Abstract
Background:
Despite decreases in adolescents’ cigarette use over the past decade, overall rates of adolescent tobacco use have increased. Research examining adolescents’ changes across a range of tobacco products reflective of the current market, as well as multilevel predictors of use trajectories is needed.
Methods:
Data derive from Waves 1—4 (W1-4; 2013—2018) of the Population Assessment of Tobacco and Health (PATH) study. Participants included 975 adolescents who used ≥1 tobacco product (cigarettes, electronic cigarettes [ECIGs], traditional cigars, cigarillos, filtered cigars, snus, smokeless tobacco [SLT], hookah) at any wave (W1 Mage = 13.29 [0.86], 54.2% male; 54.5% White, 25.9% Hispanic).
Results:
Utilizing latent growth curve modeling (separate models per product), adolescents displayed increases in their past 30-day use of all tobacco products from W1-4. Greater W1 use was predicted by identifying as non-Hispanic (cigarettes); lower parent education (SLT); greater externalizing problems (cigarillos); greater motives (all products except cigarillos); greater youth-reported household smoking rules (cigarillos); and greater isolation (ECIGs). More use across time (i.e., higher slope) was predicted by older age (cigarettes); identifying as male (ECIGs, SLT), Black (vs. White; cigarillos), White (vs. Black, Hispanic; ECIGs, SLT); fewer externalizing problems (SLT); fewer motives (ECIGs); fewer youth-reported rules (cigarillos, SLT); and greater geographic isolation (cigarettes, SLT).
Discussion:
Although some individual-level factors (i.e., motives, externalizing problems) predicted greater W1 use (i.e., intercept) only, interpersonal- (parent rules) and community-level (geographic isolation) factors were associated with changes in use over time (i.e., slope). Intervention efforts may address such factors to reduce adolescents’ escalations in use.
Keywords: Tobacco use, Adolescence, Socioecological framework, Latent growth modeling
Rates of adolescent use of most tobacco products have declined over the past decade, as evidenced by decreases in past 30-day cigarette (15.8% in 2011 to 8.1% in 2018), smokeless tobacco (SLT) (7.9% in 2011 to 5.9% in 2019), and cigar (11.6% in 2011 to 7.6% in 2018) use [1]. However, adolescents’ overall use of any tobacco product has increased from 23.2% in 2011 to 27.0% in 2018, largely due to the emergence of electronic cigarettes (ECIGs) (1.5% in 2011 to 20.8% in 2018) [1]. Less research has investigated changes in adolescents’ frequency of use of these products, as the majority of studies have aggregated across products or focused on cigarettes or ECIGs only. Much existing work has examined trajectories of adolescent cigarette or ECIG use utilizing person-centered analyses, yielding classes of low-level users and users who increase use throughout adolescence (cigarettes) [2], and classes of nonusers, infrequent users, moderate users, and adolescent-onset frequent users (e-cigarettes) [3]. Regarding change examined with latent growth modeling (LGM), findings suggest increases in aggregated (i.e., cigarettes, ECIGs, cigars, SLT, hookah) tobacco use from middle to late adolescence [4], but stability in ECIG use across late adolescence [5].
To our knowledge, only one study has examined changes in the use of a range of tobacco products (i.e., cigarettes, ECIGs, cigars, SLT, hookah), examined separately using LGM [6]. Among young adult college students in Texas, findings revealed decreases in past-month use of ECIGs, cigars, and hookah, with no changes in cigarette or SLT use, from 2014 to 2017. It is unclear whether these same patterns exist during adolescence. Because increased frequency of tobacco use is associated with greater likelihood of nicotine dependence, difficulties with cessation, and continued use [7], research documenting changes in a range of tobacco products is critical to inform policy, prevention, and intervention efforts.
According to the socioecological framework, tobacco use is influenced by factors at the individual, interpersonal, and community levels [8]. At the individual level, adolescents’ internalizing (e.g., depression, anxiety) and externalizing (e.g., aggression, delinquency) problems [9] are associated with greater cigarette and ECIG use [10]. Similarly, greater sensation seeking (i.e., the need for novel and complex sensations and experiences) [11] and positive tobacco-related motives (e.g., boredom relief, affect regulation, self-enhancement) [12] are associated with the use of a range of products (i.e., cigarettes, ECIGs, traditional cigars, filtered cigars, cigarillos, SLT, hookah) among adolescents. At the interpersonal level, parents’ own tobacco use is consistently associated with increased risk for adolescents’ cigarette and any tobacco use [13] and greater escalation of any tobacco use [14], whereas greater parental tobacco-related rules are associated with lower levels of cigarette use [15], noncombustible product (e.g., ECIGs, SLT) use [16], and aggregated any tobacco (i.e., cigarettes, ECIGs, cigars, SLT, hookah aggregated) use [16]. Community-level factors influencing adolescents are also important to consider. Adolescents residing in rural areas consistently report higher levels of cigarette and SLT use [17,18], with some evidence suggesting lower levels of ECIG use [18]. Whether adolescents vary in their likelihood for increasing their use of a range of tobacco products as a function of geography remains largely unexplored.
To expand upon previous literature, which has largely utilized cross-sectional data and has not assessed products separately and/or comprehensively [13-15], the current study assessed: (1) adolescents’ Wave 1 (W1) use and changes in use of: cigarettes, ECIGs, traditional cigars, cigarillos, filtered cigars, snus, SLT, and hookah across the first four waves (2013—2018) of the Population Assessment of Tobacco and Health (PATH) study; and (2) individual- (i.e., internalizing/externalizing problems, sensation seeking, tobacco use motives), interpersonal- (i.e., parental monitoring, parental rules), and community-level (i.e., geographic isolation) predictors of W1 use and changes in use, accounting for sociodemographic covariates.
