2.1. Data Source
Our study uses data from the Taiwan Adolescent to Adult Longitudinal Study (TAALS), a large nationally representative long-term longitudinal cohort study of health behaviors among adolescents in Taiwan that was conducted between 2014 and 2016 [25]. At baseline, the TAALS survey interviewed students in the first year of middle school and students in the first year of high school. Since our participants are minors (age less than 16 years), informed consent form was collected from both a parent/or legal guardian and the participant for study participation. During the first wave of the formal cohort study, 6903 middle school students and 11,742 high school students were interviewed, for a total of 18,645 students. During the second wave, the same cohort of students was re-interviewed in their third year of middle school and third year of high school. A total of 16,265 students were interviewed (6381 middle school students and 9884 high school students), representing a follow-up completion rate of 87.21%. After excluding participants with incomplete records, a final analysis was conducted on 14,109 students. Among those 14,109 students, 474 students were current smokers at the baseline, of whom 331 students used only combustible cigarettes, while the remaining 143 students used both combustible cigarettes and e-cigarettes. Among the 474 students who were current smokers, 362 students were still current smokers and the remaining 112 students had quit smoking at follow-up (See Figure 1). A more complete description of the TAALS Cohort and methodology is described in one of our previous publications [25]. This study was approved by the Joint Institutional Review Board of Taipei Medical University, Taiwan (TMU-JIRB-201410043).
2.2. Measures
2.2.1. Definitions of Smoking Cessation and Changes in Total Monthly Combustible Tobacco Consumption
Two main variables were being examined: smoking cessation and total monthly combustible tobacco consumption. The following methods were used to measure smoking cessation. First, participants were being asked “Have you ever smoked a cigarette?” and if they answered yes, they were being asked “In the last 30 days how many days have you smoked cigarettes?”. Participants who had smoked any number of cigarettes in the last 30 days were defined as current smokers. At follow-up, participants were identified as either current smoker or quit smoking. Smoking cessation was defined as participants who indicated that they were current smokers at baseline, but then indicated that they hadn’t smoked any cigarettes in the last 30 days at follow-up.
The following self-reported questions were used to measure total monthly combustible tobacco consumption: 1. “In the last 30 days, how many days did you smoke cigarettes?”, and 2. “In the last 30 days, on the days you smoked cigarettes, on average how many cigarettes did you smoke each day?”. From these questions, we measured the number of smoking days per month and average number of cigarettes smoked per smoking day. We multiplied the average number of cigarettes smoked per smoking day by the number of smoking days per month to calculate the total monthly combustible tobacco consumption. Next we subtracted the total number of cigarettes smoked each month at follow-up from the number of cigarettes smoked each month at baseline to calculate the change in total monthly combustible cigarette consumption.
2.2.2. Independent variables at baseline: current e-cigarette use and current use of other tobacco products
In this study, we asked participants at baseline if they are current users of e-cigarettes or other tobacco products. We measured current e-cigarette use by asking: “In the last 30 days how many days have you used e-cigarettes?” and participants responding between 1–30 days, were classified as current e-cigarette users. We measured current other tobacco products use by asking “In the past 30 days have you used other tobacco products (such as: cigars, cigarillos, pipes or water pipes)?”
2.2.3. Demographic characteristics variables
This study also recorded the following demographic characteristic variables for all participants: sex, depression, peer support, age, father’s highest education level (junior high or below, high school, university or higher), mother’s ethnicity (Han Chinese, Aboriginal Taiwanese, or New Immigrant), father’s occupation status (full-time, part-time, unemployed), and family living arrangement status (living with both mother and father, living with both parents and grandparents, living with a single parent, living with only grandparents, and living with someone other than direct kin).
2.3. Statistical Analysis
The logistic regression was used to evaluate the causal inference between the use of e-cigarettes and future smoking cessation among adolescents. To ensure our analytic results remain nationally representative of adolescents in Taiwan, we used weighted values to conduct weighted adjustment. For a detailed description on how weighted adjustment was performed, please refer to Chein et al. [25]. We used risk ratio (RR) and risk difference (RD) instead of odds ratio (OR) because the use of RR and RD can better reflect the relative and absolute probability change in smoking cessation in the presence of e-cigarettes. Here the adjust risk ratio (aRR) and adjust risk difference (aRD) were performed using the “adjrr” command in STATA statistical software (version 16 MP, StataCorp LLC)[26]. P-value of < 0.05 was considered significant.