Design, sample, data collection

The ESPAD survey is the largest quadrennial cross-national project on adolescent substance use in the world, having the overall aim of repeatedly collecting comparable data on substance use among young people in some 35 European countries (www.espad.org). In the 2019 Ireland (Republic) arm of the ESPAD survey, a total of 3565 students aged 15, 16 and 17 years from a nationally representative, stratified random sample of 50 schools were surveyed. Data were collected between March and May 2019 and full accounts of the sampling, data collection and cleaning procedures have been reported elsewhere². Following data cleaning, the final sample comprised 3495 students.

Dependent variable

Prevalence of e-cigarette ever use was measured by the question: ‘Have you ever used e-cigarettes?’, No; Yes, more than 12 months ago; Yes, in the last 12 months; and Yes, in the last 30 days; recoded as ever use, no versus yes.

Independent variables

Sociodemographic, personal, peer and familial characteristics are shown in detail with full question and answer categories in the Supplementary file and we summarize them here.

Sociodemographic variables included age, sex, parental education level, perceived family wealth, and household composition. Variables measuring personal risk behaviors as well as potentially protective behaviors were: use of cigarettes, alcohol, cannabis (including problem cannabis use (CAST [Cannabis Abuse Screening Test], a 6-item, 5-point scale, Cronbach’s alpha 0.83), inhalants, tranquilizers (with prescription); age of smoking and alcohol initiation; problems with social media use (3-item, 5-point Likert scale, Cronbach’s alpha 0.67), internet use (14-item, 5-point Likert scale, Cronbach’s alpha 0.92), online gaming (12-item, 5-point Likert scale, Cronbach’s alpha 0.95), and gambling; missing school due to truancy; reading books other than school books, actively participating in sport, having other hobbies, and average school grade. Variables measuring peer risk activities were: how many of their friends smoke, drink alcohol, get drunk, use cannabis, tranquilizers/sedatives, ecstasy, and inhalants. The variable ‘peer support’ measured friends’ help, support, sharing and communication (4-item, 7-point Likert scale, Cronbach’s alpha 0.94). Familial variables measured familial support (a 4-item, 7-point Likert scale, Cronbach’s alpha 0.92), relationship with parents, and household rules about smoking (smoking regulation).

Statistical analysis

Pearson’s χ² test (for categorical variables) and Student’s t-test (for continuous variables) were conducted to compare differences in e-cigarette use between ever users and never users in relation to the independent variables above (Table 1). All variables in the study were adjusted by the dependent variable (e-cigarette ever use) using Spearman’s correlation coefficient and variance inflation factor (VIF) as appropriate between variables, and a VIF <5 was used to detect multicollinearity. A univariate logistic regression analysis was then performed to assess the association of e-cigarette ever use with the sociodemographic, personal, peer and familial characteristics of respondents. The results are presented in Table 2 as crude odds ratios (ORs) and their respective 95% confidence intervals (95% CIs). This was followed by a stepwise logistic regression analysis to assess predictors of e-cigarette ever use after entering all the variables in the model, and only variables with a p<0.7 were retained in the final model (Table 2). Adjusted odds ratios (AORs) and their 95% confidence intervals were estimated and associations with a p<0.05 were considered statistically significant. All analyses were performed using Stata version 16.

Characteristics of e-cigarette ever users, bivariate analyses

A total of 3495 students were included in the analysis. Sample characteristics of e-cigarette ever use among people aged 15–17 years are shown in Table 1. Overall, 36.6% (n=1278) of students in the sample had ever used e-cigarettes. Girls were more likely than boys to be never users (65.6%, n=1200). In these bivariate analyses, there were significant differences between e-cigarette ever users and never users according to: sex, parental education level, household composition, absenteeism, ever and current smoking, skipping school; ever, current, and binge alcohol use; ever, current, cannabis use and problem cannabis use; ever use of tranquilizers with or without prescription; perceived risks of smoking e-cigarettes; and familial regulation (p<0.05). E-cigarette ever use was higher than cigarette ever use. More than a third (36.6%) of the sample were e-cigarette ever users compared with 31.2% who were ever smokers and 17.3% (n=415) of those who had never tried combustible cigarettes were e-cigarette ever users. Bivariate analyses, shown in Table 2, indicate that respondents’ other risk behaviors – cigarette, alcohol, cannabis, and inhalant use – have the strongest associations with e-cigarette ever use.

Multivariable analysis of e-cigarette ever use

As with the bivariate analyses, multivariable analysis (Table 2) also shows that a respondent’s other risk behaviors have the strongest associations with e-cigarette ever use. Those who had ever tried cigarettes had an AOR of 4.15 (95% CI: 1.29–13.41, p<0.05) for e-cigarette ever use while those who had ever used cannabis had an AOR of 2.21 (95% CI: 1.11–4.41, p<0.05), and those who had ever used inhalants had an AOR of 2.51 (95% CI: 1.07–5.88, p<0.05).

