This study sought to determine whether recent attention on social connection and loneliness brought on by the COVID-19 pandemic may influence risk perception and whether these perceptions were heightened among those who are lonely.

Methods

Two waves of online survey data were collected. The first included data from 1,486 English-speaking respondents in the US, UK, and Australia, and a second sample of 999 nationally representative US adults, with a final sample of 2392 respondents from the US and UK.

Results

Perceptions of risk have remained consistent, underestimating the influence of social factors on health outcomes and longevity, even among respondents who reported moderate-to-severe levels of loneliness.

Conclusions

Despite heightened awareness and discourse during the COVID-19 pandemic, public perception in the US and UK continues to significantly underestimate the impact of social factors on physical health and mortality. This underestimation persists regardless of individual loneliness levels, underscoring the need for enhanced public education and policy efforts to recognize social connection as a crucial determinant of health outcomes.

Introduction

Lifestyle factors such as smoking cigarettes, physical inactivity, and inadequate sleep are widely recognized as risk factors for poorer health and have been met with national campaigns to help people stop smoking and national health guidelines to promote physical activity and adequate sleep. Social connection is widely recognized as a significant contributor to emotional well-being [1], but it is less clear whether the public recognizes it as important to physical health.

Across the life course, social factors are among the strongest predictors of morbidity and mortality among social species, including humans, with experimental studies demonstrating that social interactions can causally alter physiology, disease risk, and life span itself [2]. Lacking social connection, whether through social isolation, loneliness, or low social support, has been linked to increased risk for cardiovascular disease [3], stroke [4], type 2 diabetes [5], dementia [6], susceptibility to cold and flu viruses [7], and several other medically relevant health outcomes [8, 9]. Robust empirical evidence indicates that having fewer and lower-quality social relationships has been associated with poorer physical health outcomes and higher risk for earlier death, while having a greater number of and higher quality (positive) relationships has been associated with more positive physical health outcomes and higher overall odds of surviving. [10–14].

Prospective epidemiological studies from both community populations and patient samples examine mortality rates as a function of social relationships. In a seminal meta-analysis of 148 prospective studies, greater social connection was associated with a 50% increased odds of survival [14]. Given little public recognition of the health relevance of social connection, this effect was benchmarked against the effect sizes reported in other published meta-analyses for other established behavioral health risk factors (e.g., smoking, excessive alcohol consumption, physical inactivity, obesity, air pollution). The magnitude of the effect of social support and social integration (markers of social connection) were on par with the effects of these well-established risk factors at the time of the 2010 meta-analysis [13–15]. Since then, multiple meta-analyses have replicated the protective effect of social connection and increased risk for earlier death when social connection is lacking—though effect sizes vary across various indicators of social connection [12, 16].

To determine whether there was public awareness of the relevance of social connection to risk for mortality, in 2018, Haslam and colleagues published data on 502 adults from the US and UK, demonstrating an underestimation of the impact of social connection on physical health outcomes compared to other well-known behavioral health factors [17]. At that time, there was already a large body of research demonstrating robust associations between social connection and morbidity and mortality [18], and studies documenting the prevalence of loneliness and social isolation across the globe [19, 20]. Despite the scientific evidence, people underestimated the impact of social factors on mortality. Since this study was published in 2018, significant global changes have occurred—most notably widespread enforced isolation during the global COVID-19 pandemic– potentially raising awareness of the importance of social connection for well-being and health.

Potential influences on awareness

The COVID-19 pandemic anecdotally seems to have heightened public attention on social isolation and loneliness. Social distancing recommendations and enforcement occurred globally, leading to widespread social disruption, economic loss, and general hardship during the more than three years in which the World Health Organization declared a Global Public Health Emergency. Furthermore, in an online survey of around 4,000 respondents representing the US general population, 49% of respondents reported that “COVID-19 has made me more aware of my physical health” [21]. However, it is unclear whether the public perception has linked the experience of isolation or loneliness to physical health or simply emotional suffering.

