Abstract
Aims
Most extant evidence has addressed between‐person differences, short‐term or cross‐sectional associations of electronic nicotine delivery systems (ENDS) use with other substance use, the majority focusing on current rather than escalated use. The present study aimed to examine within‐person changes in escalated ENDS use and their associations with individual and combined substance use over a 6‐year period.
Design, Setting and Participants
This study used a longitudinal cohort design with US young adults. A generalized linear mixed‐model approach was employed to fit a series of weighted logistic regression models. Data were drawn from waves 1–5 of the Population Assessment of Tobacco and Health (PATH) study in the United States. Of the 9110 young adults at baseline, aged 18–24 years, a total of 5042 individuals had matched data across all five waves of assessments.
Measurements
Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated, if the value was greater than zero (otherwise, coded as 0 = not escalated).
Findings
Escalated ENDS use gradually decreased over time, with the lowest prevalence at wave 4 (4.0%) but sharply increasing at wave 5 (8.4%). Escalated ENDS use was associated with increased odds of using each substance (binge drinking, marijuana use, marijuana vaping, prescription and illicit drugs) and different combinations of polysubstance use between cigarette smoking, binge drinking and marijuana use (Ps < 0.05). In addition, sweet/fruit flavor use (versus menthol/mint) was associated with increased likelihood of reporting co‐use of cigarettes and marijuana.
Conclusions
In the United States, the prevalence of young adults using electronic nicotine delivery systems appears to have increased steadily between 2013 and 2019, although the rate of increase may have started to accelerate in recent years. Escalated electronic nicotine delivery systems use and time‐lagged established electronic nicotine delivery systems use appear to be prospectively associated with individual and combined substance use, particularly between cigarettes, alcohol and marijuana. Among established electronic nicotine delivery systems users, sweet/fruit flavor appears to be associated with increased risk of co‐using cigarettes and marijuana.
Keywords: Electronic nicotine delivery systems, escalated use, polysubstance use, tobacco, within‐person design, young adults
INTRODUCTION
Electronic nicotine delivery systems (ENDS) use in young adults in the United States has significantly increased in recent years [1]. In 2017–19, the prevalence of past 30‐day ENDS use in college students and non‐college‐attending peers in the United States increased 6–22% and 8–18%, respectively [1]. Young adulthood involves various social transitions [2] and is often marked by increased exposure to substance use, including ENDS [3].
Evidence indicates that ENDS use in young adults is associated with the use of combustible tobacco products or other substances such as marijuana [4, 5, 6, 7], alcohol [4, 8, 9] and prescription or illegal drugs [9, 10]. However, most extant evidence is based on cross‐sectional data [4, 9, 10], and longitudinal studies of between‐person differences assess spans of a year or two [11, 12, 13] or associations with individual substances [6, 12, 13]. These findings may be useful in providing snapshots of ENDS use behaviors at different time‐points or revealing individual factors that co‐vary with ENDS use behaviors. However, such information rarely captures the temporal dynamics of substance use behaviors and provides little insight into plausible pathways of substance co‐use. The landscape of tobacco use is rapidly changing (e.g. disposable ENDS replacing JUUL as the most popular device), and there is a lack of data concerning within‐person multi‐year dynamics related to ENDS use in young adults.
Within‐subject design is a statistical method in which study participants are exposed to multiple conditions (i.e. repeated measures) of an independent variable [14]. This design has several analytical advantages, including its ability to provide more reliable estimates by accounting for unbalanced data and by controlling for person‐specific confounders by design when analyzing data with hierarchy [15]. Given that such benefits may be translated into effective and valid regulatory and/or prevention strategies, research is warranted to examine recent data of multi‐year, within‐person changes in ENDS use and their associations with use of other substances, including polysubstance use.
Research indicates that mental health problems, curiosity and perceived social norms are the leading reason for initiation of ENDS use in young people [16, 17, 18, 19]. Evidence suggests that having internalizing and externalizing problems is associated with an increased likelihood of initiating ENDS use [16, 17]. Many of the ENDS‐using young people are experimenters or light users whose use does not persist over time [20]. Not surprisingly, past research has shown that the vast majority of ENDS experimenters do not keep using ENDS in subsequent years [20]. This clearly illustrates that commonly used metrics of ENDS use in young people, such as any ENDS use or the number of days of ENDS use within the past 30 days [21], not only overestimate current ENDS use prevalence by classifying experimental use into current use [20, 22, 23] but also fail to adequately discern patterns of ENDS use and its association with other substances. In conventional cigarette tobacco research, current established use (i.e. every‐day/some‐day use among those who have smoked more than 100 cigarettes) or escalated use (e.g. more cigarettes smoked a day than before) has been widely used to capture stable patterns of use [24, 25, 26].
The primary aim of this study was to assess within‐person changes in escalated and established ENDS use and their associations with individual and combined substance use during a 6‐year period in a longitudinal sample of population‐representative US young adults. We hypothesized that while the prevalence of current established ENDS user would increase, within‐person escalated use of ENDS would decrease during the study period and that escalated ENDS use and time‐lagged established ENDS use would predict an increased risk of using individual and combined substance use. It was also hypothesized that escalated ENDS use would be associated with an increased risk of initiating marijuana vaping among those who have never vaped marijuana. As a secondary aim, we examined association between ENDS use characteristics (e.g. flavor) and multiple substance use behaviors among current established ENDS users.
METHODS
Study design and participants
Data were drawn from waves 1 to 5 of the Population Assessment of Tobacco and Health (PATH) study (2013–19). The PATH study comprises ongoing nationally representative longitudinal panel data that are designed to collect information about various substance use behaviors among non‐institutionalized US adolescents (12–17 years) and adults (18 years or older). The PATH data oversampled minority populations (e.g. tobacco users, African Americans and young adults) and used a four‐stage stratified area probability sampling method. In this study, young adults aged 18–24 years at wave 1 (n = 9110) who had matched data across all five waves of assessments were extracted as an analytical sample (n = 5042). The number of individuals who were lost to follow‐up at each wave were as follows: wave 2 = 1785, wave 3 = 847, wave 4 = 738 and wave 5 = 698 (Supporting information, Fig. S1). This study used publicly available de‐identified data (adult public‐use files for the PATH study) and was exempted from the review of the Indiana University’s Institutional Review Board. The analyses in this study were not pre‐registered, and therefore the results should be considered exploratory.