Compared with respondents whose parents were less well-educated, respondents with mothers who had college or university education had significantly higher odds of e-cigarette ever use (AOR=3.46; 95% CI: 1.40–8.54, p<0.05). A small number of variables had significantly lower adjusted odds ratios for e-cigarette ever use, including reading books (excluding schoolbooks) for enjoyment (AOR=0.32; 95% CI: 0.16–0.64, p<0.05), living in a household where some rules or restrictions pertained in relation to smoking cigarettes in the house (AOR=0.53; 95% CI: 0.30–0.94, p<0.05), and perceiving moderate (AOR=0.20; 95% CI: 0.07–0.67, p<0.05) risk in using e-cigarettes even once or twice. Current alcohol use was also negatively associated with e-cigarette ever use (AOR=0.44; 95% CI: 0.20–0.96, p<0.05).

Sociodemographic influences: social class, sex, and household composition

To date, findings about e-cigarette use and social class have been ambivalent. Recent Irish research found that, in a sample in which smoking was patterned by social class, e-cigarette ever use was not²³. This study found that perceived relative wealth was not statistically significantly associated with e-cigarette ever use but that parental education level was. This suggests some differences in teenagers’ views and motivations regarding e-cigarettes compared with cigarettes. The association between smoking and lower socioeconomic status is well-established, but the association with other substances is more ambivalent. For example, young adults with the highest family background SES have been found to be most prone to alcohol and marijuana use, even after adjusting for covariates²⁴. They have also been found to be more likely to use other drugs, and to use alcohol and other substances to cope with stress²⁵. Our findings about increased e-cigarette ever use among people aged 15–17 years with higher-educated mothers (AOR=3.46; 95% CI: 1.40–8.54) may indicate that e-cigarette ever use has more in common with alcohol and other drug use than it has with smoking or that more-educated parents have different attitudes to e-cigarettes than less-educated parents.

Other familial behaviors and supports were also implicated in adolescent e-cigarette ever use. Even after adjusting for covariates, living in a household where some rules or restrictions were in place regarding whether or where people could smoke in the house lowered the odds of people aged 15–17 years e-cigarette ever use (AOR=0.53; 95% CI: 0.30–0.94)

Parental anti-smoking communication and encouragement reduces teenage smoking²⁶. Our findings extend the role of parental influence, including having no-smoking rules in the home, from cigarette smoking to e-cigarette ever use, suggesting an encouraging role for parents in reducing nicotine consumption in teenagers.

Personal behaviors

Polysubstance use is highly prevalent among adolescents who use e-cigarettes²⁷. E-cigarette ever use was strongly associated with ever use of tobacco, cannabis, and inhalants, and the association was especially strong for cigarette smoking. Risk-taking, indicated by experimenting with many substances, may be implicated in e-cigarette ever use and these findings provide some support for the common liability theory¹⁷. We agree²⁸ that e-cigarette screening should include the assessment of other substances, especially cigarettes, alcohol, and cannabis, with a view to identifying and implementing prevention efforts and improving population health.

When we adjusted for covariates in our regression model, reading books for enjoyment remained protective against e-cigarette ever use (AOR=0.32; 95% CI: 0.36–0.51). Perceiving moderate risk in using e-cigarettes is indicated as protective against e-cigarette ever use (AOR=0.20; 95% CI: 0.07–0.67), suggesting a role for health education in providing clear, focused, up-to-date information for adolescents about the risks of e-cigarette ever use. Efforts should be stepped up in the junior cycle of post-primary schooling to develop health education curricula that are appropriate in terms of content, pedagogy, resources and evaluation²⁹.

Peer influences

Studies have shown that adolescents are more likely to engage in risky behaviors in the presence of peers^30,31. In our model, we tested correlations between all peer substance use [use of tobacco, alcohol (including getting drunk), cannabis, tranquilizers/sedatives, ecstasy, inhalants] and e-cigarette ever use and increased odds ratios were noted but data were not sufficiently sensitive to detect differences.

Experimentation, continuation, and cessation

E-cigarette ever use is largely associated with teenagers’ other substance use. Experimentation is widely recognized as a feature of adolescence and others have drawn ‘a positive, linear relation between substance use and psychopathology, such that the more frequently children and adolescents use illegal substances, the greater their risk for exhibiting internalizing or externalizing psychiatric disorders’³², suggesting that complete abstinence among children and adolescence is seen as a desirable outcome. At a minimum, an increased proportion of those who ever use e-cigarettes experimentally will go on to become addicted users of nicotine³³, or possibly dual users of e-cigarettes and combustible tobacco^34,35. Marijuana use has also been found to increase at a faster rate among e-cigarette users when compared to their peers who used cigarettes or a combination of cigarettes and e-cigarettes (dual users)³⁶. Unlike tobacco, e-cigarettes in Ireland are currently largely unregulated. The findings suggest a role for health education highlighting the important link in e-cigarette ever use with peer and personal polysubstance use.

Limitations

The study captures the associations of ever use of e-cigarettes at a point in time of school-going people aged 15–17 years in Ireland. It cannot tell us about the small percentage of students who do not attend school in Ireland at that age and may be different. ESPAD questionnaires are completed by students in a school setting therefore self-reporting bias is a consideration. Social desirability bias is a further consideration given the sensitive nature of the behavior under study – teenage substance use. Also, e-cigarette use is changing rapidly so advice on regulation and control needs regular up-dating. The use of the stepwise regression method produces confidence intervals around the parameter estimates that are too narrow and p-values that are too low due to multiple comparisons. Longitudinal and qualitative studies are certainly needed to improve our understanding and predictions for the future, but these exploratory analyses give us valuable insights in the present situation.