Increased personal exposure to isolation or loneliness may influence perceptions of risk. Even before the pandemic, many markers of social connection, such as time spent with family and friends and group participation, had been declining, while perceptions of loneliness and time spent in isolation increased [22]. Thus, more people may be experiencing or personally exposed to lower social connection. Brief periods of isolation can accentuate the significance of social connections, akin to how hunger highlights the need for food [23, 24]. Because positive social relationships may be as fundamental as other basic human needs like food and sleep [25, 26], lonelier individuals may be more sensitive and therefore place a higher value on social connections for their health than less lonely individuals. Thus, it is possible that if more people are experiencing isolation and loneliness, lonelier individuals may place higher value on the impact of social factors on health risks or beneficial health outcomes.

Over the past several years, several formalized efforts have also been made to increase awareness around social connection, isolation, and loneliness. For example, in the US, the CDC published a report entitled “Loneliness and Social Isolation Linked to Serious Health Conditions” and the 2020 NASEM report entitled “The Health and Medical Dimensions of Social Isolation and Loneliness in Older Adults” [18]. In the UK, a Minister for Loneliness was created to address the issue of social isolation and loneliness [27]. The UK has launched national campaigns, created a loneliness helpline, made funds available for research, and even published an annual report on loneliness in the UK [28]. Such formalized public health efforts presumably would also influence public perception.

Importance of awareness

Awareness of social health factors, such as connection, isolation, and loneliness, is an important outcome to consider given awareness is a critical precursor to behavior change. Although awareness alone is insufficient, several theoretical models, including the Health Belief Model (HBM) and the Stages of Change Model, underscore the importance of awareness in motivating health-related actions [29, 30]. Similarly, awareness-raising interventions targeting breast cancer screening or smoking cessation have effectively employed strategies to promote behavior change [31]. Thus, awareness of the potential personal health risks associated with social connections (e.g., public health campaigns, health education, medical screening) could encourage individuals to make behavioral changes, such as prioritizing relationships and social engagement.

Aims of the study

With changes in society, potential increases in personal experience or exposure, and public awareness campaigns focused on social isolation and loneliness [17], this study aims to determine whether the public still underestimates the importance of social connection for health and longevity. The secondary aim is to examine whether lonely individuals perceive social connections as more crucial, given their personal experience.

Preliminary content analysis

To examine whether there is evidence of potential changes in public awareness of social isolation and loneliness, we examined change in public discourse via an online content analysis. To do this, we first used Google Trends software trends for “social isolation,” “loneliness,” and “social distancing.” Further, we performed a content analysis using an industry-leading content analysis tool called BuzzSumo.com that allows market researchers to use their database (of over 8 billion articles, 3 million influencers, and an index of over 300 trillion digital engagements) to find and analyze the most engaging and popular articles, blogs, and social media content across the internet. Examining the trends across the search terms “loneliness” and “social isolation.”

Global Google trends showed the search trends for “social isolation,” “loneliness,” and “social distancing” as having increased in volume between April 2020 and January 2021, with “social isolation” and “social distancing” showing the most drastic spikes in interest [32] (See Fig. 1).

Fig. 1.

Google Trends topical analysis from January 2017 to January 2023 for search terms “loneliness,” “social isolation,” and “social distancing.” Note the spike in popularity of search terms in March 2020. The y-axis represents search interest relative to the highest point on the chart for the given region and time, with 100 as the highest popularity for the term. A value of 50 means that the term is half as popular. A 0 indicates there was not enough data for this term. Replicate these data at https://trends.google.com

The BuzzSumo.com online content engagement trends across the search terms “loneliness” and “social isolation” over the past five years resulted in a similar pattern to the global Google search trends. These trends revealed a spike in online articles published with accompanying social engagement based on shares, saves, blog backlinks (like receiving a citation), likes, and comments on popular social media platforms such as Facebook, Twitter, Reddit, Pinterest, and YouTube. Articles, blogs, and social media content containing “Social isolation” showed the most significant spike in March 2020, when COVID-19 was declared a global pandemic. Articles went from 207 published in February 2020 with 24,687 digital engagements to 3,375 articles published and 1,307,136 digital engagements in March 2020 (see Fig. 2). Although engagement with the content containing “loneliness” showed a similar spike, both the amount of content and engagement with the content were high even prior to the pandemic and have continued.

