All study participants will receive brief advice to quit vaping and an introduction to free, anonymous, text-based vaping cessation support through the Truth Initiative (This is Quitting [TIQ])¹ reported to have efficacy for promoting abstinence in this population (44, 45).

Participants randomized to V+BC and P+BC will receive generic varenicline or identical appearing placebo. In adolescents aged 12–17 years, steady-state systemic exposure in those who weighed >55 kg was comparable to adults at the same dose; however, steady-state daily exposure of 0.5 mg bid of varenicline was, on average, higher (by approximately 40%) in adolescents who weighed ≤55 kg compared to adults (32). Therefore, for the current study, participants ages 16–17 years old (≤ 55 kg) will take one 0.5 mg dose once daily for the first 7 days before taking one 0.5 mg dose twice daily for 11 weeks. For participants 16–17 years old (>55 kg) and participants 18 years and older, regardless of weight, they will take one 0.5 mg dose once daily for 3 days, one 0.5 mg dose twice daily for 4 days, and then one 1.0 mg dose twice daily for the remaining 11 weeks.

Medications will be adjusted to a lower dose if 1.0 mg bid is not well tolerated. Efficacy of doses lower than 0.5 mg bid have not been reported for any population, thus participants who do not tolerate the dose of varenicline or placebo 0.5 mg bid will be discontinued from medication, but retained in the study and included in the intent-to-treat analyses. Study staff will distribute varenicline or placebo with instructions on how to take the study medication as follows: 2-week supply at baseline, 2-week supply at week 2, and 4-week supplies at weeks 4 and 8.

Participants randomized to V+BC and P+BC will be enrolled in a novel, 12-week manualized behavioral counseling intervention (Quit Vaping) based on the American Lung Association smoking cessation program (29, 48–50), modified to be adolescent-friendly and relevant to vaping cessation. The intervention contains 12 modules, each of which is designed to be delivered by lay/non-clinically trained personnel in a 20 min session and can be delivered either via videoconference or in-person (see Table 3). Content is delivered with psychoeducational videos, guided discussions, and take-home worksheets to consolidate information and provide opportunities for skill building. Example module topics include: identifying pros and cons of vaping, industry tactics, refusal skills, and establishing a healthy lifestyle. All sessions will follow the same general structure, including: 1. check-in on adverse events, vaping behavior, and medication adherence (approximately 5–10 min); 2. presentation of new educational content (via pre-recorded psychoeducational videos; approximately 3 min); and 3. review of educational material and relevance to the participant (approximately 7–10 min). Participants will be asked to set a quit date before Session 3 and will work with the counselor during Sessions 1 and 2 to create a detailed quit plan. All session materials will be provided on the intervention website,² along with additional supplementary resources for the participant to access.

Quit Vaping counselors will complete the Tobacco Treatment Specialist (TTS) training program.³ Counselors will also receive 45 min of weekly supervision from a doctoral-level clinical psychologist on study staff for the first six months of the study with monthly supervision thereafter. Supervision will provide opportunities for case consultation as well as information on vaping behavior and cessation in adolescents, study methods, as well as the theoretical rationale and skills relevant to the behavioral intervention.

The primary outcome is cotinine-verified, continuous nicotine vaping abstinence, operationalized by self-report of no nicotine vaping since the last study visit and salivary cotinine <30 ng/mL (51) at each visit from study week 9 to end of treatment (week 12) (primary) and from study week 9 to the end of follow-up (week 24; secondary). Self-reported vaping will be assessed at all visits using the timeline follow-back (TLFB) (52, 53). A 60-day recall period will be queried at enrollment, and a modified TLFB will be administered at subsequent visits to ascertain vaping in the period between visits. Saliva samples will be collected weekly during the treatment phase and then monthly during follow-up for qualitative cotinine measurement. Study staff will mail collection kits in advance of virtual visits to participants, and participants will provide the sample during the videoconferencing session and show results to the assessor. Those who report no vaping since the last study visit but who have a cotinine >30 ng/mL will be considered non-abstinent at that visit for analyses.

Using TLFB procedures and salivary cotinine assays, we will assess for differences in 7-day point-prevalence vaping abstinence at end of treatment. Point prevalence abstinence is operationalized as self-report of no vaping in the prior seven days and salivary cotinine <30 ng/mL at week 12.