Measures
Outcome variables
The major outcome variables included: (1) current established cigarette smoking (current every‐day or some‐day use of cigarettes among those who had smoked more than 100 cigarettes in their life‐time), (2) past 30‐day binge drinking (four or more drinks for female and five or more drinks for male respondents on the same occasion), (3) past 12‐month marijuana use [hash, tetrahydrocannabinol (THC), grass, pot, weed or cigar/cigarillo/filtered cigar containing marijuana], (4) past 12‐month prescription drug (stimulants such as Adderall and Ritalin, painkillers, sedatives or tranquilizers) misuse and (5) past 12‐month other illicit drugs (cocaine/crack, hallucinogens, heroin, inhalants, solvents, hallucinogens or methamphetamine/speed) use. We used the past 30‐day binge drinking measure, given that the number of alcoholic drinks consumed each day on the days of drinking was collected only for those who drank in the past 30 days. All these variables were dichotomous (yes versus no) and repeatedly measured at each wave. Following the literature [9, 10, 27], the combined use of substances other than ENDS included: (1) cigarettes + alcohol, (2) cigarettes + marijuana, (3) alcohol + marijuana, (4) cigarettes + alcohol + marijuana and (5) cigarettes + alcohol + marijuana + drugs (including prescription and illegal drugs). These variables were repeated measures and binary (e.g. coded as 1 for the cigarette + alcohol if respondents reported use of both substances, otherwise 0). Marijuana vaping initiation (yes/no) was separately measured according to respondents’ ever use of marijuana, marijuana concentrates/waxes and THC/hash oils in an ENDS device.
ENDS use behaviors and characteristics
Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated). We defined current established ENDS use as those who said ‘yes’ to both questions: ‘Have you ever used electronic nicotine products fairly regularly?’ and ‘Do you currently use every day or some days?’ [20]. This measure was a time‐lagged regressor (i.e. regressor on outcomes). In the analyses for the secondary aim, ENDS use characteristics included: flavor used in the past 30 days (tobacco, menthol/mint, fruit/sweet or another flavor), nicotine concentration level usually used (range = 0.1–6.0+ %), price paid for ENDS (> $20, ≤ $20) or do not own an ENDS) and ENDS device type used most often (disposables, device with replaceable pre‐filled cartridge or tank/mod).
Mental health problems
Variables that measured mental health conditions were assessed according to the Global Appraisal of Individual Needs–short screener (GAIN‐SS) [28]. Following established literature [17, 20, 29], we assessed two mental health subscales: internalizing and externalizing symptoms. Internalizing problems were assessed using four items (i.e. depression, distress, anxiety and having difficulty with sleeping), and externalizing symptoms were measured using five items (i.e. had a hard time paying attention, had a hard time listening to instructions, lied/conned to gain something wanted, bullied or threatened others and started physical fights). Each subscale of mental health problems was used as the count of symptoms in the 12 months [17, 20, 29].
Covariates
Time‐fixed covariates included baseline socio‐demographic characteristics such as sex (female versus male), ethnicity (Hispanic versus non‐Hispanic), race (white versus non‐white), enrollment in a degree program (currently enrolled versus not enrolled) and federal poverty level (≥ 200, 100–199 or < 100% of poverty guideline) [20].
Statistical analysis
We calculated unweighted frequencies and weighted proportions for categorical variables and weighted means and standard errors for count variables at each wave. Each variable included in this study was a repeated measure from waves 1 to 5, except baseline socio‐demographic variables. Weighted Pearson’s χ2 tests were conducted to assess the differences in the proportions of categorical variables of interest by escalated ENDS use status.
In our primary analyses, a series of weighted logistic regression models assessed the associations of escalated ENDS use with individual and combined substance use outcome variables. We also examined the prospective associations between time‐lagged established ENDS use and substance use outcomes. The generalized linear mixed model (GLMM) approach was utilized to account for the clustering (i.e. correlations) of repeated measures within individuals [30]. This approach has been considered as one of the most useful analytical methods in analyzing discrete longitudinal data (i.e. non‐Gaussian data) in modern statistics by accommodating heterogeneity between clusters using random effects, which enables researchers to deal with many complications in longitudinal panel data analysis with the linear model framework [31]. Given that marijuana vaping measures are only available in later waves of the PATH data, using the latest wave (i.e. wave 5), a separate cross‐sectional weighted multivariable logistic regression was employed to test if escalated ENDS use is associated with marijuana vaping initiation among never marijuana vapers. Similarly, for the secondary aim, we fitted a series of cross‐sectional weighted multivariable logistic regression models to examine if ENDS use characteristics are associated with multiple substance use behaviors.
Associations are presented as adjusted odds ratios (aORs) with the corresponding 95% confidence interval (CI). The generalized variance inflation factor (GVIF) indicated no multicollinearity issues for all analytical models (all GVIFs < 2). We incorporated sampling weights (i.e. wave 5 adult wave 1 cohort all‐waves weights) for all analyses to account for different probabilities of sample selection and to calculate nationally representative estimates. Throughout the five waves, missing values were observed on outcomes (substance use outcomes: 0.1–5.7%), main explanatory variables (escalated ENDS use: 0.1–0.7%), mental health (0.8–1.5%) and baseline socio‐demographic characteristics (0.0–11.0%) (see Supporting information, Table S1 for details). We imputed missing data using the multiple imputation with chained equations (MICE) technique, which does not demand a certain pattern of missingness [32], with five imputed data sets created via 40 iterations for each set. Sensitivity analyses were conducted using complete data (without missing values) to demonstrate the robustness of primary model estimates. Analyses were performed in R version 4.2.0 using the lme4, survey and mice packages.
RESULTS
Table 1 displays the analytical sample characteristics of the 5042 US young adult respondents (waves 1–5; 2013–19). The sample (50.1% female; 70.8% white; 20.6% Hispanic; 38.6% enrolled in a degree program; 45.9% below the federal poverty level) was socio‐demographically diverse and generally mirrors the demographic composition of young adults in the United States. Depicted in Figure 1, established ENDS use generally increased over time (4.0–9.1%). Escalated ENDS use gradually decreased over time, with the lowest prevalence at wave 4 (4.0%) but sharply increasing at wave 5 (8.4%). In contrast, cigarette smoking gradually increased with a peak at wave 4 (20.4%) and decreasing at wave 5 (18.4%). Over time, the prevalence of marijuana use increased and that of prescription drug misuse and binge drinking decreased. Illegal drug use was relatively stagnant over time. As presented in Table 2, male respondents and those with high internalizing and externalizing mental health problems showed escalated ENDS use over time (Ps < 0.05). For all study outcomes, escalated ENDS use played a significant role at each wave in discriminating between respondents who used and who did not use other substances (Table 3).
TABLE 1.
