Social isolation rose worldwide after 2019, and it started with the poor

Claims about a global epidemic of loneliness usually rest on national surveys that ask different questions in different years. A cross-sectional analysis published in JAMA Network Open on September 2 uses one question, asked the same way, in more than 150 countries, every year from 2009 to 2024 [s1].

The measure is deliberately spare: the proportion of respondents who answered "no" when asked whether they had relatives or friends available to help in times of trouble [s1]. That is isolation in a hard, practical sense — not a feeling but a missing resource.

What was analysed

The study used the Gallup World Poll, a globally representative repeated cross-sectional survey with consistent methodology across more than 150 countries, sampling approximately 1,000 adults aged 15 or older in each country at each annual assessment [s1].

The analysis covers 2,483,935 person-level assessments across 16 time points and 159 countries; mean age was 41.7 years and 53.1% of respondents were women [s1]. Household income was collected in local currency and coded into quintiles within each country, so "low income" means low relative to one's own country, not globally [s1].

The findings

Global mean isolation was stable before the pandemic [s1].

A marked increase occurred between 2019 and 2020, coinciding with the onset of COVID-19, and it fell disproportionately on lower-income groups: β = 2.6 percentage points (95% CI 0.9 to 4.4; p = .003), an 11.0% increase [s1].

From 2020 to 2024, isolation kept rising, but the steeper increases were now among higher-income groups: β = 1.9 (95% CI 0.7 to 3.1; p < .001), a 12.3% increase [s1].

Over the full period, global prevalence of social isolation rose 13.4%, from 19.2% (95% CI 17.3% to 21.6%) in 2009 to 21.8% (95% CI 19.4% to 24.2%) in 2024 — with the entire increase occurring after 2019 [s1].

The income gap peaked in 2020, when 26.4% (95% CI 23.6% to 29.2%) of lower-income respondents were isolated against 15.6% (95% CI 13.6% to 17.7%) of higher-income respondents [s1]. By 2024 global isolation stood 2.6 percentage points (95% CI 0.7 to 4.5) above pre-pandemic levels and the income disparity was 8.6 percentage points (95% CI 5.1 to 12.1) [s1].

Countries diverged: 54 experienced worsening isolation with widening disparities, while 41 improved [s1].

Why the shape of the curve is the story

The finding that carries most information is not the headline increase but its sequencing. The first shock hit poorer people hardest, which is what one would expect if the mechanism was disruption to the informal networks — extended household, neighbourhood, workplace — that lower income households rely on more heavily. The second, slower phase spread upward through the income distribution [s1].

That two-stage pattern is hard to explain with a single cause, and the study does not attempt to identify one. It is a description, not an explanation.

What the design cannot do

This is repeated cross-sectional data. The same people are not followed over time, so nothing here tracks an individual's trajectory into or out of isolation [s1].

The exposure is a single binary item. Having someone to call in a crisis is a meaningful threshold, but it is not the same construct as loneliness, which is subjective and can coexist with a full social network. Conflating the two is a common error in coverage of this literature; the study itself does not make it.

Survey mode changed for many countries during the pandemic period across international polling generally, and any shift from face-to-face to telephone interviewing can move responses. The abstract does not report how mode effects were handled, which is a real limitation for a study whose central finding is a 2019-to-2020 discontinuity.

Fifteen years of annual national samples of roughly 1,000 people give reasonably tight global estimates but wide ones for individual countries. The 54-versus-41 country split should be treated as descriptive.

What is at stake

The reason this measure attracts health researchers is a body of evidence linking social connection to survival. A 2015 meta-analytic review found that, across studies controlling for several possible confounds, weighted average effects were odds ratio 1.29 for social isolation, 1.26 for loneliness, and 1.32 for living alone — corresponding to 29%, 26%, and 32% increased likelihood of mortality [s2]. That review found no difference between objective and subjective measures, and found the associations were more predictive of death in samples averaging under 65 years old [s2].

Those are observational estimates with the usual reverse-causation problem: illness isolates people as well as the reverse. But an effect of that magnitude, if even partly causal, makes a 2.6 percentage point global increase in isolation a population-health quantity rather than a social observation.

What to watch

Whether the post-2020 rise plateaus or continues in the 2025 and 2026 waves; whether country-level policy differences explain any of the divergence between the 54 worsening and 41 improving countries; and whether longitudinal cohorts confirm that the people newly isolated are the ones the cross-sectional data implies.

This article describes cross-sectional survey research and observational meta-analysis. It is informational only and is not medical advice.

Sources

• [s1] Fuller-Rowell TE, Sultana S, Kawachi I, Global Trends and Disparities in Social Isolation, JAMA Network Open, 2025;8:e2532008, published 2025-09-02.
• [s2] Holt-Lunstad J, Smith TB, Baker M, Harris T, Stephenson D, Loneliness and social isolation as risk factors for mortality: a meta-analytic review, Perspectives on Psychological Science, 2015;10:227-237, published 2015-03-01.