Materials and Methods
Sample
PATH is a national-level longitudinal survey designed to inform US tobacco policy and regulation. W1 of data collection (2013—2014) resulted in a sample of 53,178 youth (12—17 years) and adults (18 + years). These individuals were subsequently contacted at three additional waves: 2 (2014—2015), 3 (2015—2016), and 4 (2016—2018). Due to relatively low rates of youth tobacco use in the PATH dataset (see Table 1), the current analysis included a sample of 975 youth (aged 12—16 at W1) who used at least one tobacco product (≥1 day of the past 30) at one or more waves (i.e., W1-4) [19]. This sample was maintained across all analyses for each tobacco product.
Table 1.
Descriptive statistics for socioecological predictors and tobacco use outcomes
| W1 |
W2 |
W3 |
W4 |
|
|---|---|---|---|---|
| N (%) or mean (SD) | ||||
| Sociodemographics | ||||
| Age (years) | 13.29 (0.86) | – | – | – |
| Gender | ||||
| Male | – | – | – | 528 (54.2%) |
| Female | – | – | – | 447 (45.8%) |
| Race/ethnicity | ||||
| White | – | – | – | 531 (54.5%) |
| Black | – | – | – | 87 (8.9%) |
| Hispanic | – | – | – | 253 (25.9%) |
| Other | – | – | – | 104 (10.7%) |
| Parent educationa | – | – | – | 2.15 (0.85) |
| Individual-level predictors | ||||
| Internalizing problemsb | 1.65 (1.06) | – | – | – |
| Externalizing problemsb | 2.39 (0.78) | – | – | – |
| Sensation seekingc | 3.99 (0.98) | – | – | – |
| Motivesd | 1.66 (0.59) | – | – | – |
| Interpersonal-level predictors | ||||
| Parental modeling — combustiblee | 201 (20.6%) | – | – | – |
| Parental modeling — noncombustiblee | 21 (2.2%) | – | – | – |
| Parental modeling - ECIGs/hookahe | 67 (6.9%) | – | – | – |
| Parental rules — combustible (youth report)f | 1.37 (0.67) | – | – | – |
| Parental rules — combustible (parent report)f | 1.57 (0.79) | – | – | – |
| Parental rules — noncombustible (parent report)f | 1.41 (0.68) | – | – | – |
| Parental rules — noncombustible (youth report)f | 1.48 (0.74) | – | – | – |
| Contextual-level predictors | ||||
| Isolationg | – | – | – | 5.37 (0.92) |
| Past 30-day tobacco use outcomes, N (%) | ||||
| Cigarettes | 65 (6.7%) | 124 (12.7%) | 173 (17.7%) | 321 (32.9%) |
| ECIGs | 47 (4.8%) | 117 (12.0%) | 237 (24.3%) | 402 (41.2%) |
| Traditional cigars | 9 (0.9%) | 15 (1.5%) | 9 (0.9%) | 29 (3.0%) |
| Cigarillos | 23 (2.4%) | 23 (2.4%) | 46 (4.7%) | 95 (9.7%) |
| Filtered cigars | 5 (0.5%) | 10 (1.0%) | 16 (1.6%) | 29 (3.0%) |
| Snus | 8 (0.8%) | 15 (1.5%) | 21 (2.2%) | 40 (4.1%) |
| SLT | 18 (1.8%) | 34 (3.5%) | 64 (6.6%) | 91 (9.3%) |
| Hookah | 12 (1.2%) | 24 (2.5%) | 29 (3.0%) | 29 (3.0%) |
| Past 30-day tobacco use outcomes, M (SD) | ||||
| Cigarettes | 0.50 (3.00) | 1.32 (5.23) | 1.92 (6.32) | 3.11 (7.62) |
| ECIGs | 0.31 (2.23) | 0.91 (3.68) | 1.97 (5.61) | 2.84 (6.67) |
| Traditional cigars | 0.03 (0.30) | 0.08 (1.06) | 0.05 (1.03) | 0.17 (1.60) |
| Cigarillos | 0.18 (1.72) | 0.09 (0.73) | 0.24 (1.72) | 0.53 (2.77) |
| Filtered cigars | 0.02 (0.34) | 0.06 (0.93) | 0.12 (1.49) | 0.27 (2.32) |
| Snus | 0.08 (1.00) | 0.10 (1.28) | 0.13 (1.29) | 0.41 (3.00) |
| SLT | 0.13 (1.43) | 0.42 (3.04) | 0.77 (4.01) | 1.09 (4.83) |
| Hookah | 0.06 (0.93) | 0.06 (0.58) | 0.13 (2.03) | 0.04 (0.50) |
ECIGs = electronic cigarettes; SD = standard deviation; SLT = smokeless tobacco.
Parent education: 1 = less than high school, 2 = high school graduate, 3 = some college, 4 = college degree or higher.
Internalizing/externalizing problems: 1 = never, 2 = over a year ago, 3 = 2—12 months ago, 4 = past month.
Sensation seeking: 1 = strongly disagree, 5 = strongly agree.
Motives: 1 = strongly disagree, 4 = strongly agree.
Parental modeling: 0 = no parental past 30-day use, 1 = parental past 30-day use.
Parental rules: 1 = not allowed anywhere/any time, 2 = allowed in some places/sometimes, 3 = allowed anywhere/any time.
Geographic isolation: mean isolation score across all four waves.
Measures
Tobacco product use.