Descriptive statistics of sample characteristics (n = 5042) over time (waves 1–5, 2013–19), Population Assessment of Tobacco and Health (PATH) study.
| Variables | Wave 1 | Wave 2 | Wave 3 | Wave 4 | Wave 5 |
|---|---|---|---|---|---|
| Escalated ENDS use | |||||
| Not escalated | – | 4679 (94.3) | 4686 (94.7) | 4778 (96.0) | 4519 (91.6) |
| Escalated | – | 340 (5.7) | 322 (5.3) | 249 (4.0) | 517 (8.4) |
| Established ENDS use | |||||
| No | 4781 (96.0) | 4679 (94.4) | 4673 (94.1) | 4696 (94.3) | 4493 (90.9) |
| Yes | 255 (4.0) | 343 (5.6) | 363 (5.9) | 345 (5.7) | 547 (9.1) |
| Established cigarette use | |||||
| No | 3773 (80.7) | 3711 (79.8) | 3709 (79.7) | 3706 (79.6) | 3812 (81.8) |
| Yes | 1265 (19.3) | 1329 (20.2) | 1331 (20.3) | 1334 (20.4) | 1225 (18.2) |
| 30D heavy alcohol use | |||||
| No | 3925 (84.7) | 4181 (85.4) | 4212 (85.7) | 4288 (87.5) | 4292 (87.5) |
| Yes | 828 (15.3) | 807 (14.6) | 780 (14.3) | 707 (12.5) | 692 (12.5) |
| 12M marijuana use | |||||
| No | 3562 (77.6) | 3143 (68.4) | 3164 (68.4) | 3121 (67.6) | 3047 (66.0) |
| Yes | 1337 (22.4) | 1884 (31.6) | 1862 (31.6) | 1905 (32.4) | 1987 (34.0) |
| 12M prescription drug misuse | |||||
| No | 4405 (87.7) | 4456 (89.3) | 4468 (89.6) | 4515 (90.4) | 4602 (92.2) |
| Yes | 619 (12.3) | 580 (10.7) | 567 (10.4) | 518 (9.6) | 427 (7.8) |
| 12M illegal drug use | |||||
| No | 4723 (94.0) | 4700 (93.4) | 4648 (93.5) | 4635 (92.9) | 4631 (93.1) |
| Yes | 303 (6.0) | 333 (6.6) | 372 (6.5) | 386 (7.1) | 397 (6.9) |
| 12M mean internalizing problems | 1.67 (0.03) | 1.61 (0.03) | 1.56 (0.03) | 1.69 (0.03) | 1.64 (0.03) |
| 12M mean externalizing problems | 1.16 (0.02) | 1.08 (0.02) | 1.00 (0.02) | 1.11 (0.03) | 1.02 (0.02) |
| Sex | |||||
| Male | 2361 (49.9) | – | – | – | – |
| Female | 2681 (50.1) | – | – | – | – |
| Race/ethnicity | |||||
| White | 3440 (70.8) | – | – | – | – |
| Non‐white | 1602 (29.2) | – | – | – | – |
| Ethnicity | |||||
| Hispanic | 1237 (20.6) | – | – | – | – |
| Non‐Hispanic | 3805 (79.3) | – | – | – | – |
| Current degree program | |||||
| Not enrolled | 3090 (58.3) | – | – | – | – |
| Enrolled | 1939 (41.7) | – | – | – | – |
| Federal poverty level a | |||||
| ≥ 200% | 1263 (32.5) | – | – | – | – |
| 100–199% | 968 (21.6) | – | – | – | – |
| < 100% | 2255 (45.9) | – | – | – | – |
Values in the table indicate unweighted frequency and those in parentheses weighted proportion for categorical variables. For count variables, values in the table indicate weighted mean and those in parentheses standard errors. Frequencies may not sum to the total due to missing observations. Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated). All the variables shown in the table are repeated measures within respondents from waves 1 to 5 except for baseline socio‐demographic characteristics, such as sex, race, ethnicity, current degree program enrollment and poverty level. ENDS = electronic nicotine delivery systems; 12M = past 12 months; 30D = past 30‐day.
≥ 200% of poverty guideline: at or above twice poverty level; 100–199% of poverty guideline: at or near poverty level; < 100% of poverty guideline: below poverty level. The poverty status is calculated based on annual household income and the US Department of Health and Human Services poverty guidelines.
FIGURE 1.
Prevalence of each substance use over a five‐wave period among US young adults. Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated); y‐axis range: 0–100. ENDS = electronic nicotine delivery systems
TABLE 2.
Weighted proportion of baseline socio‐demographic characteristics and mental health by ENDS use escalation status.
| ENDS use frequency | Sex | Race | Ethnicity | Degree program | Federal poverty level a | Internalizing problems b | Externalizing problems b | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Male | Female | White | Non‐white | Hispanic | Non‐Hispanic | Not enrolled | Enrolled | ≥ 200% | 100–199% | < 100% | Low/moderate | High | Low/moderate | High | |
| Wave 2 | P < 0.001 | P < 0.001 | P = 0.269 | P = 0.008 | P = 0.293 | P < 0.001 | P < 0.001 | ||||||||
| Not escalated | 48.9 | 51.1 | 70.2 | 29.8 | 20.8 | 79.2 | 57.7 | 42.3 | 32.7 | 21.7 | 45.6 | 80.4 | 19.6 | 96.4 | 3.6 |
| Escalated | 64.1 | 35.9 | 82.0 | 18.0 | 18.2 | 81.8 | 66.1 | 33.9 | 29.3 | 19.9 | 50.8 | 66.1 | 33.9 | 91.7 | 8.3 |
| Wave 3 | P < 0.001 | P = 0.807 | P = 0.786 | P < 0.001 | P = 0.508 | P = 0.004 | P = 0.003 | ||||||||
| Not escalated | 49.1 | 50.0 | 70.9 | 29.1 | 20.6 | 79.4 | 57.5 | 42.5 | 32.8 | 21.5 | 45.7 | 80.4 | 19.6 | 97.0 | 3.0 |
| Escalated | 61.0 | 39.0 | 71.8 | 28.2 | 21.2 | 78.8 | 69.8 | 30.2 | 28.9 | 22.5 | 48.6 | 72.2 | 27.8 | 93.4 | 6.6 |
| Wave 4 | P = 0.019 | P = 0.112 | P = 0.504 | P = 0.170 | P = 0.218 | P < 0.001 | P = 0.010 | ||||||||
| Not escalated | 49.4 | 50.6 | 70.7 | 29.3 | 20.5 | 79.5 | 58.0 | 42.0 | 32.8 | 21.4 | 45.8 | 76.2 | 23.8 | 95.8 | 4.2 |
| Escalated | 59.6 | 40.4 | 75.6 | 24.4 | 22.2 | 77.8 | 63.3 | 36.7 | 27.7 | 26.2 | 46.1 | 61.6 | 38.4 | 91.1 | 8.9 |
| Wave 5 | P = 0.015 | P < 0.001 | P = 0.001 | P = 0.251 | P = 0.212 | P < 0.001 | P < 0.001 | ||||||||
| Not escalated | 49.3 | 50.7 | 70.0 | 30.0 | 21.1 | 78.9 | 58.1 | 41.9 | 32.8 | 21.7 | 45.5 | 78.1 | 21.9 | 96.5 | 3.5 |
| Escalated | 56.3 | 43.7 | 79.3 | 20.7 | 14.7 | 85.3 | 60.6 | 39.4 | 29.6 | 20.7 | 49.7 | 68.2 | 31.8 | 93.1 | 6.9 |
Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated). Mean (standard error) of the number of ENDS‐use days within the past 30 days among escalated ENDS users was 14.0 (0.7), 13.0 (0.7), 14.3 (0.9) and 16.5 (0.6) from waves 2 to 5, respectively. Weighted Pearson’s χ2 test was used to test differences between groups. ENDS = electronic nicotine delivery systems; P = P‐value.
