Has the 'Smoking 15 Cigarettes a Day' Claim Ever Been Replicated?

The line appears in nearly every article, speech, and policy document on loneliness written since 2023: social disconnection carries a mortality risk comparable to smoking up to 15 cigarettes a day. It appeared in the U.S. Surgeon General’s 2023 advisory, “Our Epidemic of Loneliness and Isolation,” and from there it has propagated through journalism, corporate wellness materials, and government press releases as a settled fact. It is worth asking where the number came from, and whether anyone has reproduced it.

The origin

The comparison traces to Julianne Holt-Lunstad’s 2010 meta-analysis in PLoS Medicine, which pooled 148 studies covering 308,849 participants and found that people with stronger social relationships had a 50% higher likelihood of survival over the follow-up periods studied. That is a substantial effect. To communicate its size to a non-specialist audience, Holt-Lunstad and colleagues compared it to the mortality effects of other well-established risk factors, smoking among them. The comparison was a way of putting an unfamiliar number next to a familiar one. It was not itself a study of smoking.

That distinction has been lost in the retelling. The Surgeon General’s advisory restates the comparison as if it were a direct empirical equivalence, and most coverage since has done the same. A rhetorical device for scaling one meta-analytic effect size against another has calcified into a figure that gets cited as though it were measured directly.

What the follow-up work actually found

Holt-Lunstad returned to this question in 2015, publishing a second meta-analysis in Perspectives on Psychological Science that separated the pooled “social relationships” measure into three components: social isolation, loneliness, and living alone. The odds ratios for early mortality were 1.29 for isolation, 1.26 for loneliness, and 1.32 for living alone. These are meaningfully smaller than what would be needed to sustain the smoking comparison at face value, and they are not the same construct as the composite “social relationships” measure used in 2010. The 2015 paper is more granular and, by breaking the exposure into parts, makes the original blunt comparison harder to defend as a single number. It does not repeat or test the smoking equivalence; it refines the underlying risk estimates in a direction that complicates the shorthand.

Holt-Lunstad’s own 2021 review in the American Journal of Lifestyle Medicine argues for treating social connection as a modifiable risk factor on par with diet, exercise, and smoking cessation, in the sense that all four are behavioral targets for prevention. That is a claim about policy framing, not a restated quantitative finding. It never re-derives the 15-cigarettes figure; it uses the smoking analogy again, this time explicitly as a rhetorical device for prevention science, not as an effect-size comparison.

The American Heart Association’s 2022 scientific statement is the most rigorous cardiovascular-specific synthesis available and it does not use the smoking comparison at all. It reports approximately 29% increased risk of heart attack or death from heart disease and 32% increased risk of stroke associated with social isolation and loneliness. These figures sit closer to the 2015 odds ratios than to anything resembling a smoking-equivalent hazard. The AHA statement is also unusually direct about what is missing: it names the absence of intervention trial evidence as the central gap in the field, a signal that the observational associations, however consistent, have not yet been tested through designs that could establish causation.

Why no one has replicated it

Part of the answer is that the claim was never designed to be replicated in its literal form. It was a comparison of relative risk magnitudes across two different bodies of literature, smoking epidemiology and social-relationship epidemiology, made for illustrative purposes in a single paper. Replicating it would require a study that measured both exposures in the same population using comparable methods and produced comparable hazard ratios, which is not what any of the subsequent work has attempted.

The 2023 BMC Public Health review of the state of loneliness and social isolation research is candid about a related problem: inconsistent measurement across studies is a persistent barrier to comparing findings, let alone reproducing a specific cross-domain comparison. Studies use different instruments for loneliness (the UCLA scale is common but not universal), different definitions of isolation, and different follow-up periods, which means that even the underlying mortality estimates being compared to smoking are not stable across the literature they are drawn from. A 2026 methodological review canvassing new approaches to measuring loneliness across time scales notes that the field is still working out how to compare findings that were generated with different instruments and designs, which is a tacit acknowledgment that a decade after the 2010 paper, measurement comparability remains unresolved.

What this means for the claim’s use

None of this means the underlying point is wrong. The weight of evidence, including the AHA’s cardiovascular-specific figures and the two Holt-Lunstad meta-analyses, supports the conclusion that chronic social disconnection carries a mortality risk large enough to matter at a population level, plausibly on the same order as several established behavioral risk factors. That is a defensible synthesis of the literature. It is a different thing from the specific claim that has been repeated verbatim in policy documents since 2023: that the size of the effect equals smoking 15 cigarettes a day. That number has one source, has not been independently derived since, and the two most relevant subsequent analyses produced effect sizes that sit below what the comparison implies.

This is not a case of a finding being overturned. It is a case of a rhetorical illustration in one 2010 paper becoming, through repetition in advisories, press releases, and journalism, indistinguishable from an established empirical result. Readers encountering the smoking comparison in a public health document are almost certainly encountering the same fifteen-year-old sentence rather than a new estimate.

A better basis for the claim would come from a study that directly compared mortality hazard ratios for social disconnection against those for smoking intensity within the same cohort, using a consistent measurement instrument for the social exposure and adjusting for the same covariates in both cases. No such study appears in the literature reviewed here. Until one does, the comparison should be treated as what it originally was: an aid to intuition, not a replicated finding.