Fig. 2.

BuzzSumo content analysis. (Left) BuzzSumo.com content analysis results for the search topic “social isolation” for the last five years show the number of mainstream media articles published and total engagements with these articles in the public discourse. Note the exponential spike between February and March of 2020. (Right): BuzzSumo.com content analysis results for the search topic “loneliness” for the last five years show the number of mainstream media articles published and total engagements with these articles in the public discourse

Although no more than five years of data is available, these preliminary findings indicate significant public discourse focused on social isolation and loneliness, with spikes suggesting potential increased awareness due to the pandemic. Nonetheless, this data does not tell us whether this discourse has changed public perception relevant to health and longevity.

Methods

Study design & participants

We collected data in a multi-country population-based survey primarily designed to estimate risk perceptions among health factors in a cross-sectional study design. Participants were recruited via the online survey portal Prolific.co. To expand upon prior research [17], we recruited 1,500 participants, including respondents from the United States (US), the United Kingdom (UK), and Australia, balanced for gender. The resulting sample was not equally representative of all countries, and the sample in Australia was too small to compare to the US and UK samples; therefore, Australian participants (n = 93) were dropped from the dataset, and an additional sample was recruited through the same platform to acquire a more diverse and nationally representative sample of US adults. The final combined sample (n = 2,392) consisted of 1,392 (58.2%) from the US and 1,000 (41.8%) from the UK. Sample 1 was drawn from approximately 90,000 active subjects who fit the demographics indicated in the filter. Sample 2 was taken from an eligible pool of 34,038 to qualify as a US nationally representative sample. The pattern of findings did not differ between the UK and US, so the data is presented combined.

Prolific.co, a fee-for-service panel, was used for data collection. Prolific attracts diverse samples across the United States and globally and has been found satisfactory in collecting high-quality data from nationally representative samples despite some drawbacks [33, 34]. The survey opportunity was listed to prospective subjects on Prolific.co as “How do health factors impact survival?” and did not use manipulations. Two attention checks were used within the survey, and responses from respondents who failed either check were excluded from the analysis. All responses were anonymous, and the dataset has been made available on Open Science Framework at https://bit.ly/blindedOSF for further analysis, replication, or critique.

Data collection

Consistent with prior research, we used a series of ranking questions to evaluate perceptions of decreased risk associated with social connection relative to other health-related factors. Among these, participants were instructed to “rank the following protective factors in terms of their importance as predictors of reduced mortality risk (living longer).” The following factors were included in the rankings: availability of social support, being socially connected, never smoking, quitting smoking, moderate alcohol consumption (< 2 drinks per day), flu vaccination, being physically active, healthy weight (BMI = 18.5–24.9), and medication adherence. These mirrored the items in the previous 2018 study, except we combined "Exercise" and "Physical Activity" to reduce survey fatigue. Participants were also asked to estimate the potential impact of each factor on an average person's life expectancy, ranging from "no effect" to “ > 7 years”. The order in which the factors were presented was randomized for each participant to minimize priming and order effects.

Going beyond previous research, additional ranking items were included to assess the perception of risk for adverse health outcomes and the perceived contribution to disease based on the same determinants of health. Participants were also asked, “To what extent does one’s level of social connectedness (loneliness, isolation, social support) influence the following” health outcomes, including mental health, physical health, chronic disease, quality of life, happiness, and cognitive impairment (0 = no influence, 10 = a significant influence).

To assess loneliness, we used the 3-Item UCLA Loneliness Scale [35]. This validated short version is strongly correlated with the longer R-UCLA (0.82, p < 0.001) and demonstrated an alpha reliability coefficient of 0.72.

We obtained demographic data, including age, gender, sex, sexual orientation, and education level attained. Consistent with prior research [17], we obtained self-reported health using a single item. This item is a robust predictor of mortality, even after adjusting for key covariates such as functional status, depression, and co-morbidities, as individuals who rated their health as "poor" had a two-fold higher risk of mortality compared to those who reported "excellent" self-rated health [36].

