Introduction

In recent years, new tobacco products such as electronic cigarettes (e-cigarettes) and heated tobacco products (HTPs) have been developed to reduce exposure to tobacco smoke. E-cigarettes are devices that heat liquid to generate a vapor for inhalation, and are available with or without nicotine1. HTPs are devices that generate aerosols by heating tobacco leaves without burning them2. In Europe and the Unites States, e-cigarettes have rapidly gained popularity, particularly among young people, including teenagers1.

In Japan, nicotine-containing e-cigarettes are not commercially available because nicotine is treated as a prescription drug, while nicotine-free e-cigarettes are not regulated. E-cigarette use in Japan has been gradually increasing3; in 2018, about 5% of young adults aged 15–29 years used e-cigarettes4. However, the rate of use among the general population is not very high. Meanwhile, HTPs have been approved as tobacco products and marketed in Japan similarly to combustible cigarettes. HTPs are rapidly spreading in Japan and account for about one-third of the tobacco market, as estimated using statistical data from the Tobacco Institute of Japan for 20215.

Combustible cigarettes contain high levels of nicotine, causing dependence, and tar in their smoke, and habitual use can lead to diseases such as pulmonary disease, cardiovascular disease, and malignant tumors of lungs, larynx, esophagus, etc.6,7. On the other hand, the effects on health of HTPs themselves remain unclear, because HTPs have only been on the market for a short time and most current users are dual users who also use combustible cigarettes3,8. Components in aerosols from HTPs are reported to show slightly lower to similar levels of nicotine and relatively higher levels of propylene glycol, glycerol, and acetol when compared to those of combustible cigarettes2,9. Several reports have described users of HTPs developing severe eosinophilic pneumonia after brief use or when changing from combustible cigarettes, as is sometimes seen in those starting combustible cigarettes, and components in aerosols may have contributed to this10,11.

Coronavirus disease 2019 (COVID-19) is an ongoing pandemic, sometimes causing severe pneumonia and death, and is a major problem worldwide. COVID-19 is caused by severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection, and angiotensin-converting enzyme 2 expressed in airway epithelium has been reported as the entry receptor12. Regarding the risk of SARS-CoV-2 infection, some reports have suggested that use of combustible cigarettes increases susceptibility to the virus by increasing receptor expression13, while others have suggested that nicotine downregulates the receptors14. Some reports on clinical cases have found that current tobacco use was associated with severe COVID-1915,16,17, while others found no such association18,19,20. Results regarding the association of tobacco use with SARS-CoV-2 infection and COVID-19 severity are thus conflicting.

During the COVID-19 pandemic, tobacco use has reportedly decreased due to the closure of bars and pubs and constraints on the ability to socialize with friends21, while use of HTPs has reportedly increased due to switching from combustible cigarettes22 and increased demand for tobacco when outings were restricted23. However, insufficient evidence has been accumulated regarding the safety of HTPs in the context of COVID-19.

In response to these circumstances, we examined the association of tobacco products use including HTPs with SARS-CoV-2 infection and severe COVID-19. Furthermore, we used data from the general population in the present study because many previous reports have focused on inpatients and have noted low proportions of tobacco users18.

Methods

Study design and participants

We aimed to examine the associations of tobacco products use including HTPs with both SARS-CoV-2 infection and severe COVID-19. We conducted a cross-sectional study using data from the Japan “Society and New Tobacco” Internet Survey (JASTIS), an annual cohort study on the internet aimed at investigating the effects on health of new tobacco products (HTPs and e-cigarettes) (Study profile is shown in a previous report24). Participants in JASTIS were selected from panelists of Rakuten Insight, one of Japan’s largest internet research companies, with 2.2 million users covering various social categories in the general Japanese population (according to education, housing tenure, marital status, etc.). A self-reporting questionnaire was administered on the internet between February 1 and 28, 2022 to randomly sampled panelists by sex and age group at the launch of JASTIS in 2015 and additionally sampled panelists in each year, and the survey was closed when the target number of respondents was reached. Questions consisted of tobacco use status including use of HTPs, history of SARS-CoV-2 infection (within the preceding year/over one year prior to the survey) and severe COVID-19 (hospitalization and oxygen administration), comorbidities, and other background characteristics of the respondents. Respondents answered all items and there were no missing values.