At each wave, individuals self-reported ever use of cigarettes, ECIGs, traditional cigars, cigarillos, filtered cigars, snus, SLT, and hookah. For those products that had ever been used (except hookah), individuals reported on their number of use days in the past month (0—30). For hookah, youth who indicated use in the past year reported whether they use every day, weekly, or monthly. Those who reported monthly use reported how often they use hookah in a month, those who reported weekly use reported how often they use hookah in a week, and those who reported daily use reported how often they use hookah in a day. Responses to weekly use were multiplied by 4 (with a maximum score of 30) and responses to daily use were recoded to 30.
Individual-level predictors
Internalizing/externalizing problems.
Internalizing and externalizing problems are subscales of the Global Appraisal of Individual Needs - Short Screener (GAIN-SS) [9]. Subscales are comprised of four items for internalizing problems (e.g., sleep trouble) and seven items for externalizing problems (e.g., lying/conning). Participants indicated when they last experienced each symptom (1 = never-4 = past month) at W1. Mean scores were calculated for internalizing (α = 0.82) and externalizing (α = 0.76) symptoms, with higher scores representing more recent symptoms.
Sensation seeking.
Sensation seeking was assessed using three items from the Brief Sensation Seeking Scale [20] (e.g., like to do frightening things; 1 = strongly disagree-5 = strongly agree) at W1. Mean scores were calculated with higher scores representing higher levels of sensation seeking (α = 0.77).
Tobacco use motives.
Six items assessed adolescents’ motives for using tobacco products (e.g., help reduce or handle stress; 1 = strongly disagree-4 = strongly agree) at W1. Mean scores were calculated with higher scores representing higher levels of positive tobacco use motives (α = 0.87).
Interpersonal-level predictors
Parent modeling.
Modeling was defined as parent tobacco use at W1 based on three items that assessed their past 30-day use of (1) combustible tobacco; (2) noncombustible tobacco; and (3) ECIGs or hookah (0 = no, 1 = yes). The item included in the models as a predictor differed according to product type: combustible tobacco products (i.e., cigarette and cigar product models), noncombustible tobacco products (i.e., snus and SLT models), or ECIGs or hookah (i.e., ECIG and hookah models).
Parent- and youth-reported household smoking rules.
Parents and youth reported on household tobacco use rules via 2 separate questions, which distinguished between rules for products that are burned (including cigarettes, cigars, and hookah) versus not burned (including ECIGs and SLT) at W1. Response options for these questions were not allowed anywhere or at any time (coded 1), allowed in some places or at some times (coded 2), or allowed anywhere and at any time (coded 3). The item included in the models as a predictor differed according to product type: burned (i.e., cigarette, cigar products, and hookah models) or not burned (i.e., ECIG, snus, and SLT models).
Contextual-level predictors
Geographic isolation.
Participants’ state-level geographic isolation scores are based on the geographic isolation rurality measure produced and validated by Doogan et al. (2018) [21]. Geographic isolation represents the degree to which an individual is isolated from health-promoting resources (e.g., medical clinics, employment opportunities) based on their location of residence. The original tract-level measure was aggregated to the state level by taking the weighted mean of all isolation values associated with each state. The weight was the size of the population in the tract; therefore, the state-level measure approximately represents the average geographic isolation experienced by people living in the state. As an extreme example, if a large-area state had one city and every state resident lived in that city, the geographic isolation of the state would approximate the geographic isolation of the city. If half of the population lived in the very isolated rural areas of the state and the other half in the city, the state-level value would be approximately the middle point between the city’s isolation and the isolation of the rural areas. The balance depends proportionally on where the population lives. Higher scores reflect higher levels of isolation. An isolation score was computed for each participant at each wave. Because the majority of participants remained stable in their geographic isolation over time (i.e., only 16 [1.6%] moved at any point), participants’ mean scores across all four waves were used for analysis.
Sociodemographic predictors.
Adolescents reported on their age (assessed continuously), gender (male [referent], female), parent education (less than high school, high school graduate, some college, and college [4 years] graduate), and race/ethnicity (Hispanic, non-Hispanic White, non-Hispanic Black, or non-Hispanic other [Asian, American Indian/Alaskan Native, Native Hawaiian/Pacific Islander, and more than one race combined due to low sample sizes for each]). Individuals who self-identified as Hispanic, independent of all other race/ethnic identifiers, were classified as such by PATH.
Data analyses
Descriptive statistics characterized the sample and screened for outliers. Using Mplus 8.6, we utilized unconditional LGM to assess changes in adolescents’ past 30-day use frequency of eight tobacco products across all four waves of data without covariates. Models with linear components only and models with linear and quadratic components were sequentially tested. Model fit was assessed using Comparative Fit Index (CFI; ≥0.90), Tucker-Lewis Index (TLI; ≥0.90), Root Mean Square Error of Approximation (RMSEA; ≤0.08), and Standardized Root Mean Squared Residual (SRMR; ≤0.08) [22]. All individual- (i.e., internalizing/externalizing problems, sensation seeking, motives), interpersonal- (i.e., parental modeling, rules), and community-(i.e., geographic isolation)level factors were then examined as predictors of intercept and slope. W1 age and W4 gender, race, and parent education were included as covariates (due to no missing data at W4) in all conditional models. A separate model was examined for each of the eight products (i.e., cigarettes, ECIGs, traditional cigars, cigarillos, filtered cigars, snus, SLT, hookah). Missing data were estimated using full information maximum likelihood.
Results
Descriptive analyses
Descriptive statistics are reported in Table 1. Use prevalence increased across Waves 1—4 for each product. At W1, rates of past 30-day use were highest for cigarettes (6.7%) followed by e-cigarettes (4.8%). However, e-cigarette use rates (24.3%) surpassed those for cigarettes at W3 (17.7%) and W4 (41.2% for e-cigarettes, 32.9% for cigarettes).