≥ 200% of poverty guideline = at or above twice poverty level; 100–199% of poverty guideline = at or near poverty level; < 100% of poverty guideline: below poverty level. The poverty status is calculated based on annual household income and the US Department of Health and Human Services poverty guidelines.
Dichotomized into none/low/moderate (fewer than symptoms) or high (for or more symptoms) in the past 12 months.
TABLE 3.
Weighted proportion of substance use by ENDS use escalation status.
| ENDS use frequency | Cigarette | Binge drinking | Marijuana | Prescription drug | Illegal drug | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| No | Yes | No | Yes | No | Yes | No | Yes | No | Yes | |
| Wave 2 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | |||||
| Not escalated | 81.9 | 18.1 | 86.1 | 13.9 | 69.8 | 30.2 | 90.0 | 10.0 | 95.0 | 5.0 |
| Escalated | 43.3 | 56.7 | 71.9 | 28.1 | 43.4 | 56.6 | 77.1 | 22.9 | 84.0 | 16.0 |
| Wave 3 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | |||||
| Not escalated | 81.2 | 18.8 | 86.3 | 13.7 | 70.0 | 30.0 | 90.1 | 9.9 | 94.1 | 5.9 |
| Escalated | 51.8 | 48.2 | 73.4 | 26.5 | 40.9 | 59.1 | 81.2 | 18.8 | 82.8 | 17.2 |
| Wave 4 | P < 0.001 | P < 0.001 | P < 0.001 | P = 0.023 | P < 0.001 | |||||
| Not escalated | 80.6 | 19.4 | 88.1 | 11.9 | 68.8 | 31.2 | 90.8 | 9.2 | 93.3 | 6.7 |
| Escalated | 54.9 | 45.1 | 72.9 | 27.1 | 40.9 | 59.1 | 84.8 | 15.2 | 86.1 | 13.9 |
| Wave 5 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | P < 0.001 | |||||
| Not escalated | 84.1 | 15.9 | 88.0 | 12.0 | 68.8 | 31.2 | 931 | 6.9 | 94.2 | 5.8 |
| Escalated | 58.2 | 41.8 | 81.9 | 18.1 | 35.4 | 64.6 | 82.0 | 18.0 | 81.5 | 18.5 |
Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated). Weighted Pearson’s χ2 test was used to test differences between groups. ENDS = electronic nicotine delivery systems; P = P‐value.
Table 4 displays the weighted associations of escalated ENDS use with polysubstance use. Escalated ENDS use predicted cigarette smoking + binge drinking (aOR = 1.38, 95% CI = 1.02–1.85), cigarette smoking + marijuana use (aOR = 2.36, 95% CI = 1.81–3.07) and binge drinking + marijuana use (aOR = 1.68, 95% CI = 1.33–2.13), controlling for covariates. For each substance use outcome, escalated ENDS use was associated with increased odds of binge drinking (aOR = 1.45, 95% CI = 1.19–1.77), marijuana use (aOR = 2.25, 95% CI = 1.82–2.78), prescription drug misuse (aOR = 1.26, 95% CI = 1.00–1.59) and illicit drug use (aOR = 1.49, 95% CI = 1.13–1.98) (Supporting information, Table S2) and of initiating marijuana vaping (aOR = 1.52, 95% CI = 1.11–2.08) (Supporting information, Table S3). In the sensitivity analyses, estimates of escalated ENDS use in predicting multiple substance use behaviors showed similar results between imputed and complete data without imputation (Supporting information, Table S4). Additionally, time‐lagged established ENDS use significantly predicted subsequent cigarette smoking and marijuana use and all examined combined use of substances except the last category (Supporting information, Tables S5 and S6). Further, sweet/fruit flavor use (versus menthol/mint) was associated with increased likelihood of reporting dual use of cigarettes and marijuana (Supporting information, Table S7).
TABLE 4.
Weighted generalized linear mixed model of multiple substance use behaviors among US young adults.