Of the 33,000 participants, 30,130 were included in this study and the first analysis (Fig. 1). Those who had been infected with SARS-CoV-2 once were included in the second analysis (n = 1097). The 2870 participants not included in the first analysis were excluded because of incorrect or unnatural answers such as: wrongly answering the question, “Choose the second one from the bottom”; reporting use of all drugs; and reporting having all comorbidities (see details in Supplementary Note). In addition, 20 individuals who reported having been infected twice were excluded from the second analysis to prevent duplicate results.

Figure 1
figure 1

Inclusion flow chart. JASTIS the Japan “Society and New Tobacco” Internet Survey, SARS-CoV-2 severe acute respiratory syndrome coronavirus 2.

Full size image

This study was conducted in accordance with the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects. All study protocols were reviewed and approved by the Research Ethics Committee of the Osaka International Cancer Institute (No. 20084-8). Informed consent was obtained from all participants at the time of study entry by electromagnetic means.

Definition of variables

Tobacco use status was determined based on the respondents’ answers to several questions about use of each tobacco product. Combustible cigarettes were defined as cigarettes and roll-your-own tobacco. HTPs included IQOS, Ploom Tech, Ploom S, Ploom X, glo, and lil HYBRID—brands sold in Japan as of February 2022. The questions and definitions of current tobacco users for each of the above tobacco products were as follows: (1) those who selected 4 or 5 in the question “Do you currently smoke or use tobacco? [(1) Never used before; (2) Tried using more than once, but did not use habitually; (3) Used to use habitually, but stopped; (4) Use occasionally; and (5) Use almost daily]”; (2) those who answered 1 or more to the question “On how many of the preceding 30 days did you smoke or use tobacco? (answer with a whole number between 0 and 30)” (Fig. 2). Particularly for the use of new tobacco products, there is no common historical definition of current use25, and Definition 2 was established with reference to previous reports25,26. Frequency of use was therefore not considered. Current users in the applicable category (i.e., combustible cigarettes, HTPs, or both) were defined as those who met more than one of the above criteria for any of the tobacco products (e.g., roll-your-own tobacco, IQOS) in the applicable category. Current dual users were defined as those who were currently using both combustible cigarettes and HTPs. Those who reported that they had never used any tobacco were considered as never-users. Those who did not meet any of these definitions were considered as former users. Those who stopped using any tobacco product in the applicable category within the preceding year were included as current users in each category for tobacco use status at one year prior to the survey. Questions about amount of tobacco used per day and year of tobacco cessation did not occur for respondents who were not habitual users (those who answered “2. Tried using more than once, but did not use habitually.” for tobacco use status). Because of the non-negligible missing values for both former users and current users (according to Definition 2), the amount and duration of tobacco use were not included in the analyses in this study.

Figure 2
figure 2

Definition of tobacco use status. Horizontal bars indicate the time course, and gray color indicates the duration of each tobacco use. Inverted triangles indicate the time of the survey (▼) and one year prior to the survey (▽). HTPs heated tobacco products.

Full size image

Regarding SARS-CoV-2 vaccination status, two messenger ribonucleic acid vaccinations or one viral vector type vaccination were defined as two doses, and these plus one booster vaccination were defined as three doses. The history of SARS-CoV-2 infection and severe COVID-19 (hospitalization and oxygen administration) were determined based on the answers to the question of whether the respondent had experienced each, either within the preceding year or over one year prior to the survey.

Details including other items are described in Supplementary Note, Tables S1 and S2.

Statistical analysis

For the first analysis, we performed multivariable logistic regression analysis to examine the association between tobacco use status and SARS-CoV-2 infection among all participants. We adjusted for the following potential confounding factors: age; sex; occupation; household annual income; education; housing tenure; whether the respondent lived in a prefecture under alert (21 prefectures in which the State of Emergency had been declared more than once, as listed in Supplementary Table S1); presence of household members other than the respondent; whether the respondent frequently interacted with people living separately; total metabolic equivalents (METs) per day; alcohol consumption; comorbidities (12 diseases associated with COVID-19 severity in past reports16,27, as listed in Table 1); engagement in infection prevention measures (percentage of 12 infection prevention behaviors preferred in Japan and listed in Supplementary Table S2 engaged in; partially modified with reference to a previous report28); and SARS-CoV-2 vaccination status.