The average time for respondents to complete the survey was approximately 8.1 min. The entire survey instrument can be found on the Open Science Framework (OSF) project page: https://bit.ly/blindedOSF. The Institutional Review Board at the author’s university approved this study.

Analysis

Prior to analysis, we checked the data for responses from respondents who failed attention checks and duplicate entries, and both were removed. Descriptive statistics for the survey items were analyzed using the Stata and R statistical software packages.

To compare the rankings of the combined samples to those published in 2018, we created a rank score based on a marginal rank frequency analysis, which looked at the frequency of occurrence of the 1 – 9 rankings across the sample; each of the nine rank positions being counted and summed, per factor. A count formula was created using an R script that gave each factor points based on the frequency count. This point score was then sorted across the factors from low to high and plotted with the order (low to high) of the odds ratios of survival previously reported by Haslam, obtained from a 2010 meta-analysis [14]. A combined rank score of each factor in the other rank items was created. The plots of these ranking comparisons can be found in Fig. 3.

Fig. 3.

Rankings from the Haslam et al. [17] study. “Ranked Effect Size” refers to the order of magnitude of Odds Ratios for survival based on the Holt-Lunstad et al. [14] meta-analysis. “Perceived rank” is the composite rank by participants collected in June 2022 & January 2023. Text of item: “Rank the following protective factors in terms of their importance as predictors of reduced mortality risk (living longer). 1 = most important, 9 = least important.” The result from the US nationally representative sample was identical to the above other than that “Social Support” was in position 7 and “Social Integration” was in position 8

To investigate whether lonelier individuals may have a heightened sensitivity to or awareness of social factors, the same analysis was performed among a subgroup of those scoring higher on the 3-item UCLA Loneliness Scale (indicating increased loneliness). A composite score was generated across the three items following the protocol outlined by Hughes et al. [35]. A score of 6 or higher indicated a higher perceived level of loneliness. The same scoring method was applied to this subgroup, akin to the analysis of the factor ranking item.

Exploratory Comparisons

Given the comparison effect sizes used in the 2018 risk perception study were more than a decade old, we explored the possibility that individuals may be considering more recent evidence in their rankings. Thus, we identified more up-to-date effect size estimates for the health comparisons. To obtain more recent comparison effect sizes, we searched APA PsychInfo, Medline (PubMed), and CINAHL databases to identify meta-analytic data published after 2010 for each health factor used to benchmark social connection in earlier research.

Results

Preliminary analyses were conducted to establish the descriptive data on our sample. A total of 2,392 participants completed the survey. The mean age was 41.1 (range 18–85) years old. See Table 1 for a more detailed breakdown of sex, education, sexual orientation, and gender identity.

Table 1.

Demographics of sample for education, sexual orientation, and gender identity

Demographics
Education	Frequency	Percent
Less than high school degree	33	1.3
High school graduate (high school diploma)	359	15
Associate degree in college (2-year)	236	9.9
Some college but no degree	497	20.8
Bachelor's degree in college (4-year)	895	37.5
Master's degree	281	11.8
Doctoral degree	35	1.4
Professional degree (JD, MD)	45	1.8
Sexual Orientation	Frequency	Percent
Asexual	29	1.2
Bisexual	171	7.12
Gay	59	2.45
Lesbian	42	1.75
Other (Please Specify)	12	0.5
Pansexual	41	1.7
Prefer not to say	36	1.5
Straight	2010	83.75
Gender Identity	Frequency	Percent
Man	1153	48.08
Other (Please Specify)	43	1.79
Prefer not to say	10	0.41
Transgender man/Trans man	7	0.29
Transgender woman/Trans woman	5	0.2
Woman	1180	49.2

Nearly a quarter (24.8%) of the sample reported not having good health (either “Poor” or “Fair” as opposed to “Good,” “Very good” or “Excellent”). The distribution of loneliness scores showed that 47.9% of the total sample reported a score of 6 or higher, indicating moderate levels of loneliness, whereas 25% reported severe loneliness scores (7—9). See Table 2 for a breakdown of scores and percentages of the entire sample.

