Use of oral moist tobacco (snus) in puberty and its association with asthma in the population-based RHINESSA study

Study design, population and sample

For this study, we cross-sectionally analysed the data from the multicentre population-based Respiratory Health in Northern Europe, Spain and Australia study cohort (RHINESSA), carried out from 2013 to 2016. The cohort includes study participants of at least 3 generations from 10 centres: Norway (Bergen), Denmark (Aarhus), Sweden (Gothenburg, Uppsala and Umea), Iceland (Reykjavik), Estonia (Tartu), Spain (Albacete and Huelva) and Australia (Melbourne). These are the offspring of the participants of two population-based cohorts: the European Community Respiratory Heath Survey (ECRHS; www.ecrsh.org) and the Respiratory Health In Northern Europe (https://rhine.w.uib.no/) (detailed information elsewhere21).

Data were obtained from 10 036 participants from the ten centres that participated in the baseline RHINESSA study (RHINESSA third generation). Information about the history of consumption of tobacco-containing products was obtained using detailed questionnaires. Eligible participants were those older than 15 years of age with information on tobacco-consumption habits. We excluded 608 participants who were younger than 15 years and 426 participants from three centres (Albacete and Huelva in Spain and Melbourne in Australia) with no information about snus. Thus, the analyses included 9002 participants, as shown in figure 1.

Covariates

We used Directed Acyclic Graphs (DAG) to identify potential confounders and effect modifiers (online supplemental figure S1) in the association between age of starting snus use and current asthma and asthma symptoms. The DAG included all the covariates that were considered potentially important for the models and that were supported by scientific evidence.

Time of starting tobacco smoking was defined by using the following questions: (1) Are you a smoker? (2) Are you an ex-smoker? If yes to (1) or (2): (3) Age when started smoking; and if yes to (2): (4) Age when stopped smoking. The time of smoking initiation was categorised as below:

1. Never tobacco smoking (‘no’ to 1 and 2).

2. Tobacco smoking starting before or at age 15 years (‘yes’ to 1 and/or 2 and ‘≤15 years’ to 3).

3. Tobacco smoking starting after age 15 years (‘yes’ to 1 and/or 2 and ‘>15 years’ to 3).

Tobacco pack-years were determined by considering the daily cigarette consumption (as per the question ‘How many cigarettes per day?’) and the number of years smoking (as per the question ‘How many years have you smoked?’). Alternatively, the calculation could be based on the age at which participants began and ceased smoking (questions 3 and 4 above).

Statistical analyses

The characteristics of the study population were described by sex and age of starting snus. The data were clustered by study centre and families, therefore, we analysed the associations between age of starting snus with current asthma and three or more asthma symptoms using multivariable multilevel (three levels and random intercept)23 logistic regression models. As for the analyses of the association of age of starting snus with the asthma symptom score (count), a negative binomial regression model was used given the overdispersed distribution of the data with the mean (0.98) being lower than the variance (2.33).24 We addressed missing values by only including participants with complete information on snus-consumption habits, without carrying out imputation of data. Only 184 persons out of 9002 were excluded from the regression models because they did not give information on age of onset of snus use (random loss of data). We made adjustments, as illustrated in the DAG in online supplemental figure S1, by considering age, the age when tobacco smoking started and the pack-years of tobacco smoking. The two latter were included given the plausible influence of the debut of tobacco smoking and intensity of smoking on the initiation of snus use at earlier or later age, as some participants may transition from smoking to using snus. Moreover, as tobacco smoking is widely recognised to be associated with respiratory symptoms, these factors were also taken into account.

In addition, we explored the potential interaction between sex and the age of starting snus use in relation to asthma and asthma symptoms. All models were stratified by sex for a comprehensive analysis. We conducted sensitivity analyses among participants who have never smoked and among those with asthma starting after age 15 years (no early-onset asthma—before age 15 years). By performing these sensitivity analyses, we excluded those participants with well-known risk factors for our outcomes.