Tobacco use changes
In separate LGMs per product, linear (vs. quadratic) growth models were chosen based on model fit criteria balanced with relatively low levels of use across products, leading to difficulties estimating complex models (i.e., conditional quadratic LGMs with multiple predictors) with only four waves of data. Additionally, chi-square differences tests were nonsignificant for each model, indicating that the model with linear and quadratic slopes did not provide a significantly better fit than the model with linear slopes only. Quadratic slopes were nonsignificant across all products, providing further justification for examining linear growth models. Moreover, relative model fit indices suggested optimal balance between fit and parsimony with linear but not quadratic slopes. Based on most model fit criteria, linear growth models provided an adequate fit to the data across products, with the exception of cigarillos for which CFI and TLI were low (0.577 and 0.492, respectively) (Table 2). Due to low mean levels of past 30-day use across waves for traditional cigars (M = 0.08 days) and hookah (M = 0.07 days) relative to other products (M = 0.12 days[filtered cigars]-M = 1.71 days[cigarettes]), as well as poor model fit, LGMs were not interpreted for these two products.
Table 2.
Unstandardized coefficients (and standard errors) for unconditional LGMs of past 30-day tobacco use from W1 to 4, N = 975
| Cigarettes |
ECIGs |
Cigarillos |
Filtered cigars |
Snus |
SLT |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| B (SE) | p | B (SE) | p | B (SE) | p | B (SE) | p | B (SE) | p | B (SE) | p | |
| Mean | ||||||||||||
| Intercept | 0.44 (0.10) | <.001 | 0.26 (0.07) | <.001 | 0.01 (0.04) | .800 | 0.02 (0.01) | .113 | 0.07 (0.03) | .022 | 0.13 (0.05) | .008 |
| Slope | 0.83 (0.08) | <.001 | 0.82 (0.06) | <.001 | 0.12 (0.03) | <.001 | 0.06 (0.02) | .004 | 0.04 (0.02) | .041 | 0.32 (0.05) | <.001 |
| Variance | ||||||||||||
| Intercept | 8.51 (0.65) | <.001 | 1.30 (0.44) | .003 | 0.31 (0.09) | <.001 | −0.01 (0.02) | .548 | 0.12 (0.07) | .077 | 1.77 (0.20) | <.001 |
| Slope | 3.60 (0.29) | <.001 | 0.90 (0.23) | <.001 | 0.21 (0.05) | <.001 | 0.23 (0.02) | <.001 | 0.04 (0.03) | .132 | 1.14 (0.10) | <.001 |
| Covariance: Int. with slope | −1.12 | .001 | −0.24 | .341 | −0.20 | .001 | 0.02 | .214 | 0.197 | <.001 | 0.150 | .167 |
| Model fit information | ||||||||||||
| Linear | ||||||||||||
| CFI | 0.932 | 0.844 | 0.577 | 0.905 | 0.940 | 0.924 | ||||||
| TLI | 0.918 | 0.812 | 0.492 | 0.886 | 0.928 | 0.909 | ||||||
| RMSEA | 0.112 | 0.048 | 0.060 | 0.088 | 0.064 | 0.109 | ||||||
| SRMR | 0.050 | 0.033 | 0.040 | 0.053 | 0.048 | 0.060 | ||||||
| Quadratic | ||||||||||||
| CFI | 0.998 | 0.997 | 1.00 | 1.00 | 1.00 | 1.00 | ||||||
| TLI | 0.991 | 0.981 | 1.00 | 1.00 | 1.00 | 1.00 | ||||||
| RMSEA | 0.038 | 0.015 | 0.000 | 0.000 | 0.000 | 0.000 | ||||||
| SRMR | 0.009 | 0.009 | 0.004 | 0.003 | 0.000 | 0.001 | ||||||
CFI = Comparative Fit Index; ECIGs = electronic cigarettes; LGM = latent growth modeling; RMSEA = Root Mean Square Error of Approximation; SLT = smokeless tobacco; SRMR = Standardized Root Mean Squared Residual; TLI = Tucker-Lewis Index.
Int = intercept. Model fit for traditional cigars and hookah was poor, precluding running further analyses on these products. Traditional cigars (linear): CFI = 0.672, TLI = 0.545, RMSEA = 0.029, SRMR = 0.019; traditional cigars (quadratic): CFI = 0.795, TLI = 0.680, RMSEA = 0.018, SRMR = 0.010; hookah (linear): CFI = 0.504, TLI = 0.073, RMSEA = 0.028, SRMR = 0.022; hookah (quadratic): CFI = 0.650, TLI = 0.091, RMSEA = 0.020, SRMR = 0.018.
Italicized values denote statistical significance at alpha = 0.05.
Significant positive linear slopes indicated increases in past 30-day use of all products (i.e., cigarettes, ECIGs, cigarillos, filtered cigars, snus, SLT), on average, across the four waves (Table 2, Figure 1). Results suggested individual variability around mean intercept (for cigarettes, ECIGs, cigarillos, and SLT) and slope (for cigarettes, ECIGs, cigarillos, filtered cigars, and SLT), which may be explained by the addition of predictors (Table 2). Because individual variability around mean intercept was not significant for filtered cigars and snus or slope for snus, findings related to predictors of W1 filtered cigar and snus use, as well as change in snus use are not reported.
Figure 1.
Average number of days used in the past month for each product across W1-4.
Predictors of W1 tobacco use and changes in tobacco use over time
Regarding individual-level factors, higher tobacco-use motives predicted greater W1 cigarette, ECIG, and SLT use and greater externalizing problems predicted greater W1 cigarillo use. Additionally, higher tobacco-use motives predicted less ECIG use across time, and greater externalizing problems predicted less SLT use across time (Table 3).