Table 2.

A score of 3 would indicate that respondents do not perceive themselves as lonely. A score of 5 indicates that in two or more of the items they indicated that they felt lonely some of the time. A score of 7 or above indicates that at least two of the three answers indicated high levels of loneliness or that they see themselves as lacking social support “often.” A score of 9 indicates all three questions were responded to as “often” when asking about a lack of social support and belonging (see actual items on the OSF page referenced)

Three item loneliness scale scores
Scores	Frequency	Percent
3	704	29.35
4	254	10.59
5	291	12.13
6	548	22.85
7	183	7.63
8	152	6.33
9	266	11.09
Loneliness ≥ 6
UK	490	49.04
USa	654	47.01
Total

^aNationally representative sample

Primary aim

To examine risk perception across determinants of health, consistent with methods used in prior research [17], the composite frequency of rankings across all respondents was plotted together with rankings from 2018 and effect sizes reported previously (see Fig. 3). As can be seen in the graph, participants continue to underestimate the importance of social determinants of health, as is evident by social support and social integration being ranked among the lowest (8 and 9 out of 10) factors contributing to reduced mortality risk (increased life expectancy). Estimates of the number of years of life added for social connection relative to other determinants of health also demonstrated that the perceived importance of social support and social integration was low and an underestimate from effect size estimates reported in meta-analyses (see Fig. 4).

Fig. 4.

2023 estimates were ordered based on estimates of 0–7 + years of life added and then ordered based on composite estimates in descending order. This scoring was used to give a ranking of estimates from most years added to least years added compared to the same ranked effect sizes for reduced mortality from Holt-Lunstad and colleagues’ 2010 meta-analysis. 2018 estimates were added in descending order of the perceived composite number of years added

To examine whether risk perceptions varied if they were framed differently, the average frequency rankings of determinants of health were plotted against previously reported effect sizes as predictors of increased mortality risk (decreased life expectancy) (See Fig. 5). Again, participants ranked social determinants of health, “social isolation,” “loneliness,” and “low social support” as among the lowest in importance (7, 8, and 9, respectively, out of 9). To determine risk perception among determinants of health for chronic diseases, we found that “social isolation,” “loneliness,” and “low or no social support” were ranked 8, 9, and 10, respectively, out of 11 factors (see Table 3 below).

Fig. 5.

Composite rankings from the item stating: “Rank the following risk factors in terms of their importance as predictors of earlier mortality (shorter lifespan).” Ranked effect sizes are in order of effect based on a 2010 meta-analysis from Holt-Lunstad et al. [14]

Table 3.

When asked, “To what extent do the following factors negatively contribute to chronic disease outcomes?” (0 = no contribution, 10 = a significant contribution). This table represents the order of composite scores given to each factor. All factor scores were summed and ordered according to perceived contribution

Perceived rank of negative impact on chronic disease outcomes
Factors	Rank
Smoking	1
Obesity (B.M.I. > 30)	2
Not quitting smoking	3
Excessive alcohol consumption	4
Physical inactivity	5
Medication non-adherence	6
Exposure to air pollution	7
Social isolation	8
Loneliness	9
Low or no social support	10
No flu vaccination	11

Ranking of the influence of social connectedness (loneliness, isolation, social support; 0 = no influence, 10 = a significant influence) on different types of health and well-being outcomes, “mental health” (depression, anxiety, etc.) and “emotional well-being” (happiness) were ranked highest and “physical health” and “chronic disease” as least impacted (see Table 4).

Table 4.