We will assess for group differences in frequency and severity of nicotine withdrawal, craving, and negative mood, as well incidence and severity of adverse events. Nicotine withdrawal will be assessed at all study visits with the Minnesota Withdrawal Scale (MNWS) (54), a 9-item scale of nicotine withdrawal symptoms (i.e., craving, irritability, anxiety, difficulty concentrating, restlessness, increased appetite or weight gain, depression, and insomnia) reported on an ordinal scale from 0 (not at all) to 4 (extreme). Craving will be assessed at all study visits using the Questionnaire of Vaping Craving (QVC) (55), a 10-item measure of vaping craving that assesses desire and intent to vape and anticipation of positive outcomes related to e-cigarette use. Negative mood will be assessed at enrollment, baseline and monthly study visits with the Mood and Anxiety Symptoms Questionnaire (MASQ-D30) (56), a 30-item adaptation of the original 96-item MASQ (57, 58), which instructs participants to rate how often in the past week they have experienced symptoms of depression and anxiety on a 5-point Likert scale, with 1 being “not at all” and 5 being “extremely.” Adverse events will be assessed at enrollment, baseline and monthly study visits with the Neuropsychiatric Adverse Events Interview (NAEI) (27), a semi-structured interview designed to assess the presence and severity of symptoms such as irritability and suicidal thoughts or behavior.

The TLFB interview will be used to assess frequency and amount of cannabis, alcohol, and illicit substance use 60 days prior to enrollment as well as interim periods at baseline, monthly study visits, and post-intervention follow-ups. Self-report of combustible tobacco use will be biochemically verified with urine NNAL (4- (methylnitrosamino)-1-(3-pyridyl)-1-butanol) at end-of-treatment and follow-up, with <10 pg./mL indicating no combustible tobacco use in the past two months (59). NNAL presence in urine is indicative of exposure to cured tobacco products and as such is not present with vaped nicotine.

Missing or incomplete outcome data from study week 9 to 12 will be imputed 40 times using multiple imputation via chained equations (60). Outcome data will be imputed using three predictors collected at the enrollment visit: biological sex (male versus female), environment (secondary school versus post-secondary school), and severity of nicotine dependence (ECDI, summed score ranging from 0–20) (61). Imputation will be done via fully conditional specification with predictive mean matching using linear models for continuous data and logistic regression for binary variables.

We will use a logistic regression applied to the binary outcome for 4-week continuous abstinence at end of treatment (weeks 9–12; primary) and end of follow-up (weeks 9–24) (1 = abstinent, 0 = not abstinent). Participants whose observed continuous abstinence status cannot be determined due to missing data will have continuous abstinence status determined via imputation. We will include sex, environment (secondary school versus post-secondary school), and severity of nicotine dependence (ECDI) as baseline covariates which will be standardized for analyses. The analysis will be intent-to-treat. We will then apply our logistic regression model to each set of observed and imputed data, and pool the regression estimates according to Rubin’s rule. Statistical significance for the contrast between V+BC and P+BC will be determined using a z-test applied to the pooled mean estimate divided by the pooled standard error. The same analyses will be applied for secondary hypotheses comparing V+BC and P+BC vs. EUC on 4-week continuous abstinence at end of treatment and follow-up as well as all groups on seven-day point-prevalence vaping abstinence at end of treatment and follow-up.

We will use a linear model applied separately to repeated, continuous weekly assessments of nicotine withdrawal symptoms, craving, and mood symptoms. The linear model will be estimated using generalized estimating equations, providing estimates and standard errors robust to violations of distributional assumptions (e.g., normality) and heteroscedasticity. The model will be applied to the summed scores at each time point per participant. The analysis will be intent-to-treat with missing data imputed. The key confirmatory effect will be a dummy-coded contrast comparing placebo (P+BC; coded as 0) against varenicline (V+BC; coded as 1). Exploratory analyses will also be conducted to evaluate V+BC vs. EUC and P+BC vs. EUC. The model will include covariates for change over time (two terms encoding linear and quadratic trends), baseline levels of the outcome variable, and sex, environment (secondary school versus post-secondary school), and severity of nicotine dependence. Covariates will be standardized. p-values will be adjusted to avoid multiple comparison issues using the Benjamini-Hochberg method.

We will conduct a Fisher’s exact test to evaluate differences between the V+BC and P+BC groups and V+BC and EUC groups in the number (%) of participants who endorse an adverse event at least once in the 12-week treatment period. We will report frequent adverse events (e.g., that occur in at least 5% of the participants at any point between weeks 1 and 12 of the trial) by group.

We will test whether our conclusion on efficacy is robust to the inclusion/exclusion of the three covariates [biological sex, environment (secondary school vs. post-secondary school), and nicotine dependence severity]. To test whether our conclusion on efficacy is robust to the approach for imputation of missing data, we will rerun the adjusted (i.e., with covariates) models on the observed data only and will rerun the adjusted models on a dataset with the observed data and all missing outcome data imputed as non-abstinent. Finally, we will conduct a per protocol analysis by evaluating results among participants who use varenicline on most days in the trial (e.g., >50 or 75%).

All analyses will be conducted using the statistical software R (version 4.1.1) (62). Multiple imputation will be carried out using the R package ‘mice’ (version 3.13.0) (60). The generalized estimating equations method will be implemented via the R package ‘geepack’ (version 1.3-2) (63).