What the WHO's 871,000 Deaths Figure Actually Counts

The number circulated fast after the WHO Commission on Social Connection released its report in June 2025: loneliness is linked to an estimated 871,000 deaths every year, or roughly 100 an hour. It has since appeared in speeches, press releases, and advocacy material as though it were a body count with the same evidentiary status as, say, deaths from malaria. It is not. It is a modeled estimate, several steps removed from any death certificate, and understanding those steps is the difference between using the figure responsibly and using it as a slogan.

Where the number comes from

The WHO report itself is explicit that this is an estimate derived from the epidemiological literature on social isolation and loneliness as risk factors for mortality, not a count of deaths coded to loneliness as a cause. No death certificate anywhere lists “loneliness” or “social isolation.” What exists instead are large meta-analyses estimating the increased relative risk of dying, over a follow-up period, associated with being socially isolated or lonely compared with being well connected. Julianne Holt-Lunstad’s 2015 meta-analysis in Perspectives on Psychological Science, pooling dozens of prospective studies, put the odds ratio for early mortality at 1.29 for social isolation and 1.26 for loneliness, with living alone at 1.32. Her earlier 2010 meta-analysis in PLoS Medicine, covering 148 studies and 308,849 participants, found that stronger social relationships were associated with a 50% greater likelihood of survival across follow-up periods — the figure most often shorthanded as “loneliness is as deadly as smoking.”

An 871,000-death global estimate is what you get when you take population-level prevalence of isolation and loneliness, apply a relative-risk figure of this kind, and multiply out across the population of countries covered. That is a legitimate exercise. Public health agencies do this routinely for tobacco, air pollution, and obesity. But it inherits every assumption embedded in the underlying relative-risk estimate, and those assumptions matter more here than in a field like tobacco epidemiology, where dose-response relationships are far better characterized.

What the relative-risk literature actually supports

The odds ratios behind the WHO estimate come overwhelmingly from prospective cohort studies, not randomized trials — no one randomizes people into isolation. That means the 1.26 to 1.32 range describes an association that persists after statistical adjustment for confounders such as baseline health, not a causal effect confirmed by controlled manipulation. Holt-Lunstad’s 2015 review notes that these associations remained after adjusting for health status, and that social deficits were, if anything, more predictive of death in samples averaging under 65 than in older cohorts — a finding that cuts against the intuitive assumption that isolation is mainly a problem for the very old. The American Heart Association’s 2022 scientific statement, reviewing cardiovascular and brain health specifically, found roughly a 30% increased risk of heart attack, stroke, or death from either associated with isolation and loneliness, and stated plainly that the absence of intervention evidence is the field’s central research gap. That is an unusual thing for a scientific statement to say outright, and it is worth taking at face value: the causal chain from social disconnection to a specific death is inferred, not observed.

None of this means the association is spurious. A relative risk in the 1.2 to 1.5 range, replicated across dozens of independent cohorts with different populations, follow-up periods, and covariate sets, is not a fluke. The U.S. Surgeon General’s 2023 advisory summarized the comparison to smoking up to 15 cigarettes a day as a way of communicating the magnitude of the mortality association, not a claim that the biological pathway is identical. The WHO figure sits on the same foundation. The question is not whether social disconnection predicts mortality — it plainly does, across a large and largely consistent literature — but whether a single global death count, precise to the hundred-thousands, is the right way to communicate that.

The coverage problem

The relative-risk estimates feeding the WHO’s calculation come almost entirely from cohort studies conducted in high-income countries, mainly the United States, the United Kingdom, and parts of Western Europe. Applying a risk ratio derived from, say, a Danish or American cohort to mortality patterns in low- and middle-income countries assumes the relationship between social disconnection and death operates the same way regardless of baseline mortality risk, health system access, family structure, and causes of death that dominate locally. The WHO report itself notes that loneliness rates are highest among young people and in low-income countries, which is precisely where the underlying epidemiological evidence is thinnest. Gallup’s 2023 survey across 142 countries found 24% of people worldwide report feeling lonely, with the highest rate — 27% — among adults aged 19 to 29, a pattern that does not obviously track the age-graded, cardiovascular-heavy mortality pathways studied in the Western cohort literature. A death-count estimate built by extrapolating a Western risk ratio onto a global prevalence map is doing more work than the underlying studies were designed to support.

There is also the isolation-versus-loneliness conflation running underneath all of this. The relative risks cited above come from studies that sometimes measure objective isolation (network size, contact frequency, living alone) and sometimes measure subjective loneliness (how connected a person feels), and sometimes both, without always distinguishing which is driving the mortality association. The WHO figure folds both into a single number. That may be defensible as a communication choice — the report is trying to convey the scale of a public health problem, not adjudicate a measurement debate — but it obscures a distinction the literature itself treats as consequential. A 2023 review in BMC Public Health mapping the state of loneliness and social isolation research names inconsistent measurement as a persistent barrier to comparing findings across studies. An 871,000 figure presented without that caveat reads as more settled than the literature it draws on.

What the figure is good for, and what it is not

As advocacy, the number does what the Commission likely intended: it forces social connection onto the same table as air pollution and tobacco in a policy conversation, and WHO’s Director-General framed connection at the report’s launch as a determinant of health warranting the same institutional seriousness as those established risk factors. That framing is defensible given the consistency of the underlying association, even if the precise death count is not something a health ministry should treat as a measured quantity.

What the figure cannot do is support claims about which interventions would reduce that mortality burden, by how much, or on what timeline. The relative risks it rests on describe association at the population level, observed retrospectively; they say nothing about whether a specific social-connection program, delivered to a specific population, would move the needle on deaths at all. A stronger version of this evidence base would report a range rather than a point estimate, disaggregate isolation from loneliness in the underlying calculation, and state plainly which regions the risk ratios were extrapolated into rather than estimated within. Until then, the honest use of the 871,000 figure is as an order-of-magnitude signal that social disconnection carries mortality weight worth taking seriously — not as a number to be quoted to the digit.