Furthermore, we conducted an additional analysis to estimate the association of age of starting snus use and asthma with hay fever. Lastly, we investigated the association, considering reverse causality, between having experienced early-onset asthma (further categorised as having had asthma at ages ≤10 years and 11–15 years) and reporting use of tobacco products (categorised as ever snus use and ever tobacco smoking).

We considered an association to be significant if the estimates in the adjusted models had a conventional p≤0.05. For the inclusion of an interaction term or stratification, a less conservative value of ≤0.1 was employed as a cut-off. The statistical analyses were performed by using Stata V.17 (StataCorp).

Patient and public involvement

Patients and the public were involved in several stages of the baseline RHINESSA study on which this analysis is based: patients were involved through a representative of the patient organisation, the Norwegian Asthma and Allergy Foundation and the public was represented through a user representative from the study participants and through stakeholder organisations; all representatives were invited to the annual meetings of the RHINESSA study. For more information, please refer to the official RHINESSA website (www.rhinessa.net) and in Svanes et al.21

Main characteristics of the study sample (demographics and tobacco-consumption habits)

The general characteristics of the 9002 participants by sex and age of starting snus can be seen in table 1. Overall, women represented 58% of the sample and 1839 (20%) of the participants reported ever having used snus. The median age at participation was 28.3 years (range 15–53 years) and the majority of participants had normal BMI, attained college/university education and reported that they were engaged in frequent physical activities.

Snus use was most common in the three Swedish centres (Gothenburg, Umea and Uppsala) and in Bergen in Norway (24%, 34%, 23% and 26%, respectively) (figure 2A). The same study centres had the highest prevalence of starting snus ≤15 years of age, ranging from 4.5% to 10.0% (figure 2B). Men were more likely to use snus (29%) and to start using snus early (26%) as compared with women. One-third of the cohort reported ever smoking, with a similar proportion in men and women. With regard to asthma outcomes, both current asthma and asthma symptoms were more commonly reported in women from all study centres, except for Tartu, Estonia, where asthma was more prevalent in men. Participants with increasing BMI showed a trend of reporting more asthma, with this trend being more noticeable in women than in men (online supplemental table S1).

Association between age of starting snus use and asthma outcomes

The prevalence of current asthma and three or more asthma symptoms according to snus habits is presented in figure 3 for men and women. In female participants, those who had initiated snus use early had the highest prevalence of both current asthma (18.0%) and three or more asthma symptoms (29%) (figure 3A). In male participants, only a weak pattern was indicated. When restricted to participants who reported never smoking, higher prevalence of asthma outcomes in those starting snus use early was suggested in both men and women, most pronounced in women (figure 3B).

The adjusted models showed consistent positive associations between early snus use initiation with all three asthma outcomes, evident in both men and women. However, the associations were strongest in women (table 2). In female participants, the risk of having three or more asthma symptoms was almost tripled (adjusted OR 2.70; 95% CI 1.46 to 5.00) and the asthma symptom score was 0.35 units higher (adjusted β-coefficient (β) 0.35; 95% CI 0.07 to 0.63). The estimates for the same outcomes were only indicated in male participants (OR 1.23; 95% CI 0.78 to 1.94; β 0.16; 95% CI −0.06 to 0.38, respectively). The association between early snus use initiation and reporting three or more asthma symptoms was statistically significantly stronger in women than in men (p value for interaction between sex and early snus use initiation=0.023). Evidence of associations between early snus use and current asthma was indicated in both sexes (OR 1.72; 95% CI 0.88 to 3.37 in women; OR 1.31; 95% CI 0.84 to 2.06 in men). Having started using snus after 15 years of age was not consistently associated with asthma outcomes (table 2). Among participants without smoking history, the results were generally consistent, but the observed associations showed somewhat higher estimates in women (table 2).

