Social connection

Studies measure social connection in several ways, and the choice of measure affects the result. In the 2010 meta-analysis, the association with survival was strongest for complex measures of social integration that combined several aspects of relationships (odds ratio, a measure of how much more likely an outcome is in one group, 1.91) and weakest for the simple measure of living alone compared with living with others (1.19, 95% confidence interval 0.99 to 1.44, meaning the result could be due to chance) [1].

Social isolation and loneliness overlap but are not the same. A person can live alone and have frequent contact with others, or be surrounded by people and feel lonely. The 2015 meta-analysis found no difference between objective measures of isolation and subjective measures of loneliness in their association with mortality [2]. In a single large study of older adults in England, however, social isolation remained associated with mortality after adjustment for baseline health while loneliness did not [7].

The 2010 meta-analysis pooled prospective studies (studies that measure relationships first and follow people over time). Across 148 studies the average odds ratio was 1.50 (1.42 to 1.59), and the association was consistent across age, sex, initial health status, cause of death and length of follow-up [1].

The 2015 meta-analysis focused on the absence of relationships. In studies that adjusted for several possible confounders (other factors that could explain the association), social isolation had an odds ratio of 1.29, loneliness 1.26 and living alone 1.32 for mortality. The associations were consistent across sex, length of follow-up and world region, were influenced by initial health status, and were stronger in samples with an average age under 65 than in older samples [2].

The English Longitudinal Study of Ageing measured contact with family and friends, participation in civic organisations and loneliness in 6,500 adults aged 52 and over, and followed them for an average of 7.25 years. Both isolation and loneliness were associated with higher mortality before adjustment. After adjustment for demographic factors and baseline health, the most isolated fifth had a hazard ratio (the risk in one group relative to another) of 1.26 (1.08 to 1.48), while loneliness was no longer associated with mortality (0.92, 0.78 to 1.09). The authors concluded that reducing isolation was likely to be more relevant to mortality than reducing loneliness, although both affect quality of life [7].

Loneliness and isolation are also linked to mental health. Mental disorders are associated with 2.22 times the mortality of the general population [8], and low mood and withdrawal from others can reinforce each other. Mortality in mental disorders is described under mental health conditions and lifespan.

A 2016 meta-analysis of longitudinal studies in high-income countries examined loneliness and social isolation as risk factors for cardiovascular disease. Drawing on 16 data sets with 4,628 coronary heart disease events and 3,002 strokes over 3 to 21 years, it found that poor social relationships were associated with a 29% higher risk of new coronary heart disease (relative risk 1.29, 1.04 to 1.59) and a 32% higher risk of stroke (1.32, 1.04 to 1.68), with no difference between men and women [3].

For dementia, a meta-analysis of 19 longitudinal studies found that low social participation (relative risk 1.41), less frequent social contact (1.57) and more loneliness (1.58) were each associated with higher risk of dementia. Low satisfaction with one's social network was not significantly associated (1.25, 0.96 to 1.62), and results for network size were inconsistent. The authors noted that these associations were comparable in strength with other established dementia risk factors such as low education, physical inactivity and depression in later life [4].

Proposed pathways include the effects of long-term stress on blood pressure and inflammation, which are described under stress and anxiety and inflammation, and the influence of other people on health behaviours and use of medical care.

Marital status is the most commonly recorded relationship measure. A meta-analysis of 53 comparisons in more than 250,000 older adults found a relative risk of death of 0.88 (0.85 to 0.91) for married compared with unmarried people. Compared with married people, relative risks were 1.11 for widowed, 1.16 for divorced or separated and 1.11 for never-married people, and the estimate did not differ by sex or between Europe and North America [6].

Participation in groups adds a further layer. In a meta-analysis of organisational volunteering by adults aged 55 and over, volunteering was associated with 47% lower mortality before adjustment and 24% lower mortality (95% confidence interval 16 to 31%) after adjustment for health and other factors; there was preliminary evidence that the association was stronger among people with higher public religious participation [5]. Group exercise and team games are discussed under sports and active hobbies.

Contact with acquaintances also relates to well-being. In a series of studies, people reported greater happiness and a greater sense of belonging on days when they interacted with more weak ties (acquaintances and peripheral members of their social network) than usual [9]. These studies measured mood, not health or survival. Companionship from animals is described under pet ownership.

A sense of purpose in life is often studied alongside social connection because both concern meaning and engagement with others. In 6,163 US adults in the Midlife in the United States study, each standard deviation higher in purpose was associated with 15% lower mortality over 14 years (hazard ratio 0.85, 0.78 to 0.93), after controlling for other measures of well-being and regardless of age or retirement status [10]. Among 6,985 US adults over 50 in the Health and Retirement Study, the lowest purpose group had 2.43 times the mortality of the highest over four years (95% confidence interval 1.57 to 3.75), with a similar association for deaths from heart, circulatory and blood conditions [11].

Full deep dive: Connection and purpose.

The evidence on social connection and health is observational. People who are ill or disabled often lose social contact, so reverse causation can inflate associations; the 2015 meta-analysis found that initial health status influenced the results [2], and in the English study the association with loneliness disappeared after adjustment for baseline health [7]. Income, education, personality and health behaviours can influence both relationships and survival.

Measures vary widely, from a single question about living alone to multi-item scales of social integration, and the strength of the association depends on the measure used [1]. Studies of perceptions add a further complication: in one US study, the association between stress and mortality depended on whether people believed stress was harming them [12]. Trials that increase social contact and measure mortality have not been done, so the number of years gained by strengthening relationships is not known.