Dr. Vivek Murthy, former Surgeon General of the United States, recently wrote: “Loneliness and weak social connections are associated with a reduction in lifespan similar to that caused by smoking 15 cigarettes a day”.


This ‘15 cigarettes a day’ figure has been reproduced and reported in the news many times, under headlines such as “Loneliness is as lethal as smoking 15 cigarettes per day”.


It is indeed quite a shocking comparison since around 7 million deaths globally are attributed to smoking every year, and back-of-the-envelope calculations published in medical journals say one cigarette reduces your lifespan by 11 minutes.


Here we dig deeper to try to understand what the data and research tell us about the link between social relations and health. In a nutshell, my reading of the evidence is as follows:


  • There is a huge amount of evidence showing individuals who report feelings of loneliness are more likely to have health problems later in their life.
  • There is credible theory and explanation of biological mechanisms, whereby isolation can set off unconscious surveillance for social threat, producing cognitive biases, reducing sleep and affecting hormones.
  • It's very likely there is a causal link, but there is no credible experimental evidence that would allow us to have a precise estimate of the magnitude of the causal effect that loneliness has on key metrics of health, such as life expectancy.
  • The fact that we struggle to pin down the magnitude of the effect of loneliness on health doesn't mean we should dismiss the available evidence. But it does show that more research is needed.


Observational studies: A first look at the data


Measuring loneliness


Psychologists and social neuroscientists often refer to loneliness as painful isolation. The emphasis on painful is there to make a clear distinction between solitude – the state of being alone – and subjective loneliness, which is the distressing feeling that comes from unmet expectations of the types of interpersonal relationships we wish to have.


Researchers use several kinds of data to measure solitude and loneliness. The most common source of data are surveys where people are asked about different aspects of their lives, including whether they live alone, how much time they spend with other people in a given window of time (e.g. ‘last week’) or specific context (e.g. ‘at social events, clubs or places of worship’); and whether they experience feelings of loneliness (e.g. ‘I have no-one with whom I can discuss important matters with’). Researchers sometimes study these survey responses separately, but often they also aggregate them in a composite index.


Surveys confirm that people respond differently to questions about subjective loneliness and physical social isolation, which suggests people do understand these as two distinct issues.


In the chart below I've put together estimates on self-reported feelings of loneliness from various sources. The fact that we see such high levels of loneliness, with substantial divergence across countries, explains why this is an important and active research area. Indeed, there are literally hundreds of papers that have used survey data to explore the link between loneliness, solitude, and health. Below is an overview of what these studies find.


The link between loneliness and physical health


Most papers studying the link between loneliness and health find that both objective solitude (e.g. living alone) and subjective loneliness (e.g. frequent self-reported feelings of loneliness) are correlated with higher morbidity (i.e. illness) and higher mortality (i.e. likelihood of death).


The relationship between health and loneliness can of course go both ways: lonely people may see their health deteriorate with time; but it may also be the case that people who suffer from poor health end up feeling more lonely later down the line.


Because of this two-way relationship it’s important to go beyond cross-sectional correlations and focus on longitudinal studies – these are studies where researchers track the same individuals over time to see if loneliness predicts illness or mortality in the future, after controlling for baseline behaviors and health status.


The evidence from longitudinal studies shows that people who experience loneliness during a period of their life tend to be more likely to have worse health later down the line. In the Netherlands, for example, researchers found that self-reported loneliness among adults aged 55-85 predicted mortality several months later, and this was true after controlling for age, sex, chronic diseases, alcohol use, smoking, self-assessed health condition, and functional limitations.


Most studies focus either on subjective loneliness, or on objective isolation. But some studies try to compare both. In a recent meta-analysis covering 70 longitudinal studies, the authors write: “We found no differences between measures of objective and subjective social isolation. Results remain consistent across gender, length of follow-up, and world region.” And in the concluding section they highlight that, in their interpretation of the evidence, “the risk associated with social isolation and loneliness is comparable with well-established risk factors for mortality”; which include smoking and obesity.


The link between mental health and subjective well-being


In another much-cited review of the evidence, Louise Hawkley and John Cacioppo, two leading experts on this topic, concluded that “perhaps the most striking finding in this literature is the breadth of emotional and cognitive processes and outcomes that seem susceptible to the influence of loneliness”.


Researchers have found that loneliness correlates with subsequent increases in symptoms related to dementiadepression, and many other issues related to mental health; and this holds after controlling for demographic variables, objective social isolation, stress, and baseline levels of cognitive function.


There is also research that suggests a link between loneliness and lower happiness, and I discuss this in more detail in a related post here.




Experiments with social animals, like rats, show that induced isolation can lead to a higher risk of death from cancer. Humans and rats are of course very different; but experts, such as Hawkley and Cacioppo, argue that these experiments are important because they tell us something meaningful about a shared biological mechanism. In a review of the evidence, Susan Pinker writes: “If our big brains evolved to interact, loneliness would be an early warning system - a built-in alarm that sent a biological signal to members who had somehow become separated from the group”.

Effect of long-term social isolation on mammary tumor growth and diagnosis at 15 months. (A) Tumor burden (mean  SEM) after living alone or in noncrowded social groups. (B) Rats with ductal carcinoma in situ (DCIS) or malignant tumors were primarily socially isolated; those with benign or no tumors lived in groups. (C–H) Wide range of naturally developing mammary tumors: (C) fibroadenoma; (D) lactating adenoma; (E) intraductal papilloma; (F) DCIS; (G) invasive ductal carcinoma; (H) fibrosarcoma.

Image: PNAS

Indeed, there’s evidence of social regulation of gene expression in humans: studies suggest perceived loneliness can switch on/off genes that regulate our immune systems, and it is this what then affects the health of humans, or other animals that evolved with similar defence mechanisms.


Causality and implications


The bulk of evidence from observational studies and biological mechanisms, described above, implies that loneliness most likely matters for our health and well being. But do we really know how much it matters relative to other important risk factors?


The key point here is that estimates are likely biased to some extent.


The findings from longitudinal studies that track individuals over time are insightful, but they cannot rule out that the relationship might be partly driven by other factors that we cannot observe. Even the studies linking loneliness and genetics can be subject to this omitted-variable bias, because a genetic predisposition to loneliness may drive both loneliness and health outcomes.


I could not find credible experimental evidence that would allow us to have a precise estimate of the magnitude of the causal effect. But the fact that we struggle to pin down the magnitude of the effect doesn't mean we should dismiss the available evidence. On the contrary – it would be great if we had evidence from randomized control trials that test positive interventions to reduce loneliness, to understand better if the ‘15 cigarettes per day’ comparison from the Surgeon General of the US is roughly correct, at least for the average person.


Having a better understanding of the magnitude of the effect is important, not only because loneliness is common, but also because it’s complex and unequally experienced by people around the world.


As the chart above shows, there are large differences in self-reported loneliness across countries. We should understand how important these differences are for the distribution of health and well-being.


Esteban Ortiz-OspinaSenior Researcher and Content Lead, Our World in Data