A representation of composite perception from the responses to the question: “To what extent does one's level of social connectedness (loneliness, isolation, social support) influence the following?” (0 = no influence, 10 = a significant influence). Order: 1 = perceived as most influential, 7 = perceived as least influential

Perceived order of influence of social connectedness on outcomes of interest
Factors	Order	Mean rating	SE
Mental Health (depression, anxiety, etc.)	1	8.24	0.04
Emotional wellbeing (Happiness)	2	8.20	0.04
Quality of Life	3	7.88	0.04
Self-care (diet, exercise, sleep, etc.)	4	7.06	0.05
Cognitive Impairment (e.g. Dementia)	5	6.49	0.05
Physical Health	6	6.23	0.05
Chronic disease	7	5.12	0.06

Furthermore, we observed a significant difference in Likert (0–10) ratings of the perceived influence of social connectedness on various health factors. A Welch Two Sample t-test revealed a significant difference in the perceived influence of social connectedness on the average of emotional well-being and average of physical health factors, t = −53.19, df (8557), p < 0.0001. Participants perceived a significantly greater influence of social connectedness on mental health factors (M = 8.22, SE = 0.039 for Mental Health (depression, anxiety, etc.); M = 8.20, SE = 0.038 for Emotional Well-being (Happiness)) compared to physical health factors (M = 6.23, SE = 0.051 for Physical Health; M = 5.12, SE = 0.057 for Chronic Disease). The effect size, as measured by Cohen’s d = 1.09, 95% CI [1.04, 1.13], indicates a large effect. The difference between physical and mental or emotional factors is depicted in Fig. 6.

Fig. 6.

Mean ratings of the importance of social connection for mental/emotional health and Physical health factors

Secondary aim

We evaluated whether individuals reporting higher levels of loneliness would rank social factors higher among determinants of health contributing to reduced mortality risk (increased life expectancy) than the overall population. The frequency test of factor rank order within a subset of the sample reporting moderate to severe loneliness, composite ranking placed social factors in the 6th position (Availability of Social Support) and 7th position (Being Socially Connected) out of 9 positions (see Table 5).

Table 5.

Perceived rankings from a subset of participants who reported having higher loneliness scores (≥ 6)

Rankings of factors contributing to reduced mortality risk among those with loneliness scores ≥ 6
Factors	Rank
Being Physically Active	1
Never Smoking	2
Healthy weight (BMI = 18.5–24.9)	3
Quitting Smoking	4
Medication Adherence	5
Availability of Social Support	6
Being Socially Connected	7
Moderate Alcohol Consumption (< 2 drinks per day)	8
Flu Vaccination	9

The ranking positions remained consistent even when focusing solely on those scoring 7 and above on the UCLA loneliness scale. This consistency persisted even when examining individuals with a score of 9 (severely lonely).

Exploratory Aim

To explore whether more updated health comparison effect sizes might be guiding the public’s perception of health risk, we identified the effect sizes for the health factors included in the 2018 study as well as additional health factors that have grown in public popularity over the last decade. The results of the search can be found in Table 6 below. Because effect sizes obtained from meta-analyses reported in different metrics (odds ratio, risk ratio, hazards ratio) are not equivalent, a combined plot was not possible.

Table 6.

Effect sizes of common risk factors, compared to social factors. OR = Odds Ratio. HR = Hazard Ratio. RR = Relative Risk

Effect sizes for all-cause mortality based on meta-analyses
	Effect Size	Effect Type	[95% CI]
Complex Social Integration [14]	1.91	OR	[1.63–2.23]
Living alone [13]	1.32	OR	[1.14–1.53]
Loneliness [13]	1.26	OR	[1.04–1.53]
Underweight [37]	1.42	HR	[1.29–1.59]
Social Isolation [38]	1.33	HR	[1.26–1.41]
Social Isolation [16]	1.32	HR	[1.26–1.39]
B.M.I. / Healthy Weight [39]	1.18	HR	[1.12–1.25]
Loneliness [16]	1.14	HR	[1.08–1.20]
Cardiac Rehabilitation [40]	0.96	HR	[0.88–1.04]
Daily Step Count [41]	0.88	HR	[0.83–0.93]
Sleep 7–8 h per day [37]	0.87	HR	[0.77–0.96]
Healthy Diet [37]	0.84	HR	[0.81–0.88]
Physical Activity (vs no physical activity) [42]	0.75	HR	[0.67–0.83]
Smoking [43]	1.83	RR	[1.65–2.03]
Alcohol Consumption ≥ 65 g/day [44]	1.32	RR	[1.23–1.41]
Air pollution (household) [45]	1.12	RR	[1.06–1.19]
Smoking cessation [46]	0.82	RR	[0.76–0.90]
Flu Vaccine (women) [47]	0.53	RR	[0.32–0.86]
Flu Vaccine (men) [47]	0.47	RR	[0.32–0.68]