Table 3.
Unstandardized coefficients (and standard errors) for conditional LGMs of effects of sociodemographics, individual-, interpersonal-, and community-level predictors of intercepts and slopes of past 30-day tobacco product use from W1-4, N = 975
| Cigarettes |
ECIGs |
Cigarillos |
Filtered cigars |
Snus |
SLT |
|||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Int. | Slope | Int. | Slope | Int. | Slope | Int. | Slope | Int. | Slope | Int. | Slope | |
| Sociodemographics | ||||||||||||
| Age | 0.01 (0.12) | 0.22 (0.09) | −0.02 (0.09) | 0.10 (0.08) | 0.02 (0.05) | 0.01 (0.04) | −0.01 (0.01) | 0.04 (0.02) | −0.03 (0.04) | 0.04 (0.03) | 0.02 (0.06) | 0.02 (0.05) |
| Female | 0.07 (0.21) | 0.03 (0.16) | −0.12 (0.16) | −0.35 (0.14) | 0.11 (0.09) | −0.10 (0.07) | −0.01 (0.03) | −0.06 (0.04) | −0.11 (0.07) | −0.03 (0.04) | −0.16 (0.10) | −0.56 (0.10) |
| Black | −0.19 (0.36) | −0.37 (0.27) | −0.14 (0.26) | −0.75 (0.23) | −0.10 (0.15) | 0.43 (0.12) | −0.06 (0.04) | −0.02 (0.07) | −0.13 (0.11) | −0.02 (0.07) | −0.22 (0.17) | −0.34 (0.16) |
| Hispanic | −0.51 (0.25) | −0.12 (0.19) | 0.22 (0.19) | −0.43 (0.17) | −0.05 (0.11) | 0.03 (0.08) | −0.05 (0.03) | 0.03 (0.05) | −0.13 (0.08) | −0.04 (0.05) | −0.08 (0.13) | −0.31 (0.11) |
| Other | 0.02 (0.34) | 0.10 (0.26) | −0.35 (0.25) | 0.41 (0.21) | −0.16 (0.14) | 0.11 (0.11) | −0.06 (0.04) | 0.07 (0.07) | −0.12 (0.11) | 0.01 (0.07) | −0.14 (0.16) | −0.16 (0.15) |
| Parent education | −0.15 (0.12) | −0.10 (0.10) | 0.14 (0.09) | 0.01 (0.08) | −0.03 (0.05) | −0.01 (0.04) | −0.01 (0.01) | −0.02 (0.03) | −0.05 (0.04) | 0.01 (0.03) | −0.13 (0.06) | −0.06 (0.05) |
| Individual-level factors | ||||||||||||
| Internalizing problems | 0.18 (0.13) | −0.04 (0.10) | 0.02 (0.09) | 0.04 (0.08) | −0.09 (0.05) | 0.06 (0.04) | −0.01 (0.02) | 0.02 (0.03) | −0.01 (0.04) | −0.02 (0.03) | −0.05 (0.06) | 0.04 (0.06) |
| Externalizing problems | −0.06 (0.17) | −0.01 (0.13) | −0.09 (0.12) | 0.20 (0.11) | 0.15 (0.07) | −0.09 (0.06) | 0.02 (0.02) | 0.02 (0.04) | 0.02 (0.06) | −0.02 (0.04) | −0.03 (0.08) | −0.19 (0.08) |
| Sensation seeking | −0.05 (0.12) | 0.05 (0.09) | 0.08 (0.09) | 0.07 (0.08) | 0.01 (0.05) | −0.01 (0.04) | −0.01 (0.01) | −0.01 (0.03) | −0.10 (0.04) | 0.01 (0.04) | 0.02 (0.06) | 0.03 (0.06) |
| Tobacco use motives | 1.08 (0.19) | 0.26 (0.15) | 0.64 (0.14) | −0.34 (0.12) | 0.06 (0.08) | −0.04 (0.06) | 0.07 (0.02) | −0.01 (0.04) | 0.34 (0.06) | 0.04 (0.02) | 0.43 (0.09) | 0.03 (0.09) |
| Interpersonal-level factors | ||||||||||||
| Parent modelinga | 0.32 (0.25) | 0.14 (0.20) | 0.37 (0.27) | 0.36 (0.25) | 0.19(0.11) | −0.10 (0.08) | 0.02 (0.03) | −0.03 (0.05) | −0.31 (0.21) | 0.30 (0.15) | −0.21 (0.31) | 0.37 (0.30) |
| Rules (parent report)b | 0.23 (0.18) | 0.17 (0.14) | 0.01 (0.10) | −0.03 (0.09) | 0.04 (0.08) | 0.01 (0.06) | 0.04 (0.02) | 0.01 (0.04) | 0.04 (0.05) | −0.07 (0.03) | −0.01 (0.07) | −0.03 (0.07) |
| Rules (youth report)b | 0.01 (0.17) | −0.01 (0.13) | −0.07 (0.11) | 0.01 (0.10) | −0.18 (0.07) | 0.17 (0.06) | −0.01 (0.02) | 0.01 (0.04) | 0.01 (0.05) | 0.01 (0.04) | −0.03 (0.07) | 0.14 (0.07) |
| Community-level factor | ||||||||||||
| Geographic isolation | −0.01 (0.11) | 0.25 (0.09) | 0.17 (0.08) | −0.09 (0.07) | −0.05 (0.05) | 0.04 (0.04) | 0.01 (0.01) | 0.04 (0.02) | 0.07 (0.04) | 0.04 (0.02) | 0.04 (0.06) | 0.15 (0.05) |
| Model fit information | ||||||||||||
| CFI | 0.996 | 0.960 | 0.998 | 0.957 | 0.997 | 0.970 | ||||||
| TLI | 0.982 | 0.900 | 0.990 | 0.924 | 0.988 | 0.930 | ||||||
| RMSEA | 0.018 | 0.025 | 0.010 | 0.032 | 0.009 | 0.044 | ||||||
| SRMR | 0.007 | 0.017 | 0.005 | 0.017 | 0.009 | 0.020 | ||||||
CFI = Comparative Fit Index; ECIGs = electronic cigarettes; LGM = latent growth modeling; RMSEA = Root Mean Square Error of Approximation; SLT = smokeless tobacco; SRMR = Standardized Root Mean Squared Residual; TLI = Tucker-Lewis Index.