Table 1 Characteristics of participants.
Full size table

For the second analysis, we performed multivariable logistic regression analyses to investigate associations between tobacco use status and severe COVID-19 (hospitalization and oxygen administration). Age, sex, comorbidities (as above), and vaccination status were adjusted as covariates. In these analyses, we also adjusted for the year of infection because SARS-CoV-2 strains predominant at the time and vaccination status differed between infection within the preceding year and infection over one year prior to the survey. Those infected within the preceding year were examined based on tobacco use status and vaccination status at the time of the survey, and those infected over one year prior to the survey were examined based on their tobacco use status at one year prior to the survey (since vaccination had started in Japan in February 202129, these respondents were considered unvaccinated).

Using these models, adjusted odds ratios (aORs) and 95% confidence intervals (CIs) were obtained, and values of p < 0.05 were considered statistically significant. We also examined the variance inflation factor (VIF) for each covariate and confirmed that covariates were not linear with each other (VIF < 2 was considered to indicate non-linearity).

The population bias due to being an internet survey was adjusted for using inverse probability weighting (IPW). To estimate statistical weightings, we used an existing nationally representative sample from the Comprehensive Survey of Living Conditions of People on Health and Welfare (CSLCPHW)30. Two surveys (JASTIS and CSLCPHW) were combined and logistic regression analysis adjusted for background factors such as residence, education, housing tenure, and health status was used to estimate the probability of “being a respondent in an internet survey,” i.e., propensity score. Details on IPW were provided in previous reports24,26. The numbers of individuals shown in the text and tables indicate the weighted numbers; unweighted numbers are shown in Supplementary Tables.

All analyses were performed using R (version 4.0.3) and EZR on R commander (version 1.54)31.

Results

A total of 30,130 individuals between 16 and 81 years of age participated in the study, comprising 14,718 men (48.8%) and 15,412 women (51.2%). Of these, 1117 (714 men, 403 women) reported having been infected with SARS-CoV-2. Characteristics of all participants and infected individuals are shown in Table 1. The mean age of all participants was 47.9 years, while that of the infected was 34.9 years, with a bias toward younger age groups. The percentages of current tobacco users at the time of the survey were 24.3% in all and 53.0% among those infected, and current users of HTPs (including dual users) among current tobacco users were 51.3% and 84.5%, respectively.

Among those infected (excluding those infected twice), 525 were infected within the preceding year and 571 were infected over one year prior to the survey (Supplementary Table S3). The percentage of current tobacco users was higher in previously infected individuals than in recently infected ones (73.0% vs. 30.9%), and many individuals were hospitalized (67.3% vs. 18.9%) and received oxygen (66.9% vs. 15.0%) due to COVID-19 in the same group. Among recently infected individuals, 82.5% had been vaccinated against SARS-CoV-2 twice or more.

Results of the first analysis of the association between tobacco use status and SARS-CoV-2 infection among all participants (n = 30,130) are shown in Table 2. Current users of HTPs (aOR = 1.65, 95%CI 1.26–2.15) and current dual users of combustible cigarettes and HTPs (aOR = 4.66, 95%CI 3.89–5.58) were more likely to be infected with SARS-CoV-2 compared to never-users. Former users and current users of combustible cigarettes showed no significant difference in infection rates when compared to never-users. In addition, individuals who were working in some occupations (healthcare worker/remote worker/other worker: aOR = 1.48/1.93/1.35; 95%CI 1.11–1.98/1.42–2.61/1.10–1.65), living in prefectures under alert (many of which were densely populated areas) (aOR = 1.72, 95%CI 1.50–1.99), and frequently interacting with people living separately (aOR = 1.63, 95%CI 1.43–1.86), as well as individuals who were highly physically active (aOR = 1.63, 95%CI 1.43–1.87), were more likely to be infected compared to those who were not. Individuals with two or more comorbidities were more likely to be infected than those without comorbidities (aOR = 2.71, 95%CI 2.26–3.25). SARS-CoV-2 vaccination status was not associated with infection history, but individuals engaging in a high percentage of infection prevention measures were less likely to be infected compared to those with a low percentage (aOR = 0.57, 95%CI 0.50–0.65). A similar analysis was performed using the data for tobacco use/SARS-CoV-2 vaccination status at one year prior to the survey (as mentioned above, no vaccinations had been administered at that time in Japan), yielding similar results (data not shown). In addition, there was no obvious change in tobacco use status at the time of the survey from one or two years prior to the survey (Supplementary Table S4). Adjustment using current status at the time of the survey for risk assessment was therefore considered appropriate.

