Trajectories of Nicotine and Cannabis Vaping and Polyuse From Adolescence to Young Adulthood

This cohort study assesses the developmental trajectories of nicotine and cannabis vaping from late adolescence to young adulthood in Southern California.


Introduction
The prevalence of electronic vaporizer use among US adolescents and young adults has substantially increased. Recent past 30-day estimates indicate marked increases for both nicotine vaping (20.9% to 25.4% among 12th graders from 2018 to 2019; 6.5% to 10.6% among young adults from 2017 to 2018) and cannabis vaping (7.5% to 14.0%% among 12th graders from 2018 to 2019; 6.6% to 9.3% among young adults from 2017 to 2018). [1][2][3] There is a wide distribution in the frequency of past 30-day use among youths, ranging from vaping 1-2 days to daily use 4,5 ; however, the extent to which this wide distribution represents individuals on escalating, deescalating, or stable use trajectories is unclear.
Growth mixture modeling (GMM) is a data-driven analytic approach for identifying unobserved subpopulations and describing distinct longitudinal change. 6,7 This approach has been applied to identify youth trajectories of combustible cigarette, alcohol, and cannabis use. [8][9][10] However, to date, only 2 longitudinal studies have sought to identify trajectories of nicotine vaping. Park et al 11 identified 3 e-cigarette trajectories from 13 to 17 years of age: never (66.6%), low and increasing (20.1%), and high and increasing (13.3%). Westling et al 12 reported 2 trajectories of e-cigarette use from 8th to 9th grade: 94.8% infrequent or no use and 5.1% accelerated use. Both studies indicated that membership in e-cigarette-using trajectories was associated with other substance use.
Although Park et al 11 and Westling et al 12 showed that adolescents can be classified into distinct trajectories of vaping, neither study examined the transition from adolescence to young adulthood. This is a population at high risk because transitions to college and/or the workforce and increased familial and financial responsibility are associated with increased risk of polysubstance use, enduring substance use problems, and substance use disorder. [13][14][15][16] Furthermore, to our knowledge, developmental trajectories of cannabis vaping have not been identified in adolescence or young adulthood; thus, it is unknown how cannabis vaping develops over time. Examining gender and racial/ethnic differences of nicotine and cannabis vaping trajectories is also warranted because there is some evidence that males are more likely to vape than females 17,18 ; racial/ethnic differences are less clear. 19,20 In addition, although cross-sectional studies have reported high rates of nicotine and cannabis vaping polyuse among adolescents and young adults, [21][22][23][24] to our knowledge, no longitudinal study has examined co-occurring development of nicotine and cannabis vaping trajectories.
Identifying common polysubstance vaping patterns may inform both nicotine and cannabis policy and prevention.
Using a prospective longitudinal design following a cohort of adolescents through young adulthood (Ն18 years of age), the current study evaluated nicotine and cannabis vaping trajectories, demographic covariates of trajectories, and co-occurrence of nicotine and cannabis vaping trajectories.
years after high school (October 2018 to October 2019) through online questionnaires. Questions pertaining to specific substance vaped were first asked at wave 5; thus, the current study assessed survey data from waves 5 to 9. All 9th graders at wave 1 were eligible to participate. This study was approved by the University of Southern California institutional review board. For waves 1 to 8, written or verbal parental consent and student assent were obtained. For wave 9, participants were contacted after turning 18 years of age and provided written informed consent. This study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) and American Association for Public Opinion Research (AAPOR) reporting guidelines.

Measures Nicotine and Cannabis Vaping
Frequency of nicotine vaping and cannabis vaping was assessed across waves 5 to 9. Participants indicated the number of days of nicotine and cannabis vaping in the past 30 days by answering the following 2 questions. respectively: "In the last 30  Covariates Age, highest parental educational level, gender, and race/ethnicity were self-reported. Highest parental educational level was recoded into a binary variable (college degree or higher vs some college or less). Race/ethnicity was recoded into 3 dummy variables (1, Asian; 0, non-Asian; 1, Hispanic or /Latino; 0, not Hispanic or Latino; 1, White; 0, non-White) for racial/ethnic groups representing 10% or more of the sample (Asian, Hispanic or Latino, and White).

Statistical Analysis
Growth mixture modeling (GMM) was used to identify latent trajectories of nicotine and cannabis vaping. This approach captures heterogeneity within the population by identifying different growth trajectories of a latent variable based on unique intercepts and slopes. 27 An increasing number of classes are estimated until an optimal model is identified. Similar to structural equation modeling in which statistical indexes (eg, confirmatory fit index, root mean square error of approximation) are used to identify the best-fitting model, the bayesian information criterion 28 and Lo-Mendell-Rubin likelihood ratio test 29 are commonly used indexes for GMM. Owing to the large number of 0s resulting in skewed distributions for past 30-day nicotine and cannabis vaping, vaping variables were treated as count outcomes; a zero-inflated Poisson model was used to account for both processes generating 0 days scores (vape users who did not use in the past 30 days and never users). Full information maximum likelihood was used to account for missing data, which allowed participants with at least 1 time point of data to be included in trajectory analyses (3322 of 3396 in the original cohort). Simulation studies report samples near 1200 as adequate for complex GMM. 30 Covariates of each trajectory model were evaluated within the GMM framework using a validated 3-step approach to account for classification error. 31 Parallel growth mixture modeling (PGMM) assessed co-occurrence of nicotine and cannabis vaping trajectories. PGMM estimates the unique developmental growth parameters of 2 distinct processes and how each process relates to the other across time. 32 PGMM was used to calculate the probability of cross-classification between a specific nicotine vaping trajectory and cannabis vaping trajectory. A 2-sided P < .05 was considered to be statistically significant. Analyses were conducted with Mplus, version 8.4. 33