Bolded values denote statistical significance at alpha = 0.05. Int = Intercept. All predictors were treated as time invariant, as several were not assessed at all waves and those that were (i.e., internalizing problems, externalizing problems, parent report of rules, parent modeling) did not significantly change over time.
Parental modeling of combustible tobacco products was included in cigarette, cigarillo, and filtered cigar models. Parental modeling of noncombustible tobacco products was included in snus and SLT models. Parental modeling of ECIGs was included in ECIG model.
Parent- and youth-report of rules for combustible products was included in cigarette, cigarillo, and filtered cigar models. Parent- and youth-report of rules for noncombustible products was included in ECIG, snus, and SLT models.
At the interpersonal level, greater youth report of parental household smoking rules predicted greater W1 cigarillo use and less cigarillo and SLT use across time.
With regard to community-level factors, adolescents living in states with greater geographic isolation reported greater W1 ECIG use and more cigarette and SLT use across time.
Several associations emerged among sociodemographics and use. Hispanic (vs. non-Hispanic White) adolescents reported less W1 cigarette use and higher parental education predicted less W1 SLT use. Older adolescents reported more cigarette use across time and female adolescents reported less ECIG and SLT use across time. Adolescents identifying as Black or Hispanic (vs. non-Hispanic White) reported less ECIG and SLT use across time. Adolescents identifying as Black (vs. non-Hispanic White) reported more cigarillo use across time.
Sensitivity analyses
We ran a series of sensitivity analyses to ensure that findings were not due to suppression (i.e., the inclusion of a variable that changes the predictive ability of another variable by its inclusion in a regression equation). All significant findings from the original LGMs (including all predictors) were also present when examining models for each predictor separately.
Discussion
Consistent with reported trends [23,24], ECIGs became the most commonly used product among adolescent users of any tobacco product by W3 (2015—2016), surpassing rates of cigarette use. Despite decreases in overall rates of adolescents’ tobacco use (with the exception of ECIGs) over the past decade [25,26], current findings suggest that, on average, adolescents who used at least one tobacco product at any wave increased their past 30-day use frequency of each product from 2013 to 2018. Although adolescents demonstrated significant increases in their past 30-day use of each product, they reported relatively few days of use across products over time. For instance, adolescents’ cigarette use increased from 0.50 days (W1) to 3.11 days (W4) and their ECIG use increased from 0.31 days (W1) to 2.84 days (W4). Days of use in the past 30 days remained even lower for other tobacco products. This is consistent with 2020 data indicating that adolescents used combustible products and SLT an average of 0.80 days and ECIGs an average of 2.60 days of the past 30 [27].
Leveraging a socioecological framework, we documented multilevel predictors of adolescents’ tobacco use changes. At the individual level, positive tobacco-use motives predicted greater W1 use across all products except cigarillos, suggesting that individuals with greater positive tobacco-related motives (e.g., stress reduction, weight management) are more likely to engage in such use, consistent with previous work [12]. Unexpectedly, positive motives were associated with decreases in ECIG use. Because the items used to assess motives asked about motives to use tobacco in general, adolescents in the PATH study may have responded to these items with more traditional tobacco products (e.g., cigarettes) in mind.
Also at the individual level, adolescents’ externalizing problems were associated positively with W1 cigarillo use. Adolescents with greater (vs. fewer) externalizing problems engage in greater deviant behaviors (e.g., substance use) with peers [28]. Notably, adolescents report a preference for using cigarillos in social settings with peers to a greater degree than other tobacco products [29]. Externalizing problems were also associated with less SLT use across time. An important consideration in interpreting this finding is polytobacco use. Previous research has primarily documented associations among externalizing problems and adolescents’ cigarette or ECIG use [10]. Thus, while these adolescents may be decreasing their SLT use, they may be initiating or continuing use of other tobacco products. However, this finding warrants further investigation.
Regarding interpersonal-level factors, youth-report of tobacco-related household rules were associated with tobacco use over and above parents’ modeling of such use. Greater youth-reported household rules for combustible tobacco products were associated with less cigarillo and SLT use across time, consistent with a large body of literature demonstrating the protective role of limit-setting in reducing adolescents’ risk for tobacco use [15,16]. Although the current study assessed parental household smoking rules (i.e., whether smoking is allowed in the home), rather than rules over the adolescent’s engagement in tobacco use, findings provide evidence for the impact of such rules on adolescents’ tobacco use. Taken together, findings stress the importance of parents’ engagement in anti-tobacco behaviors over and above their own use in predicting adolescent tobacco use.