| Variable | CIG + ALC | CIG + MJ | ALC + MJ | CIG + ALC + MJ | CIG + ALC + MJ + DR | |||||
|---|---|---|---|---|---|---|---|---|---|---|
| aOR (95% CI) | P | aOR (95% CI) | P | aOR (95% CI) | P | aOR (95% CI) | P | aOR (95% CI) | P | |
| ENDS use frequency | ||||||||||
| Not escalated | Ref | Ref | Ref | Ref | Ref | |||||
| Escalated | 1.38 (1.02, 1.85) | 0.035 | 2.36 (1.81, 3.07) | < 0.001 | 1.68 (1.33, 2.13) | < 0.001 | 1.41 (0.98, 2.03) | 0.065 | 0.93 (0.52, 1.69) | 0.819 |
| Time | ||||||||||
| Wave 2 | Ref | Ref | Ref | Ref | Ref | |||||
| Wave 3 | 0.89 (0.71, 1.13) | 0.345 | 1.10 (0.89, 1.36) | 0.374 | 0.86 (0.72, 1.02) | 0.091 | 0.81 (0.61, 1.08) | 0.157 | 0.61 (0.39, 0.96) | 0.034 |
| Wave 4 | 0.61 (0.48, 0.79) | < 0.001 | 1.13 (0.92, 1.40) | 0.238 | 0.63 (0.52, 0.77) | < 0.001 | 0.50 (0.37, 0.69) | < 0.001 | 0.29 (0.18, 0.49) | < 0.001 |
| Wave 5 | 0.56 (0.44, 0.72) | < 0.001 | 0.61 (0.49, 0.77) | < 0.001 | 0.70 (0.58, 0.85) | < 0.001 | 0.45 (0.32, 0.62) | < 0.001 | 0.25 (0.15, 0.43) | < 0.001 |
| Sex | ||||||||||
| Male | Ref | Ref | Ref | Ref | Ref | |||||
| Female | 0.58 (0.44, 0.75) | < 0.001 | 0.52 (0.38, 0.70) | < 0.001 | 0.70 (0.58, 0.84) | < 0.001 | 0.50 (0.36, 0.68) | < 0.001 | 0.44 (0.26, 0.75) | 0.003 |
| Race | ||||||||||
| White | Ref | Ref | Ref | Ref | Ref | |||||
| Non‐white | 0.49 (0.36, 0.67) | < 0.001 | 0.66 (0.48, 0.91) | 0.012 | 0.67 (0.54, 0.83) | < 0.001 | 0.54 (0.37, 0.80) | 0.002 | 0.26 (0.12, 0.57) | < 0.001 |
| Ethnicity | ||||||||||
| Hispanic | Ref | Ref | Ref | Ref | Ref | |||||
| Non‐Hispanic | 1.19 (0.87, 1.62) | 0.228 | 2.80 (1.96, 4.02) | < 0.001 | 0.61 (0.49, 0.76) | < 0.001 | 0.85 (0.59, 1.23) | 0.393 | 1.64 (0.82, 3.29) | 0.163 |
| Current degree program | ||||||||||
| Not enrolled | Ref | Ref | Ref | Ref | Ref | |||||
| Enrolled | 0.48 (0.36, 0.63) | < 0.001 | 0.24 (0.18, 0.33) | < 0.001 | 0.20 (0.99, 1.46) | 0.059 | 0.44 (0.31, 0.63) | < 0.001 | 0.51 (0.29, 0.90) | 0.021 |
| Federal poverty level | ||||||||||
| ≥ 200% | Ref | Ref | Ref | Ref | Ref | |||||
| 100–199% | 1.89 (1.32, 2.72) | < 0.001 | 3.80 (2.49, 5.81) | < 0.001 | 1.01 (0.78, 1.30) | 0.950 | 1.69 (1.10, 2.60) | 0.018 | 1.32 (0.66, 2.66) | 0.438 |
| <100% | 1.60 (1.17, 2.20) | 0.004 | 3.28 (2.29, 4.70) | < 0.001 | 0.84 (0.67, 1.04) | 0.114 | 1.20 (0.82, 1.77) | 0.348 | 1.12 (0.60, 2.09) | 0.727 |
| Established cigarette use | ||||||||||
| No | Ref | Ref | Ref | Ref | Ref | |||||
| Yes | – | – | – | – | 2.48 (2.05, 3.00) | < 0.001 | – | – | – | – |
| 30D heavy alcohol use | ||||||||||
| No | Ref | Ref | Ref | Ref | Ref | |||||
| Yes | – | – | 2.04 (1.62, 2.58) | < 0.001 | – | – | – | – | – | – |
| 12M marijuana use | ||||||||||
| No | Ref | Ref | Ref | Ref | Ref | |||||
| Yes | 2.47 (1.96, 3.11) | < 0.001 | – | – | – | – | – | – | – | – |
| 12M prescription drug misuse | ||||||||||
| No | Ref | Ref | Ref | Ref | Ref | |||||
| Yes | 1.51 (1.14, 2.01) | 0.004 | 2.36 (1.81, 3.07) | < 0.001 | 1.90 (1.54, 2.34) | < 0.001 | 2.79 (2.03, 3.83) | < 0.001 | – | – |
| 12M illegal drug use | ||||||||||
| No | Ref | Ref | Ref | Ref | Ref | |||||
| Yes | 2.51 (1.84, 3.43) | < 0.001 | 4.99 (3.64, 6.83) | < 0.001 | 3.33 (2.65, 4.20) | < 0.001 | 3.89 (2.75, 5.52) | < 0.001 | – | ‐ |
| 12M internalizing problems | 1.22 (1.13, 1.32) | < 0.001 | 1.39 (1.29, 1.49) | < 0.001 | 1.12 (1.06, 1.19) | < 0.001 | 1.20 (1.09, 1.32) | < 0.001 | 1.37 (1.17, 1.61) | < 0.001 |
| 12M externalizing problems | 1.07 (0.98, 1.16) | 0.155 | 1.01 (0.93, 1.10) | < 0.001 | 1.19 (1.11, 1.27) | < 0.001 | 1.18 (1.06, 1.31) | 0.003 | 1.53 (1.30, 1.82) | < 0.001 |
Variables in the table are repeated measures within respondents from waves 2 to 5. Socio‐demographic characteristics such as sex, race, ethnicity, current degree program enrollment and poverty level are measured at baseline. Escalated ENDS use was computed by subtracting the number of days of ENDS use within the past 30 days at wave from that at wave and coded as 1 = escalated if the value was greater than zero (otherwise, coded as 0 = not escalated). ALC = binge drinking; aOR = adjusted odds ratio; CI = confidence interval; CIG = established cigarette smoking; DR = prescription or illegal drugs; ENDS = electronic nicotine delivery systems; MJ = marijuana; P = P‐value; Ref = reference group; 12M = past 12 months; 30D = past 30 days.
DISCUSSION
The current study is one of the first longitudinal analyses to characterize within‐person associations of escalated ENDS use with individual and combined use of other substances in a recent national sample of US young adults. Our hypothesis was that, while the prevalence of current established ENDS users would increase, the within‐person escalated use of ENDS would decrease during the 6‐year study period, was partly supported. Established ENDS use generally increased while escalated ENDS use decreased during the first four waves, but both abruptly increased between last two waves (December 2016–December 2019), while cigarette smoking decreased during the same period. This may imply that ENDS use may be diverting young adults away from combustible cigarettes after the emergence of new types of ENDS (e.g. JUUL, Puff Bar). This possible acceleration should be confirmed with future waves of data.
ENDS use escalation appears to be increasing after several years of a downward trend (from waves 1 to 4 of the PATH study). Whereas many young adults between 2013 and 2018 did not increase their ENDS use frequency over time or possibly experimented with ENDS, they may have increased their ENDS use intensity/frequency in recent years. This implies potentially higher abuse potential of recent ENDS products (e.g. via nicotine salt products with high nicotine) or some unknown mechanism by which young adults are increasing their ENDS use frequency. For example, ENDS sales in the United States remained relatively steady in 2015 and 2016, but it grew substantially from 2017 to 2018 as JUUL (pod‐based nicotine salt ENDS) sales increased [33], and the number of US young adults who tried JUUL increased more than 400% from July 2017 to October 2018 [34]. It is also plausible that new types of disposable ENDS (e.g. Puff Bar) with a variety of flavors appealing to young people (e.g. blueberry ice) might have contributed to the abrupt increase in escalated and established ENDS use between waves 4 and 5 in this study. During August 2019–May 2020, retail sales for disposable ENDS increased from 10.3 to 19.8% [35].