Use of large language models

During the preparation of this work, the authors used ChatGPT to improve the readability and language of small portions within this manuscript. The authors also used ChatGPT to check the logic of statistical analyses and debug code in the R statistical software package. After using this service, the authors reviewed and edited the content and take full responsibility for the content of the publication. When limitations of large language models are avoided or mitigated, artificial intelligence can be used to improve and accelerate scientific research and writing [48, 49].

Discussion

This study examined whether greater awareness efforts around social connection would result in greater recognition as a protective factor for mortality. Although preliminary content analysis confirmed increases in public discourse in the media, participants’ ranking of perceived importance among determinants of health on mortality risk revealed remarkably little change from 2018 to the current study. We found that social connection, a well-established social determinant of health, continues to be underestimated for its impact on reduced mortality risk—with participants ranking it among the lowest in importance. This underestimation of social connection as a protective factor has persisted despite significant media publicity, increases in global search volume, national campaigns in the UK, CDC reports in the US, and heightened attention to the impacts of social isolation and loneliness during the global pandemic, which necessitated physical distancing (commonly referred to as social distancing). This underestimation held regardless of framing the influence on life expectancy (increased risk vs. reduced risk). Given the rankings of other well-recognized risk factors were more aligned with true risk, this data suggests that participants aren’t simply bad at understanding risk but are less aware of the risk associated with social factors.

Going beyond prior research, we further examined the perceived importance of social connection, isolation, and loneliness relative to other health determinants for morbidity, and rankings of the impact of social connectedness (loneliness, isolation, social support) on different health outcomes. Consistent with the mortality rankings, participants underestimate the importance of these social factors relative to other determinants of health and rank them among the lowest in importance. Despite robust evidence that social factors influence physical health and chronic illnesses [50–52], according to the rankings found in this study, the public views social connection, isolation, and loneliness as highly influential for mental and emotional health but significantly less relevant to physical health or chronic disease.

We further examined whether one’s own level of loneliness might influence these perceptions of risk. Consistent with previously reported prevalence rates [53], nearly half of respondents (47.9%) reported experiencing moderate to severe loneliness (scoring above 6). Although evidence suggests loneliness is associated with heightened social monitoring and lonelier individuals may be more sensitive to social information [54, 55], lonelier individuals did not rank social factors as more important for health relative to the overall sample. Even respondents who reported high (a score of 7 or more) and severe loneliness (a score of 9) still underestimated the impact of social connection, loneliness, and social integration on health outcomes. This suggests that even among those who may be most sensitive to social information and most at risk, the risk is underestimated. Thus, regardless of one’s level of loneliness, social connection is ranked low in importance among factors contributing to health and mortality risk.

Given the underestimation of the relevance of social connection to health and risk for mortality, additional education and awareness efforts are needed to explicitly focus on health and mortality outcomes. Future research should also determine how healthcare providers perceive the importance of social factors for health outcomes. The healthcare sector and providers are critical frontline agents for communicating information relevant to health and may provide an additional avenue for raising awareness. Additional research on the effectiveness of government-driven reports and other national campaigns and reports on their potency in changing the perception and awareness of general audiences would also be meritorious.

Limitations

There are some limitations to this study. To have the most direct comparison between the current data and the data reported in 2018 by Haslam and colleagues [17] we used the same health comparisons reported; however, we did not have access to their data, which precluded a precise comparison across the two studies. Nonetheless, the total sample size in the current research is significantly larger than the previous study (N = 2392 vs. N = 502 respectively) and includes a US nationally representative subset, which is a strength of the present study. Although comparisons of perceived rankings to estimated effect sizes from meta-analytic data provide us with the best available estimates of what the actual risk may be, nonetheless, because each estimate is based on aggregate data, it may not reflect absolute risk [56] or a comparison of risks within the same population [57]. Thus, we are cautious in describing these estimates as the “true effect.”