Table 2 Factors associated with SARS-CoV-2 infection among all participants (n = 30,130).
Full size table

Results of the second analysis of the associations between tobacco use status and severe COVID-19 (hospitalization and oxygen administration) among those infected (n = 1097) are shown in Table 3. Of the infected respondents, 483 (44.0%) were hospitalized and 461 (42.0%) were administered oxygen. Current dual users of combustible cigarettes and HTPs compared to never-users (aOR = 3.17, 95%CI 2.11–4.77) and individuals who had two or more comorbidities compared to those without comorbidities (aOR = 1.46, 95%CI 1.01–2.11) were more likely to be hospitalized due to COVID-19. All tobacco users (including former users) were more likely to be administered oxygen compared to never-users (former user/current user of combustible cigarettes/current user of HTPs: aOR = 1.88/3.17/1.90, 95%CI 1.11–3.19/1.77–5.67/1.01–3.59), and dual users required oxygen administration the most at the time of infection (aOR = 4.15, 95%CI 2.70–6.36). Individuals with multiple comorbidities were also more likely to be administered oxygen than those without comorbidities (aOR = 1.86, 95%CI 1.27–2.72). Individuals vaccinated twice or more had almost half the aORs for hospitalization and oxygen administration compared to those who received no or only one dose of vaccination (hospitalization/oxygen administration: aOR = 0.57/0.49, 95%CI 0.33–0.98/0.27–0.87) even after taking into account differences in severity rate by year of infection. Neither age group nor sex were associated with severe disease in this population.

Table 3 Factors associated with hospitalization and oxygen administration due to COVID-19 among participants infected with SARS-CoV-2 once (n = 1097).
Full size table

The results of the second analysis adjusted for each comorbidity are shown in Supplementary Table S5, and the results for the unweighted population are shown in Supplementary Tables S6, S7, and S8.

Discussion

We examined the association of tobacco use including HTPs with SARS-CoV-2 infection and severe COVID-19 using data from a large-scale questionnaire. Logistic regression modeling revealed that use of HTPs alone or in combination with combustible cigarettes was associated with higher rates of SARS-CoV-2 infection and severe COVID-19 than never-use, and both rates were markedly higher in the combination group than in the other tobacco status groups. To the best of our knowledge, this represents the first report to examine associations between use of tobacco including HTPs and severe COVID-19.

While many COVID-19 patients were asymptomatic and therefore undiagnosed32, one report suggested that use of e-cigarettes had increased the percentage of symptomatic patients at the time of SARS-CoV-2 infection33. In the present study, many users of HTPs were diagnosed as infected, and this may be due to an increase in symptomatic patients among users of HTPs. In addition, Japan had no lockdown period during the COVID-19 outbreak, and Japanese citizens were only asked to refrain from going out through the declaration of the State of Emergency when the number of infected people increased (Supplementary Table S1). In this study population, users of HTPs were more physically active than non-users (average 47.1 vs. 38.8 METs per day; data not shown). Although we adjusted for confounding factors such as occupation and interaction with people living separately, high activity with frequent outings may have contributed to the increase in infection. Meanwhile, engagement in infection prevention measures was higher among current users of HTPs than among others, including users of combustible cigarettes (average 90.6% vs. 83.7%; data not shown), suggesting that health awareness may have increased the number of those tested for SARS-CoV-2.

Regarding history of hospitalization, no significant associations were evident among tobacco users other than dual users. In Japan, there was a period during which all COVID-19 patients were hospitalized at the beginning of the outbreak, and thus hospitalization may not necessarily reflect disease severity. Because many tobacco users were among those infected over one year prior to the survey (Supplementary Table S3), corresponding to the above period, the association with hospitalization may not have been significant, unlike with oxygen administration.

Use of combustible cigarettes is generally considered a risk factor for severe COVID-19, including increased risk of respiratory failure requiring mechanical ventilation, and death15,16,17, whereas few reports have described the association between new tobacco products and COVID-19. Similar to the abovementioned papers, current use of combustible cigarettes was associated with severe COVID-19 (i.e., oxygen administration) in the present study. As for HTPs, current single users displayed an aOR for oxygen administration almost equal to or slightly higher than that in former tobacco users. Since 86.0% of single users of HTPs had a history of using combustible cigarettes, the inference was that they were affected by past use of combustible cigarettes, like the majority of former tobacco users. In addition, one report from Japan noted a decrease in the immunoglobulin G antibody titer after SARS-CoV-2 vaccination in current users of tobacco, including HTPs34, which may have been associated with severe disease in those infected within the preceding year in the present study. Interestingly, however, dual users showed a markedly higher aOR for oxygen administration than those in other tobacco status groups. Dual users may have used combustible cigarettes in situations where its use was possible and have used HTPs where it was not35, and may have had contact with others without a mask in areas designated for cigarette smoking and other spaces. Wearing a mask is hypothesized to reduce the amount of inhaled virus and make COVID-19 less severe36,37, and so the amount of viral exposure might have been related to disease severity. However, if dual use itself increases the risk of severe disease, the mechanism awaits further study.