The main finding of the current study is the confirmation of within‐person associations between escalated ENDS use and other substances during a 6‐year time‐frame. Escalated ENDS use was associated with a greater likelihood of using each (except cigarette smoking) and multiple substances between cigarettes, alcohol (binge) and marijuana, supporting our a priori hypothesis. This finding not only supports previous findings [4, 6, 8, 9, 10], but also meaningfully extends the literature by demonstrating the associations between escalated ENDS use and other‐ and polysubstance use. These results can be explained by the common liability model suggesting that a common liability (e.g. biobehavioral features) may increase the risk of using both ENDS and other substances [36, 37]. Time‐lagged established ENDS use was prospectively associated with an increased risk of using cigarettes and marijuana and combined use between cigarettes, alcohol and marijuana. This finding may support the catalyst effect that suggests that use of a substance could lead to other substance use [38]. In addition, escalated ENDS use was associated with marijuana vaping initiation, which is increasingly prevalent among young adults recently [1] and can potentially increase e‐cigarette or vaping product use‐associated lung injury (EVALI) [39]. This is alarming, given the surging popularity of ENDS in young people. This surge in ENDS use and subsequent or concurrent co‐use of other substances may emerge as a grave public health threat due to the ongoing renormalization of nicotine use via ENDS use. As supported by the association between escalated ENDS and marijuana use in the present study, this may be exacerbated by the burgeoning renormalization of marijuana use expedited by the increasing trend of legalization of recreational marijuana, further contributing to drug misuse/abuse and addiction problems. Indeed, recent US regional evidence has shown that more than 95% of college students who currently use ENDS were characterized by multiple or often problematic substance use behaviors [10], which are associated with poor physical and mental health [40, 41, 42]. The present study adds another layer of evidence why escalating ENDS use should be addressed, particularly among young people.
Another finding of note was that sweet/fruit flavor (versus menthol/mint) was associated with an increased likelihood of co‐using cigarettes and marijuana. ENDS flavoring has been a central target for tobacco regulatory policies [43]. Among various ENDS flavors, sweet flavors (e.g. fruit and candy) have enhancing effects on sensory perception/experience and product appeal [44] and were the leading ENDS flavor types in young people [45]. What is more concerning is the exposure to these flavors may increase nicotine dependence and promote dose escalation [46], which is supported by the finding of the current study. Regulation of sweet/fruit ENDS flavors may discourage young people from escalating ENDS use and therefore using each and multiple substances as well as ENDS.
Both internalizing and externalizing problems played a significant role in discriminating escalated ENDS use over time and in predicting multiple substance use behaviors. These results are congruent with a recent finding that mental health problems are associated with higher rates of ENDS use escalation in a cohort sample of adolescents [16]. Our findings meaningfully expand the literature by demonstrating the associations between mental health, escalated ENDS use and multiple substance use behaviors. Also importantly, throughout the analyses in this study, being male was associated consistently with a higher risk of ENDS use escalation and multiple substance use. This might be due, in part, to more positive expectancies about ENDS products (e.g. taste, social facilitation and energy) [47] and preference of fruited‐flavored ENDS [48] among males. Future research is needed to examine the degree to which the effects of mental health and sex on multiple substance use is explained through escalated ENDS use using a mediation approach.
It is also interesting to note the difference in the association of cigarette smoking with escalated and established ENDS use. Time‐lagged established ENDS use, but not escalated ENDS use, was prospectively associated with greater likelihood of smoking combustible cigarettes. A speculation about this finding is that those who escalated ENDS use may be still experimenters or may not be motivated to smoke cigarettes to consume nicotine. In contrast, established ENDS use in previous years may be addicted to nicotine and exploring other nicotine substances such as cigarettes, which supports prior work on the association between ENDS use and subsequent use of combustible tobacco in young populations [12, 25, 49].
This study has limitations. First, the use of repeated measures of self‐reported data taken from the same respondents may have reduced measurement error; however, our findings may be subject to respondent bias such as measurement errors attributable to reporting bias and recall bias. Secondly, we cannot preclude the possibility that unmeasured variables may have influenced the longitudinal associations of escalated ENDS use with other substance use. Thirdly, although we analyzed longitudinal cohort data and included time‐lagged exposure, causal interpretation is not warranted due to a lack of randomized experimental design. Lastly, the use of imputed data might have produced biased estimates, although sensitivity analyses with complete case data did not change the main findings.
CONCLUSIONS
The present study finds that while the prevalence of ENDS‐using young adults steadily increased between 2013 and 2019, the rate of increase may have started accelerating in recent years. Importantly, escalation in ENDS use appears to have risen after several years of a downward trend in a nationally representative cohort of young adults, possibly due, in part, to the higher abuse potential of recent ENDS products with more pleasurable sensory attributes with high nicotine concentrations. Escalated ENDS use and time‐lagged established ENDS use were prospectively associated with individual and combined substance use, particularly between cigarettes, alcohol and marijuana. Among established ENDS users, sweet/fruit flavor was associated with increased risk of co‐using cigarettes and marijuana use. The present study adds another layer of evidence of a need for preventing young people from using ENDS products to discourage them from co‐using other substances. This study also suggests that regulation of sweet/fruit ENDS flavors may help thwart young people from escalating ENDS use and therefore using each and multiple substances.
DECLARATION OF INTERESTS
None of the authors have any conflicts of interest to disclose, including relevant financial interests, activities, relationships and affiliations.
AUTHOR CONTRIBUTIONS
Dae Hee Han: Conceptualization; formal analysis; investigation; methodology; writing‐original draft. Kit Elam: Validation; writing‐review and editing. Patrick Quinn: Validation; writing‐review and editing. Chunfeng Huang: Validation; writing‐review and editing. Dong‐Chul Seo: Conceptualization; investigation; methodology; writing‐review and editing; supervision; interpretation.