Regarding the contextual level, greater isolation predicted higher levels of W1 ECIG use. Some findings suggest that adolescents residing in more (vs. less) rural areas who use other tobacco products (primarily cigarettes and SLT) are at greater risk for also using ECIGs [30]. Greater state-level isolation also predicted more cigarette and SLT use across time, consistent with an established body of research suggesting elevated rates of both cigarette and SLT use among adolescents who reside in more rural areas [17,18]. Although rural adolescents report trying cigarettes and SLT with the intentions of using such products for the short-term only [31], our findings suggest that these intentions may not reflect their longitudinal patterns of use, which suggest increases over time. Because more isolated youth live in contexts with weaker tobacco control policies, greater tobacco advertising, and greater pro-tobacco social norms combined with less access to cessation [32,33], these findings highlight the need for tailored intervention efforts specifically focused on cigarettes and SLT for these rural youth.
Findings regarding sociodemographics and W1 use are supported by other research suggesting that non-Hispanic White (vs. Hispanic) adolescents are more likely to use cigarettes [34], and build on findings suggesting associations between socioeconomic status and SLT use among adults [35]. However, expanding upon the more limited literature regarding sociodemographic predictors of changes in a range of tobacco products, older adolescents reported more cigarette use across time. Additionally, male and White (vs. Black and Hispanic) adolescents reported more ECIG and SLT use across time, whereas Black (vs. White) adolescents reported more cigarillo use across time.
The current results should be interpreted in light of a few limitations. First, certain predictors (e.g., parental rules, parental tobacco use) were aggregated across products within the survey (e.g., combustible products, noncombustible products) and we were unable to examine associations among product-specific predictors and adolescents’ use of specific products. Second, adolescents in the current study reported relatively low levels of past 30-day use for some products (i.e., traditional cigars, hookah), interfering with our ability to assess socioecological predictors of changes in such use. However, given such low rates of use of these products in the current study, it may not be clinically meaningful to examine predictors of their use. Third, due to low levels of past 30-day tobacco use, our analytic sample was limited only to those adolescents who reported using any tobacco product at any wave, limiting the generalizability of our findings. However, because associations among known predictors (i.e., internalizing problems [9,10], sensation seeking [11]) of adolescent tobacco use were not observed in the current study, future work utilizing additional measures of internalizing symptoms and sensation seeking, as well as samples with greater variability in adolescent tobacco use should explore these associations. Such work may clarify whether null findings observed in the current study are related to measurement, reduced power to detect associations, or whether these factors are not clinically meaningful predictors of use. Moreover, low levels of past 30-day use precluded us from examining how adolescents varied in their trajectories of use of each product through the use of other statistical approaches (e.g., growth mixture modeling).
Taken together, our findings indicate that while associations among multilevel predictors and W1 use were largely similar across products (i.e., motives associated with greater W1 use across most products), associations among predictors and changes in use over time were more product specific. Thus, prevention and intervention efforts should be tailored to specific products in which adolescents engage and should consider the demographic characteristics of intended participants. Specifically, prevention efforts should focus on targeting externalizing problems for cigarillos and tobacco use motives across all products. However, intervention efforts aimed at reducing escalation in use may benefit from targeting older adolescents for cigarettes, non-Hispanic White males for ECIGs and SLT, Black adolescents for cigarillos, and adolescents in more isolated geographic contexts for cigarettes and SLT. Important targets for change in intervention include parental rules for cigarillos and SLT. Consideration of these factors enhances the likelihood that prevention and cessation efforts will help enhance health equity among vulnerable adolescent populations. As ECIGs and cigarettes remained the most commonly used products among adolescents, findings stress the importance of developing tailored prevention and intervention efforts aimed specifically at preventing and reducing youth’s engagement in these products in particular.
IMPLICATIONS AND CONTRIBUTION.
Findings suggest that adolescents displayed increases in use over time for each tobacco product. Interventions aimed at reducing escalations in use should target older adolescents (cigarettes), non-Hispanic White males (ECIGs, SLT), Black adolescents (cigarillos), adolescents residing in geographically isolated contexts (cigarettes, SLT), and adolescents with fewer parental rules (cigarillos, SLT).
Funding Sources
This publication was supported by the National Institute on Drug Abuse of the National Institutes of Health (NIH) and the Center for Tobacco Products of the U.S. Food and Drug Administration (FDA) (R21DA051628; PI: Blank). Dr. Romm is supported by the National Institute on Drug Abuse (R25DA054015; MPIs: Obasi, Reitzel), the Oklahoma Tobacco Settlement Endowment Trust (TSET) contract #R22-03, and the National Cancer Institute grant awarded to the Stephenson Cancer Center (P30CA225520).
Footnotes
Conflicts of interest: The authors declare no conflicts of interest.
Disclaimer: The content is solely the responsibility of the authors and does not necessarily represent the views of the NIH or FDA.