We also identified updated effect sizes for use as comparisons; however, they were not in the same metric—and odds ratios, risk ratios, and hazards ratios may not be directly comparable. While converting to a standard metric is relatively easy if based on a single study, these were obtained from meta-analyses. Thus, a common metric is impossible without re-analyzing each meta-analysis's original data. While the ranking method avoids using precise effect size estimates, which the public may not be aware of, this method may be limited if confidence intervals overlap. Regardless of any true discrepancy between perceived and actual risk, the reported rankings help us better understand the current public perception of health risks.

The marginal rank frequency analysis discussed was an analytical technique that is mathematically sound [58] but could not be used to perform more granular statistical comparisons with ranking scores across participants compared to other participants for cross-tabulation and correlational analysis with statistical significance. Further investigation of rank order studies and statistical techniques may reveal more about relationships between demographics such as sex and gender, and other findings. Given that this study aimed to determine whether individuals underestimate the importance of social factors for health outcomes, the marginal rank frequency analysis was sufficient to determine this.

We further acknowledge that the US and UK data may not represent the global population. While social connection, isolation, and loneliness are global public health issues, much of the evidence and attention to this issue has come from North America, the UK, and Europe. Thus, presumably, awareness of health risks and protection should be higher. Nonetheless, further research is needed to assess risk perception globally.

Conclusion

These findings suggest that public perception in the US and UK continues to underestimate the importance of social factors for health and mortality risk. Despite increased public discourse, formalized national efforts in multiple countries, published reports from national and international organizations, and a global pandemic with sustained and enforced isolation and loneliness –public perception within the US and UK appears to view this as highly relevant to emotional or mental health but of less importance for physical health and mortality. This low perceived health relevance of social connection is consistent regardless of one’s level of loneliness. These findings highlight the continued need to raise public awareness of the robust scientific evidence documenting social connection as an independent determinant of morbidity and mortality. This data may assist with major policy decisions, public awareness and education efforts, healthcare provider education and training, and broader national and global efforts focusing on the importance of social connection for physical health. To be effective, awareness raising must also address perceived barriers, including societal stigmas and the cognitive biases that loneliness exacerbates, and incorporate actionable strategies to mitigate these risks [59, 60].

Given the robust evidence documenting that social factors are a critical determinant of leading health outcomes and survival, this study brings to light the need to further educate and consistently increase awareness on a large scale of the importance of social connection broadly, including community, social support, close and positive relationships, and their impact on health outcomes.

Abbreviations

BMI

Body Mass Index

CDC

Centers for Disease Control and Prevention

COVID-19

Coronavirus Disease 2019

Fmri

Functional Magnetic Resonance Imaging

NASEM

National Academies of Sciences, Engineering, and Medicine

OSF

Open Science Framework

R-UCLA

Revised University of California, Los Angeles Loneliness Scale

UK

United Kingdom

US

United States

Authors’ contributions

J.H.L.: Writing – review & editing, Writing—original draft, Methodology, Conceptualization, Project administration, Supervision, Funding acquisition. A.S.P.: Writing – original draft, Writing-- review & editing, Data curation, Formal analysis, Visualization.

Funding

Funding for this project was obtained from university or departmental sources including the BYU Department of Gerontology and the BYU School of Family Home and Social Sciences.

BYU Department of Gerontology and the BYU School of Family Home and Social Sciences

Data availability

All data generated or analyzed during this study are included in this published article. Anonymized data is available publicly on the Open Science Framework website here: https://osf.io/j425a/. R code to analyze this dataset is also available on this OSF project page and we welcome criticism and replication.

Declarations

Ethics approval and consent to participate

This study was approved by the Brigham Young University Institutional Review Board under IRB2022-183. Informed consent was given by all participants before they were allowed to participate in the survey. This study adhered to the Declaration of Helsinki (http://www.wma.net/en/30publications/10policies/b3/index.html).

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.