Supporting information
Table S1. Frequency and proportion of missing data
Table S2. Association of escalated ENDS use with each substance use behavior among U.S. young adults
Table S3. Association of escalated ENDS use with marijuana vaping initiation at Wave 5 among never marijuana vapers in previous waves (N = 4195)
Table S4. Association of escalated ENDS use with multiple substance use behaviors among U.S. young adults using complete data (N = 4032)
Table S5. Association of established ENDS use with multiple substance use behaviors among U.S. young adults
Table S6. Association of established ENDS use with each substance use behaviors among U.S. young adults
Table S7. Association of ENDS use characteristics with multiple substance use behaviors among established ENDS‐using young adults at Wave 5 (N = 1620)
Figure S1. Flowchart for the final sample selection
ACKNOWLEDGEMENTS
None.
Han D‐H, Elam KK, Quinn PD, Huang C, Seo D‐C. Within‐person associations of escalated electronic nicotine delivery systems use with cigarette, alcohol, marijuana and drug use behaviors among US young adults. Addiction. 2023;118(3):509–519. 10.1111/add.16082
Funding information This research did not receive any specific grant from funding agencies in the public, commercial or not‐for‐profit sectors.
REFERENCES
- 1. National Institute on Drug Abuse . Vaping and cannabis trends among young adults (19–22). Available at: https://www.drugabuse.gov/drug-topics/trends-statistics/infographics/vaping-cannabis-trends-among-young-adults-19-22 (accessed 2 April 2022).
- 2. Arnett JJ. Emerging adulthood: a theory of development from the late teens through the twenties. Am Psychol. 2000;55:469–680. [PubMed] [Google Scholar]
- 3. Doran N, Myers MG, Correa J, Strong DR, Tully L, Pulvers K. Marijuana use among young adult non‐daily cigarette smokers over time. Addict Behav. 2019;95:91–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Buu A, Hu YH, Wong SW, Lin HC. Comparing American college and noncollege young adults on e‐cigarette use patterns including polysubstance use and reasons for using e‐cigarettes. J Am Coll Health. 2020;68:610–6. [DOI] [PubMed] [Google Scholar]
- 5. Chadi N, Schroeder R, Jensen JW, Levy S. Association between electronic cigarette use and marijuana use among adolescents and young adults: a systematic review and meta‐analysis. JAMA Pediatr. 2019;173:e192574. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Lanza HI, Barrington‐Trimis JL, McConnell R, Cho J, Braymiller JL, Krueger EA, et al. Trajectories of nicotine and cannabis vaping and polyuse from adolescence to young adulthood. JAMA Netw Open. 2020;3:e2019181. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Han DH, Seo DC. Identifying risk profiles for marijuana vaping among US young adults by recreational marijuana legalization status: a machine learning approach. Drug Alcohol Depend. 2022;232:109330. [DOI] [PubMed] [Google Scholar]
- 8. Lanza HI, Motlagh G, Orozco M. E‐cigarette use among young adults: a latent class analysis examining co‐use and correlates of nicotine vaping. Addict Behav. 2020;110:106528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Temple JR, Shorey RC, Lu Y, Torres E, Stuart GL, Le VD. E‐cigarette use of young adults motivations and associations with combustible cigarette, alcohol, marijuana, and other illicit drugs. Am J Addict. 2017;26:343–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Smith PH, Ward RM, Bartoszek LA, Branscum PW. College students’ patterns of electronic nicotine delivery system use and other substance use. J Am Coll Health. 2020;70:1–7. [DOI] [PubMed] [Google Scholar]
- 11. Dai H, Catley D, Richter KP, Goggin K, Ellerbeck EF. Electronic cigarettes and future marijuana use: a longitudinal study. Pediatrics. 2018;141:e20173787. [DOI] [PubMed] [Google Scholar]
- 12. Evans‐Polce RJ, Patrick ME, McCabe SE, Miech RA. Prospective associations of e‐cigarette use with cigarette, alcohol, marijuana, and nonmedical prescription drug use among US adolescents. Drug Alcohol Depend. 2020;216:108303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Evans‐Polce RJ, Veliz PT, Boyd CJ, McCabe SE. E‐cigarette and cigarette use among US adolescents: longitudinal associations with marijuana use and perceptions. Am J Prev Med. 2020;58:854–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Charness G, Gneezy U, Kuhn MA. Experimental methods: between‐subject and within‐subject design. J Econ Behav Organ. 2012;81:1–8. [Google Scholar]
- 15. Hox JJ. Multilevel Analysis: Techniques and Applications 2nd ed. New York, NY: Routledge/Taylor & Francis Group; 2010. [Google Scholar]
- 16. Audrain‐McGovern J, Rodriguez D, Testa S, Alexander E, Pianin S. Adolescent e‐cigarette onset and escalation: associations with internalizing and externalizing symptoms. J Adolesc Health. 2021;68:801–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Riehm KE, Young AS, Feder KA, Krawczyk N, Tormohlen KN, Pacek LR, et al. Mental health problems and initiation of e‐cigarette and combustible cigarette use. Pediatrics. 2019;144:e20182935. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Cho H, Li W, Shen L, Cannon J. Mechanisms of social media effects on attitudes toward e‐cigarette use: motivations, mediators, and moderators in a national survey of adolescents. J Med Internet Res. 2019;21:e14303. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Kinouani S, Leflot C, Vanderkam P, Auriacombe M, Langlois E, Tzourio C. Motivations for using electronic cigarettes in young adults: a systematic review. Subst Abuse. 2020;41:315–22. [DOI] [PubMed] [Google Scholar]
- 20. Han DH, Seo DC. Financial strain and electronic nicotine delivery systems use among US young adults: a longitudinal panel analysis, 2013–2018. Addict Behav. 2021;114:106716. [DOI] [PubMed] [Google Scholar]
- 21. Weaver SR, Kim H, Glasser AM, Sutfin EL, Barrington‐Trimis J, Payne TJ, et al. Establishing consensus on survey measures for electronic nicotine and non‐nicotine delivery system use: current challenges and considerations for researchers. Addict Behav. 2018;79:203–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Amato MS, Boyle RG, Levy D. How to define e‐cigarette prevalence? Finding clues in the use frequency distribution. Tob Control. 2016;25:e24–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23. Farsalinos KE, Niaura R. E‐cigarettes and smoking cessation in the United States according to frequency of e‐cigarette use and quitting duration: analysis of the 2016 and 2017 national health interview surveys. Nicotine Tob Res. 2020;22:655–62. [DOI] [PubMed] [Google Scholar]
- 24. Cavazos‐Rehg P, Li X, Kasson E, Kaiser N, Borodovsky JT, Grucza R, et al. Exploring how social media exposure and interactions are associated with ENDS and tobacco use in adolescents from the PATH study. Nicotine Tob Res. 2021;23:487–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Hair E, Bennett M, Williams V, Johnson A, Rath J, Cantrell J, et al. Progression to established patterns of cigarette smoking among young adults. Drug Alcohol Depend. 2017;177:77–83. [DOI] [PubMed] [Google Scholar]
- 26. Huggett SB, Keyes M, Iacono WG, McGue M, Corley RP, Hewitt JK, et al. Age of initiation and transition times to tobacco dependence: early onset and rapid escalated use increase risk for dependence severity. Drug Alcohol Depend. 2019;202:104–10. [DOI] [PubMed] [Google Scholar]
- 27. Stevens AK, Gunn RL, Sokolovsky AW, Colby SM, Jackson KM. Examining the heterogeneity of polysubstance use patterns in young adulthood by age and college attendance. Exp Clin Psychopharmacol. 2022;30:701–13. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Dennis ML, Chan YF, Funk RR. Development and validation of the GAIN short screener (GSS) for internalizing, externalizing and substance use disorders and crime/violence problems among adolescents and adults. Am J Addict. 2006;15:s80–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Blanco C, Krueger RF, Hasin DS, Liu SM, Wang S, Kerridge BT, et al. Mapping common psychiatric disorders: structure and predictive validity in the national epidemiologic survey on alcohol and related conditions. JAMA Psychiatry. 2013;70:199–207. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. McCulloch CE, Neuhaus JM. Encyclopedia of Biostatistics. New York, NY: Wiley; 2005. [Google Scholar]
- 31. Zhao Y, Staudenmayer J, Coull BA, Wand MP. General design Bayesian generalized linear mixed models. Stat Sci. 2006;21:35–51. [Google Scholar]
- 32. White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30:377–99. [DOI] [PubMed] [Google Scholar]
- 33. Liber A, Cahn Z, Larsen A, Drope J. Flavored e‐cigarette sales in the United States under self‐regulation from January 2015 through October 2019. Am J Public Health. 2020;110:785–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Truth Initiative . Behind the explosive growth of JUUL. Available at: https://truthinitiative.org/research-resources/emerging-tobacco-products/behind-explosive-growth-juul (accessed 27 September 2022).