References
- [1].Gentzke AS, Creamer M, Cullen KA, et al. Vital Signs: Tobacco product use among middle and high school students - United States, 2011-2018. MMWR Morb Mortal Wkly Rep 2019;68:157–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Ahun MN, Lauzon B, Sylvestre MP, et al. A systematic review of cigarette smoking trajectories in adolescents. Int J Drug Policy 2020;83:102838. [DOI] [PubMed] [Google Scholar]
- [3].Lanza HI, Barrington-Trimis JL, McConnell R, 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]
- [4].Defoe IN, Semon Dubas J, Somerville LH, et al. The unique roles of intra-personal and social factors in adolescent smoking development. Dev Psychol 2016;52:2044–56. [DOI] [PubMed] [Google Scholar]
- [5].Harrell MB, Chen B, Clendennen SL, et al. Longitudinal trajectories of e-cigarette use among adolescents: A 5-year, multiple cohort study of vaping with and without marijuana. Prev Med 2021;150:106670. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [6].Loukas A, Marti CN, Perry CL. Trajectories of tobacco and nicotine use across young adulthood, Texas, 2014-2017. Am J Public Health 2019;109:465–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Ebrahimi Kalan M, Bahelah R, Bursac Z, et al. Predictors of nicotine dependence among adolescent waterpipe and cigarette smokers: A 6-year longitudinal analysis. Drug Alcohol Depend 2020;217:108346. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [8].Szapocznik JCJ, Glantz M, Hartel C. An ecodevelopmental framework for organizing the influences on drug abuse: A developmental model of risk and protection. In: CH M Glantz, ed. Drug abuse: Origins and interventions. Washington, DC: American Psychological Association; 1999. [Google Scholar]
- [9].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(Suppl 1):80–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Riehm KE, Young AS, Feder KA, 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]
- [11].Hiemstra M, Rozema A, Jansen M, et al. Longitudinal associations of substance use risk profiles with the use of alternative tobacco products and conventional smoking among adolescents. Int J Environ Res Public Health 2021;18:13248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Creamer MR, Delk J, Case K, et al. Positive outcome expectations and tobacco product use behaviors in youth. Subst Use Misuse 2018;53:1399–402. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].Vuolo M, Staff J. Parent and child cigarette use: A longitudinal, multigenerational study. Pediatrics 2013;132:e568–77. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [14].Simons-Morton B, Chen R, Abroms L, Haynie DL. Latent growth curve analyses of peer and parent influences on smoking progression among early adolescents. Health Psychol 2004;23:612–21. [DOI] [PubMed] [Google Scholar]
- [15].Emory K, Saquib N, Gilpin EA, Pierce JP. The association between home smoking restrictions and youth smoking behaviour: A review. Tob Control 2010;19:495–506. [DOI] [PubMed] [Google Scholar]
- [16].Wu TS, Chaffee BW. Parental awareness of youth tobacco use and the role of household tobacco rules in use prevention. Pediatrics 2020;146:e20194034. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Chaffee BW, Cheng J. Cigarette and smokeless tobacco perception differences of rural male youth. Tob Regul Sci 2018;4:73–90. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Noland M, Rayens MK, Wiggins AT, et al. Current use of e-cigarettes and conventional cigarettes among US high school students in urban and rural locations: 2014 national youth tobacco survey. Am J Health Promot 2018;32:1239–47. [DOI] [PubMed] [Google Scholar]
- [19].Romm KF, Childers MG, Douglas AE, et al. Transitions in tobacco use profiles among adolescents: Results from the Population Assessment of Tobacco and Health (PATH) study waves 3 and 4. Drug Alcohol Depend 2022;232:109272. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Stephenson MT, Hoyle RH, Palmgreen P, Slater MD. Brief measures of sensation seeking for screening and large-scale surveys. Drug Alcohol Depend 2003;72:279–86. [DOI] [PubMed] [Google Scholar]
- [21].Doogan NJ, Roberts ME, Wewers ME, et al. Validation of a new continuous geographic isolation scale: A tool for rural health disparities research. Soc Sci Med 2018;215:123–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Kline RB. Principles and practice of structural equation modeling. New York: The Guilford Press; 2005. [Google Scholar]
- [23].Wang TW, Gentzke A, Sharapova S, et al. Tobacco product use among middle and high school students - United States, 2011-2017. MMWR Morb Mortal Wkly Rep 2018;67:629–33. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [24].U.S. Department of Health and Human Services. Smoking cessation. A report of the surgeon general. Atlanta, GA: U.S. Department of Health and Human Services, Centers for Disease Control and Prevention, National Center for Chronic Disease Prevention and Health Promotion, Office on Smoking and Health, 2020. [Google Scholar]
- [25].Gentzke AS, Wang TW, Jamal A, et al. Tobacco product use among middle and high school students - United States, 2020. MMWR Morb Mortal Wkly Rep 2020;69:1881–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [26].Wang TW, Gentzke AS, Creamer MR, et al. Tobacco product use and associated factors among middle and high school students - United States, 2019. MMWR Surveill Summ 2019;68:1–22. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [27].Sun R, Mendez D, Warner KE. Trends in nicotine product use among US adolescents, 1999-2020. JAMA Netw Open 2021;4:e2118788. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Chen D, Drabick DA, Burgers DE. A developmental perspective on peer rejection, deviant peer affiliation, and conduct problems among youth. Child Psychiatry Hum Dev 2015;46:823–38. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Mead EL, Chen JC, Kirchner TR, et al. An ecological momentary assessment of cigarette and cigar dual use among African American young adults. Nicotine Tob Res 2018;20:S12–S21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Blank MD, Ozga JE, Romm KF, et al. Geographic isolation predicts tobacco product use among youth: A latent class analysis. J Rural Health 2022;38:373–81. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [31].Couch ET, Darius E, Walsh MM, Chaffee BW. Smokeless tobacco decision-making among rural adolescent males in California. J Community Health 2017;42:544–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Roberts ME, Doogan NJ, Kurti AN, et al. Rural tobacco use across the United States: How rural and urban areas differ, broken down by census regions and divisions. Health Place 2016;39:153–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [33].Henriksen L, Schleicher NC, Johnson TO, et al. Retail tobacco marketing in rural versus nonrural counties: Product availability, discounts, and prices. Health Promot Pract 2020;21:27S–36S. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [34].Cho B, Hirschtick JL, Usidame B, et al. Sociodemographic patterns of exclusive, dual, and polytobacco use among U.S. High school students: A comparison of three nationally representative surveys. J Adolesc Health 2021;68:750–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [35].White TJ, Redner R, Bunn JY, Higgins ST. Do socioeconomic risk factors for cigarette smoking extend to smokeless tobacco use? Nicotine Tob Res 2016;18:869–73. [DOI] [PMC free article] [PubMed] [Google Scholar]