Sensitivity analyses and test for reverse causality

When excluding participants with early-onset asthma, the outcomes remained crudely consistent, but the sex difference was less clear (online supplemental table S2). We further analysed the associations of snus use with two different phenotypes of asthma, ‘allergic asthma’ is defined as asthma with hay fever, and ‘non-allergic asthma’ is defined as asthma without hay fever. The results were consistent for both asthma phenotypes, with increased OR for both allergic and non-allergic asthma as associated with early snus use initiation among women (online supplemental table S3).

Finally, analyses on possible reverse causality for early-onset asthma with the ever use of snus and/or smoking tobacco showed no clear tendency for participants to report neither higher nor lower use of these tobacco-containing products (online supplemental table S4).

Statement of principal findings

We investigated whether the age of starting the use of snus, a nicotine-containing tobacco product, was associated with asthma and asthma symptoms, in a cross-sectional analysis in 9002 participants of a large multinational population-based cohort. We found that starting snus use before or at age 15 years was associated with more asthma and asthma symptoms, most pronouncedly in female participants. The results were consistent across several measures of asthma and in participants who reported never smoking. Associations with asthma outcomes could not be identified for snus use starting after age 15 years. Thus, our analyses suggest that using snus early in puberty may harm respiratory health more than if such use starts at a later age, independent of tobacco smoking.

Study findings in context

Strengths and limitations of this study

This study used data from a large multicentre population-based cohort, with a high prevalence of exposure to snus, ensuring robust statistical power to detect associations. We characterised the age of snus use initiation, considering covariates crucial to disentangle the relationship with asthma (eg, age of starting smoking and pack-years). Importantly, we addressed the impact of smoking tobacco through sensitivity analysis among participants who have never smoked, yielding consistent findings. The asthma definitions used were previously validated and each grasps different aspects of the disease; the asthma symptom score, in particular, has proved to be more specific,22 29 thus minimising potential misclassification biases (eg, diagnostic bias). This outcome, in addition, increases the statistical power of the analysis. The fact that the analyses using the different definitions show consistent results, strengthen the support for our hypothesis. Finally, we did not find any clear tendency for reverse causality of our observations, and we were able to analyse a clear temporal link between early initiation of snus use and respiratory outcomes, by including only outcomes being present in the last 12 months, therefore, enhancing the associations of early snus use.

The cross-sectional design, however, introduces some limitations, including misclassification of snus use and age of initiation due to self-reports, the stigma associated with its use, the stipulated legal age for snus consumption, legal restrictions on selling tobacco-containing snus in some of the countries where some study centres (eg, Aarhus, Reykjavik and Tartu) were located30 and the asthma status of the participants. In addition, recall bias about age of initiation may be another misclassification source, however, we expect it to be minor given the evidence supporting strong validity of self-reported smoking habits,31 which could potentially be similar to self-reports of snus use habits. Any misclassification, if present, is likely non-differential and may lead to underestimation of the tested associations. Other limitations include insufficient data on snus consumption amounts, which could impact dose–response associations, and potential residual confounding by factors like dual or intermittent use of snus and tobacco smoking. Finally, although the analyses included the main known covariates identified by DAG that could influence the associations, there may possibly be other factors that we have not accounted for, and unknown confounding cannot be ruled out.

Implications

Our results show that snus use is popular in Northern Europe, even in countries without legal protection for snus trade such as Estonia, Denmark and Iceland.30 These findings add substantial knowledge to the existing by showing that early initiation of snus use is widespread and that there are indications it may increase the risk of asthma. This study highlights the susceptibility of puberty, during which consumption of snus could significantly influence respiratory health, particularly among young women. Additionally, it cannot be ruled out that other nicotine-containing products, such as e-cigarettes and nicotine pouches, may have similar effects as what we have found for snus, carrying important common public health implications. This information can guide future preventive interventions for high-burden diseases like asthma in settings where oral moist tobacco use is increasingly common. It also calls for attention from healthcare professionals and policymakers to address the alarming trend of early-life use of alternative tobacco and nicotine products.