New insights into the paradox between smoking and the risk of SARS-CoV-2 infection (COVID-19): Insufficient evidence for a causal association.

Abstract

Previous studies have reported a 'smoker's paradox', where people who smoke appear to be protected against Severe Acute Respiratory Syndrome CoronaVirus-2 (SARS-CoV-2) infection (COVID-19). This conflicts with well-established evidence that people who smoke are generally more vulnerable to respiratory infections. In this study, we aimed to validate the association between smoking and SARS-CoV-2 infection in a general Dutch population, and to evaluate the evidence underlying the possible causal relationship between smoking and SARS-CoV-2 infection by applying a modern adaptation of the Bradford Hill criteria. In total, 57,833 participants from the Lifelines Cohort Study were included in the analysis. Smoking status, including never smoker, current smoker, and former smoker, was derived from the Lifelines general assessment between 2014 and 2017, while SARS-CoV-2 infection status was derived from an additional COVID-19 questionnaire from 2021 to 2022. Logistic regressions were used for the association between smoking status and infection status. The adapted Bradford Hill's criteria, including the strength of association (including an analysis of plausible confounding), plausibility, temporality and study design suitability, were applied to evaluate the existing literature. We found, compared with never smokers, an increased risk of SARS-CoV-2 infection for former smokers (odds ratio (OR)=1.07, 95% confidence interval (CI)=1.01-1.13), but a reduced risk for current smokers (OR=0.85, 95% CI=0.79-0.92), after adjusting for several relevant covariates. However, we discerned a possible explanation of the smoker's paradox since we observed that current smokers were more likely to be non-responders to the COVID-19 questions and, more importantly, these non-responders were more likely to have other established risk factors for SARS-CoV-2 infection. There is insufficient evidence to suggest that smoking protects against SARS-CoV-2 infection. According to the adapted Bradford Hill's criteria, we observed a high inconsistency between study results, a high possibility for residual confounding and no clear evidence for biological plausibility. Future studies should include linkage with the confirmed testing results from national healthcare registries to mitigate avoidable bias.