This study demonstrated the significant role of smoking behavior in COVID-19-related educational inequality among SARS-CoV-2-infected individuals in China. This study revealed a strong association between lower educational attainment and increased odds of COVID-19 hospitalization. Smoking behavior significantly mediated this association, with effects varying by sex. Specifically, individuals with a lower educational status had higher rates of smoking history or current smoking, increasing their risk of hospitalization. Addressing this knowledge gap, our findings suggest that public health initiatives focused on tobacco control could help mitigate educational disparities in COVID-19 outcomes.
There is currently little evidence to explain educational differences in COVID-19 outcomes in terms of smoking behavior. In one study, the association between education and hospitalization due to COVID-19 was partially attenuated by lifestyle factors, which included smoking[6]. Nevertheless, the mediating effect of smoking is still unclear. The differentiation of educational attainment concerning smoking and resulting health inequalities has been identified across multiple countries[22]. Individuals aged 35–69 years from disadvantaged groups, including those with low levels of education, presented higher mortality rates due to smoking[23]. Tobacco use accounts for approximately half of the difference in mortality within this age range due to socioeconomic status [24]. A study indicated that there are educational disparities in the reduction in smoking prevalence among male adults in China across different cohorts[25]. This divergence in trends has resulted in a significant increase in educational disparities in smoking behavior among recent cohorts and plateau in the overall decline in smoking prevalence[25]. These findings underscore the importance of examining education disparities in smoking and assessing inequities in the burden of disease attributable to smoking in China.
Smoking behavior has been demonstrated to act as a mediator in the relationship between educational attainment and diseases such as coronary heart disease [26, 27]. Our findings on COVID-19 hospitalization indicated that smoking had a mediating effect ranging from 7.62–13.34%, which was smaller than the effects reported in the studies on specific chronic diseases or cancers. This difference may be attributed to different disease outcomes, assessment indicators, and measurement methods of education.
Importantly, our findings provide critical insights into the role of smoking in the educational stratification of diseases in China, particularly highlighting its detrimental effects on COVID-19 outcomes. A recent review confirmed that current and former smokers infected with SARS-CoV-2 face increased risks of hospitalization, severity, and mortality compared with those who never smoked[28]. This underscores the significant public health challenge posed by smoking, particularly as a pathway that exacerbates health disparities among different educational groups.
Following the World Health Organization (WHO) declaring COVID-19 as an ongoing health issue rather than a public health emergency[29], it is an opportune time to focus on mitigating factors that increase vulnerability. Our research emphasizes the urgent need for robust tobacco control policies and smoking cessation programs, especially those that target less educated populations. These interventions are essential not only for reducing the burden of COVID-19 but also for addressing broader health disparities exacerbated by smoking. By prioritizing these measures, we can better protect vulnerable groups from current and future public health challenges.
Our study also revealed that the mediating effect of smoking was more pronounced in females than in males. Although the smoking prevalence among women in China is much lower than that among men, recent national surveys have shown a higher smoking prevalence ratio between the lowest and highest education levels in females than in males[30, 31], suggesting that educational disparities have a more pronounced impact on smoking behavior in women. Compared with 35.1% of males, 59.6% of females with tobacco dependence have a primary school education or below[32]. These findings underscore the need for targeted public health interventions aimed at low-educated female smokers, who represent a particularly vulnerable group. Effective interventions and tobacco control policies should prioritize education and support specifically tailored for low-educated female smokers, providing them with resources and assistance in quitting smoking. Such targeted efforts are essential for reducing health disparities and promoting equity in health outcomes, particularly in the context of ongoing and future public health challenges such as COVID-19. By addressing the unique needs of this population, tobacco control initiatives can make significant strides in reducing the burden of tobacco-related diseases and improving overall community health.
We also found that ex-smokers had a greater risk of COVID-19 hospitalization than did current smokers. This increased risk in ex-smokers may be due to diseases caused by previous smoking, which can persist for years after cessation. Studies have indicated that smokers often quit smoking due to smoking-related diseases, suggesting that ex-smokers might have poorer health than current smokers do [33]. Our data revealed that ex-smokers had significantly greater proportions of patients with chronic respiratory diseases, chronic liver diseases, and cancers than did current smokers (P < 0.05). Additionally, individuals who quit smoking immediately before or after hospitalization due to severe COVID-19 symptoms might be recorded as ex-smokers, introducing reverse causality. Last, collider bias might explain this finding, where conditioning on a collider (such as testing or hospitalization) could create an artificial link between smoking status and adverse COVID-19 outcomes[34].
Our findings revealed that ex-smokers had a greater risk of COVID-19 hospitalization than didi current smokers. This increased risk among ex-smokers may be attributed to the long-term effects of diseases caused by previous smoking, which can persist even after cessation. Studies suggest that many smokers quit smoking due to smoking-related diseases, indicating that ex-smokers might generally have poorer health than current smokers[33]. Our data revealed significantly greater proportions of chronic respiratory diseases, chronic liver diseases, and cancer among ex-smokers than among current smokers (P < 0.05). Additionally, individuals who quit smoking immediately before or after hospitalization for severe COVID-19 symptoms might be classified as ex-smokers, potentially introducing reverse causality. Moreover, collider bias could also explain this finding, where conditioning on factors such as testing or hospitalization could artificially link smoking status with adverse COVID-19 outcomes[34]. These insights underscore the importance of targeted smoking cessation programs, especially for those with existing health conditions. By addressing the unique health challenges faced by ex-smokers and supporting current smokers in quitting, public health initiatives can enhance overall health outcomes and progress toward tobacco control goals.
Several potential limitations of our study should be noted. First, our research sample was obtained through an online survey. Hence, the potential bias of this methodology may be inherent in our study. For example, people with the lowest income and illiteracy may be excluded from the survey[35]. Although our study subjects were distributed throughout the Chinese mainland, the average education level of the sample may be higher than the national average because of the data collection method. Considering the issue of sample representativeness, caution should be exercised when generalizing the research findings to populations with lower education levels. However, we performed a sensitivity analysis of an inverse-probability weighted method to eliminate the selection bias. Second, in our retrospective survey, self-reported information regarding COVID-19 hospitalizations may pose a greater risk of misreporting than medical records do. Nevertheless, we implemented rigorous measures for data quality control. Third, certain confounding factors, such as COVID-19 medications, treatment, and proximity to healthcare facilities, could not be collected in the questionnaire. Finally, our results do not establish causality and, more importantly, do not guarantee that controlling smoking among individuals with lower educational status will necessarily reduce their risk of contracting COVID-19 and other pandemics. However, our findings suggest that this scenario is plausible. Therefore, extending alternative approaches, such as Mendelian randomization, could be employed to further clarify the causality.