- 35. Ali ERM, Diaz MC, Vallone D, Tynan MA, Dordova J, Seaman EL. E‐cigarette unit sales, by product and flavor type—United States, 2014–2020. Morb Mortal Wkly Rep. 2020;69:1313–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Van Leeuwen AP, Verhulst FC, Reijneveld SA, Vollebergh WA, Ormel J, Huizink AC. Can the gateway hypothesis, the common liability model and/or, the route of administration model predict initiation of cannabis use during adolescence? A survival analysis—the TRAILS study. J Adolesc Health. 2011;48:73–8. [DOI] [PubMed] [Google Scholar]
- 37. Vanyukov MM, Tarter RE, Kirillova GP, Kirisci L, Reynolds MD, Kreek MJ, et al. Common liability to addiction and ‘gateway hypothesis’: theoretical, empirical and evolutionary perspective. Drug Alcohol Depend. 2012;123:S3–S17. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Schneider S, Diehl K. Vaping as a catalyst for smoking? An initial model on the initiation of electronic cigarette use and the transition to tobacco smoking among adolescents. Nicotine Tob Res. 2016;18:647–53. [DOI] [PubMed] [Google Scholar]
- 39. Centers for Disease Control and Prevention . Outbreak of lung injury associated with the use of e‐cigarette, or vaping, products. Available at: https://www.cdc.gov/tobacco/basic_information/e-cigarettes/severe-lung-disease.html (accessed 27 September 2022).
- 40. Bailey AJ, Farmer EJ, Finn PR. Patterns of polysubstance use and simultaneous co‐use in high risk young adults. Drug Alcohol Depend. 2019;205:107656. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41. Morley KI, Lynskey MT, Moran P, Borschmann R, Winstock AR. Polysubstance use, mental health and high‐risk behaviours: results from the 2012 Global Drug Survey. Drug Alcohol Rev. 2015;34:427–37. [DOI] [PubMed] [Google Scholar]
- 42. Rodzlan Hasani WS, Saminathan TA, Ab Majid NL, Miaw Yn JL, Mat Rifin H, Abd Hamid HA, et al. Polysubstance use among adolescents in Malaysia: findings from the National Health and Morbidity Survey 2017. PLOS ONE. 2021;16:e0245593. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 43. Leventhal A, Cho J, Barrington‐Trimis J, Pang R, Schiff S, Kirkpatrick M. Sensory attributes of e‐cigarette flavours and nicotine as mediators of interproduct differences in appeal among young adults. Tob Control. 2020;29:679–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44. Goldenson NI, Kirkpatrick MG, Barrington‐Trimis JL, Pang RD, McBeth JF, Pentz MA. Effects of sweet flavorings and nicotine on the appeal and sensory properties of e‐cigarettes among young adult vapers: application of a novel methodology. Drug Alcohol Depend. 2016;168:176–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 45. Soneji SS, Knutzen KE, Villanti AC. Use of flavored e‐cigarettes among adolescents, young adults, and older adults: findings from the population assessment for tobacco and health study. Public Health Rep. 2019;134:282–92. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 46. Patten T, Dreier A, Herman RJ, Kimball BA, De Biasi M. Exposure to fruit‐flavoring during adolescence increases nicotine consumption and promotes dose escalation. Neuropharmacology. 2021;195:108672. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47. Piñeiro B, Correa JB, Simmons VN, Harrell PT, Menzie NS, Unrod M, et al. Gender differences in use and expectancies of e‐cigarettes: online survey results. Addict Behav. 2016;52:91–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48. Pang RD, Goldenson NI, Kirkpatrick M, Barrington‐Trimis JL, Cho J, Leventhal AM. Sex differences in the appeal of flavored e‐cigarettes among young adult e‐cigarette users. Psychol Addict Behav. 2020;34:303–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49. Leventhal AM, Strong DR, Kirkpatrick MG, Unger JB, Sussman S, Riggs NR, et al. Association of electronic cigarette use with initiation of combustible tobacco product smoking in early adolescence. JAMA. 2015;314:700–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Table S1. Frequency and proportion of missing data
Table S2. Association of escalated ENDS use with each substance use behavior among U.S. young adults
Table S3. Association of escalated ENDS use with marijuana vaping initiation at Wave 5 among never marijuana vapers in previous waves (N = 4195)
Table S4. Association of escalated ENDS use with multiple substance use behaviors among U.S. young adults using complete data (N = 4032)
Table S5. Association of established ENDS use with multiple substance use behaviors among U.S. young adults
Table S6. Association of established ENDS use with each substance use behaviors among U.S. young adults
Table S7. Association of ENDS use characteristics with multiple substance use behaviors among established ENDS‐using young adults at Wave 5 (N = 1620)
Figure S1. Flowchart for the final sample selection