Tobacco smoking and the risk of Long COVID: a prospective cohort study with mediation analysis

Publication type

Journal Article

Series Number

Authors

Publication date

October 15, 2025

Summary:

Background:

Tobacco smoking is a well-established risk factor for severe acute COVID-19 outcomes, but evidence regarding its role in Long COVID is limited and inconsistent. This study investigated whether pre-pandemic smoking independently predicted Long COVID and assessed mediation by long-standing illness or disability in a nationally representative cohort.

Methods:

We analysed data from Waves 10 (2018–19) and 14 (2022–23) of the UK Household Longitudinal Study. Smoking status (current vs non-smoker) and covariates (age, sex, education, income satisfaction, ethnicity, rural/urban residence) were measured at baseline (Wave 10). Long COVID, defined as symptoms lasting ≥12 weeks following initial COVID-19 infection, was assessed at follow-up (Wave 14). Logistic regression was used to estimate the total association between smoking and Long COVID. We then applied generalized structural equation modelling and parametric causal mediation analysis, specifying long-standing illness or disability at baseline as the mediator.

Results:

Among 11,944 participants, 1097 (9.2 %) reported Long COVID symptoms at follow-up. In the unadjusted model, smoking was associated with increased odds of Long COVID (odds ratio [OR] = 1.22, 95 % CI: 1.00–1.48, p = 0.05), although this was only borderline significant. After adjusting for demographic and socioeconomic factors, the association was no longer statistically significant (adjusted OR = 1.11, 95 % CI: 0.91–1.35, p = 0.32). The structural equation model indicated that smoking was associated with higher likelihood of long-standing illness or disability at baseline (β = 0.461, 95 % CI: 0.33–0.59, p <0.001, log-odds scale), which in turn predicted Long COVID (β = 0.435, 95 % CI: 0.30–0.57, p <0.001, log-odds scale). Mediation analysis revealed a small but statistically significant indirect effect of smoking on Long COVID operating through long-standing illness or disability (risk difference = 0.0057, 95 % CI: 0.0020–0.0095, p = 0.003), but no significant direct effect (risk difference = 0.0027, 95 % CI: –0.0144 to 0.0199, p = 0.76).

Conclusion:

Smoking did not independently predict Long COVID, but may increase vulnerability indirectly through pre-existing long-standing illness or disability.

Published in

Journal of Epidemiology and Population Health

Volume

Volume: 73

DOI

https://doi.org/10.1016/j.jeph.2025.203142

ISSN

29504368

Subjects

Notes

Open Access

Under a Creative Commons license

Uses Understanding Society data (not Understanding Society - COVID-19 Study, 2020)

#588797

News

Latest findings, new research

Publications search

Search all research by subject and author

Podcasts

Researchers discuss their findings and what they mean for society

Projects

Background and context, methods and data, aims and outputs

Events

Conferences, seminars and workshops

Survey methodology

Specialist research, practice and study

Taking the long view

ISER's annual report

Themes

Key research themes and areas of interest