A dual analytical approach combining observational and Mendelian randomisation data suggests smoking linked to worse outcomes in COVID-19.
Throughout the COVID-19 pandemic, several important risk factors for more severe disease have become apparent including male sex, increased age and cardiovascular co-morbidities. Whether or not being a current smoker affects COVID-19 outcomes is less clear with some analyses indicating a higher risk of worse outcomes, whereas others reviews suggest a reduced risk . One solution to untangling this ambiguity is the adoption of a dual analytical approach to data analysis, for example, by comparing the results from observational studies with those from a Mendelian randomisation study. In a Mendelian randomisation (MR) study, the underlying assumption is that a genetic variant influences only the variable of interest and since genetic variants are randomly allocated at birth, MR is less susceptible to confounding which is a problem in observational studies. Genetic analysis can be used to identify variants associated with particular traits. For example, tobacco and alcohol use are known to have heritable behaviours and in one study, researchers identified 566 genetic variants in 406 loci associated with multiple stages of tobacco use, e.g., initiation, cessation, and heaviness (i.e., the number of cigarettes smoked per day).
Using a dual analytical approach to examine the relationship between smoking and outcomes in COVID-19, a team from the Nuffield Department of Primary Care Health Sciences, University of Oxford, UK, turned to data held within the UK Biobank. The researchers separately explored the relationship between smoking and COVID-19 using findings from both observational studies and Mendelian randomisation. For the MR study, researchers used established genetic proxies for smoking initiation and smoking heaviness (i.e., the number of cigarettes smoked per day). The results were analysed using multivariate logistic regression analysis adjusted for several variables including age, sex, ethnicity and several co-morbidities.
For the observational analysis, there were 421,469 individuals with a median age of 68.6 years (55.1% female). Current smokers were found to have an overall higher risk of hospitalisation (adjusted odds ratio, aOR = 1.80, 95% CI 1.26 – 2.29) and mortality (aOR = 4.89, 95% CI 3.41 – 7.0).
Similarly, in the Mendelian randomisation analysis, genetic propensity to initiate smoking was associated with a higher risk of hospitalisation (aOR = 1.60, 95% CI 1.13 – 2.27), but not mortality (aOR = 1.35, 95% CI 0.82 – 2.22, p = 0.23). For genetically predicted higher number of cigarettes smoked per day, the risk of hospitalisation was also higher (aOR = 5.08, 95% CI 2.04 – 12.66) and risk of death (aOR = 10.02, 95% CI 2.53 – 39.72).
Discussing their findings, the authors said that this was the first study to adopt a dual analytical approach to assess the relationship between smoking and outcomes from COVID-19. They concluded that the congruence between the two methods indicated that a lifelong predisposition to smoking and smoking heaviness supported a causal effect (found in the observational analysis) of smoking on COVID-19 severity.
Clift AK et al. Smoking and COVID-19 outcomes: an observational and Mendelian randomisation study using the UK Biobank cohort. Thorax 2021