Descriptive Statistics
A total of 4100 students were eligible to enroll in the study; parental consent and student assent were obtained for 3396 adolescents (82.8%). Of the 3396 participants originally enrolled in the study, those with at least 1 time point of past 30-day vaping data from waves 5 to 9 were included in the analysis (n = 3322). Data distinguishing between nicotine vaping and cannabis vaping were available from waves 5 (fall of 11th grade) to wave 9 (young adulthood). Data were missing for 134 (4.0%) students in wave 5, 257 (7.7%) in wave 6, 179 (5.4%) in wave 7, 213 (6.4%) in wave 8, and 839 (25.3%) in wave 9.
For cannabis vaping, any use in the past 30 days steadily increased from 134 (4.2%) participants in

Cannabis Vaping Trajectories
Similar to the nicotine vaping trajectories, the bayesian information criterion and Lo-Mendell-Rubin likelihood ratio test indicated that the 5-class model was optimal (eTable 2 in the Supplement). The identified trajectories for cannabis vaping were similar to those identified for nicotine vaping (Figure 2)

Covariates of Nicotine and Cannabis Vaping Trajectories
Covariates were added to each vaping trajectory model to assess the odds of trajectory membership based on age, highest parental educational level, gender, and race/ethnicity (

Co-occurring Nicotine and Cannabis Vaping Trajectories
PGMM was used to assess co-occurring trajectories of nicotine and cannabis vaping. Before examining conditional probabilities of membership in the cannabis vaping trajectories given membership in the nicotine vaping trajectories (Table 3), an overview of classification across both sets of trajectories indicated that 57.6% belonged to nonusers nicotine and cannabis vaping trajectories, 7.5% were classified into a nicotine-use but not a cannabis-use vaping trajectory, 9.8% were classified into a cannabis-use but not a nicotine-use vaping trajectory, and 25.1% belonged to both nicotine-use and cannabis-use vaping trajectories. As shown in   Those in the adolescent-onset escalating frequent users nicotine vaping trajectory had the highest probability of membership in the adolescent-onset escalating frequent users cannabis vaping trajectory (45.5%).

Discussion
The current study corroborates prior research showing associations between nicotine and cannabis vaping in adolescence. 21 For both nicotine and cannabis vaping, 5 trajectories distinguished between frequency and developmental timing of use, which is consistent with previous studies that identified adolescent e-cigarette trajectories characterized by frequency and increasing use. 11,12 Of note, the proportion of the sample in each trajectory and type of trajectory were similar between nicotine and cannabis vaping. These similarities suggest that developmental trajectories of nicotine and cannabis vaping may share underlying risk processes, which has been observed in polyuse of combustible tobacco and cannabis use. 34,35 In this study, males (vs females) had close to double (cannabis) and triple (nicotine) the odds of belonging to the adolescent-onset escalating frequent users trajectory compared with the nonusers trajectory. These results align with previous findings indicating that male adolescents have a higher likelihood of vaping than female adolescents 17,18 and suggest that males who begin vaping in adolescence may be more likely than females to escalate their intake substantially. Race/ethnicity was not associated with membership in similar nicotine and cannabis vaping trajectories, but Latino individuals (vs non-Latino individuals) had lower odds of belonging to the adolescent-onset escalating frequent users nicotine trajectory than to the nonusers trajectory.
As public health initiatives expand from their focus on nicotine vaping to other vaped substances, such as cannabis, understanding shared processes and pathways to polysubstance vaping is critical to appropriately tailor public health and clinical interventions to those most at risk.

Limitations
This study has limitations. The use of a sample specific to Southern California limits generalizability to other regions; however, a regionally specific sample increases the likelihood that participants were experiencing similar regulatory policies and trends in vaping use. The study also relied on self-report of vaping use, but self-report is the most common method of measuring substance use. Although the sample was racially/ethnically diverse, evaluating racial/ethnic differences between Black individuals and other ethnicity/race participants was not feasible owing to low representation. There were more missing data in wave 9 (young adulthood) compared with waves 5 to 8 (adolescence); however, full information maximum likelihood enabled participants with at least 1 wave of data to be included in the analysis. Although the study examined the transition from adolescence to young adulthood, additional waves in young adulthood would better inform young adult vaping use.

Conclusions
In this study, nicotine and cannabis vaping trajectory models were similar; this, additional work appears to be needed to evaluate shared risk processes. A significant proportion of individuals initiated and participated in both nicotine and cannabis vaping during young adulthood, suggesting that research is warranted to identify developmentally appropriate interventions. Further study of the substantial polysubstance vaping observed between nicotine and cannabis vaping trajectories appears to be needed to develop more effective regulatory practices and interventions.