Infected respondents in this study skewed younger than in previous studies of hospitalized patients20,27. With regard to severity, older age groups had higher aORs than younger age group, but unlike previous reports16,27, no significant association with age was seen. This may be because elderly individuals at higher risk of severe disease than healthy ones, such as those already hospitalized or in nursing homes, were unable to participate. Similarly, no association between sex and COVID-19 severity was identified; this may be because tobacco use and other factors were adjusted for as covariates in the analysis. The tobacco use rate was higher among males than among females (Supplementary Table S9), and males had a significantly higher odds ratio for severe disease than females in univariable analysis (Supplementary Table S10).

Presence of one comorbidity was also not associated with severity. In this study, comorbidities were selected from those associated with COVID-19 severity in previous reports16,27. Diseases among individuals with one comorbidity were hypertension in 46.6% and hyperlipidemia in 24.3%, neither of which are listed by the United States Centers for Disease Control and Prevention as risk factors for severe COVID-1938 because of inconsistent conclusions among reports. We thus consider that having only a single comorbidity was not associated with severe disease.

SARS-CoV-2 vaccination, once adjusted for engagement in infection prevention measures, was not associated with infection history. A past report has indicated that post-vaccination breakthrough infections were common, particularly with mutant strains of SARS-CoV-2, but vaccination prevented severe disease39. Similarly, in this study, even after adjusting for the year of infection (i.e., the viral strains predominant at the time), individuals vaccinated twice or more were less likely to develop severe disease compared to those who received no or only one dose of vaccination.

Limitations

This study had several limitations that should be kept in mind.

First, this was a cross-sectional study that only compared tobacco use status and other background characteristics with history of SARS-CoV-2 infection and severe COVID-19, and so did not examine causal relationships.

Second, administration as an internet survey introduced bias in participant selection. The obtained data were weighted using IPW, but because this was a study of a sample from the general population, not hospitalized patients, severely ill or deceased COVID-19 patients (particularly elderly individuals) would not have had access to this survey. This may help explain the bias toward younger age among the infected and why older age was not associated with severe disease in this study.

Third, this was a questionnaire-based survey, and the diagnosis of COVID-19 was self-reported. However, because of the great interest in COVID-19 at this time, while asymptomatic individuals may not have been diagnosed, at least symptomatic ones were likely to have been examined or tested for the virus and would have known their results. History of severe disease was also self-reported, and no questions investigated interventions beyond oxygen administration (e.g., use of ventilators or extracorporeal membrane oxygenation). In this regard, since we defined severe disease as hospitalization or oxygen administration, which were easily understood by the general population, the patients themselves were likely to have known whether they had been in these situations.

Fourth, actual dates of infection were not known and the exact circumstances at the time of infection may not have been reflected in the analyses. For infection, similar results were obtained by adjusting for tobacco use/vaccination status at the time of the survey and at one year prior to the survey. For severe disease, results were adjusted by the status of the year of infection, which likely reflected the actual circumstances. We therefore believe that the possibility of reverse causation such as “tobacco cessation/vaccination because of infection” was eliminated to the extent possible.

Fifth, the amount and duration of tobacco use was not considered, and users of HTPs may have been using combustible cigarettes for longer than HTPs, regardless of their current use of combustible cigarettes. Nevertheless, dual users displayed the highest aORs for infection and severe disease, suggesting that some effect may have been introduced with combined use.

Finally, we could not exclude the remaining confounding factors related to infection and severe disease, including unknown factors.

Conclusion

In a sample of the Japanese study population in 2022, current tobacco users comprised 24.3%, of which 21.2% were single users of HTPs and 30.1% were dual users of combustible cigarettes and HTPs. Compared to never-users, current users of HTPs (including dual users) were more likely to be infected with SARS-CoV-2, and current or former users of combustible cigarettes and/or HTPs were more likely to require oxygen administration due to COVID-19. Our results are useful in considering the safety of tobacco products use during the ongoing COVID-19 pandemic in the context of the current situation in Japan and, by extension, in other regions where the use of HTPs is expanding.