Associations Between Family and Peer E-Cigarette Use With Adolescent Tobacco and Marijuana Usage: A Longitudinal Path Analytic Approach

Abstract

Introduction

Research indicates a link between adolescent e-cigarette use and combustible tobacco cigarette (CTC) initiation, and recent studies suggest their connection with marijuana uptake. Our 3-year longitudinal cohort study investigated the implications of adolescent, peer, and family e-cigarette use with adolescents’ expectations and willingness to initiate CTC use, and subsequent CTC and marijuana use.

Aims and Methods

Relationships were examined in a secondary analysis of a 3-year longitudinal cohort subsample involving adolescents enrolled in alternative California high schools (N = 1025). Analyses examined responses over three yearly observations. Family, peer, and respondents’ e-cigarette use, respondents’ positive cigarette expectancies and willingness to use CTCs were assessed in the study’s first year (T1). CTC use in the survey’s second year (T2) and marijuana use in the third year (T3) were assessed via path analysis.

Results

Respondents reporting at least one family member or peer using e-cigarettes were more likely to use e-cigarettes at T1 than those whose peers/family members did not. They reported more positive expectancies about CTCs and greater willingness to initiate use. These variables predicted CTC use at T2, which directly anticipated marijuana use in the survey’s third year (T3), as did adolescents’ use of e-cigarettes at T1. All model relations were statistically significant.

Conclusions

Analysis demonstrated the strong association of family members’ and peers’ behaviors with adolescent e-cigarette use, and the temporal precedence of e-cigarette use with subsequent CTC and marijuana uptake. The predictive implications of e-cigarettes for other dangerous substance use should be examined in future prevention campaigns.

Implications

The presented study expands upon existing literature connecting adolescent e-cigarette use and later CTC and marijuana use. The findings indicate the significant implications of exposure to e-cigarette use by parents and peers and demonstrate in a longitudinal 4-year panel survey the direct and indirect predictive implications of e-cigarette use for CTC and marijuana uptake. The research illustrates the utility of programs and campaigns that target peer and family groups to maximize impacts on adolescent willingness to try CTCs, positive expectancies, and possible onset of CTC and marijuana use.

Introduction

Few topics in public health today are as contentious as those involving the use of electronic cigarettes (also termed e-cigarettes, vapes, ENDS, etc.), handheld battery-operated devices that range from instruments resembling conventional combustible tobacco cigarettes (CTCs) to vape pens and vaporizers.^1,2 Unlike CTCs that burn tobacco to generate smoke, e-cigarettes heat liquids containing nicotine, propylene glycol or glycerin, flavorings, and other compounds to create inhalable vapor. Both CTCs and e-cigarettes deliver nicotine, tobacco’s addictive agent, but e-cigarette manufacturers insist their products do not expose users to many chemical constituents responsible for tobacco-related diseases, thus offering the tantalizing prospect of reducing the harms of CTCs, a leading cause of preventable death and disability worldwide.^2,3 In addition to research on these claims, recent research also has focused attention on the association of e-cigarette use in adolescents and onset of CTC and marijuana use, the focus of present concerns.

Adolescent E-Cigarette Use

A Monitoring the Future (MTF) analysis of e-cigarette use by US adolescents found ever-use rates of 36% in 12th-grade students,⁴ representing a clear danger signal, as e-cigarettes may attract young smokers for whom regular CTC use is initially unappealing.⁵ Youth subsequently may become addicted to nicotine, and once smoking is normalized, transition to harmful tobacco products.^6–8 Further, adolescents may be more prone than adults to nicotine addiction after e-cigarette exposure due to the nature of the still-developing brain, which during adolescence is especially sensitive to nicotine.⁹

The possibility that adolescent e-cigarette use may lead CTC use has been supported in prior research, with some studies examining the relation longitudinally.^10,11 Soneji et al.¹⁰ examined this relationship in a meta-analysis of nine studies that supported the link between e-cigarette use and later CTC initiation. Their findings demonstrated the combined odds ratio for later CTC initiation was nearly four times greater for ever versus never e-cigarette users after controlling for known demographic, psychosocial, and behavioral risk factors.

Research also has examined the connection between e-cigarette use and marijuana initiation. Studies by Dai et al.¹² and Audrain-McGovern et al.¹³ examined this link longitudinally, and both found that adolescents who reported ever using an e-cigarette were significantly more likely to report initiation of marijuana in 1- and 2-year follow-ups. These studies did not investigate motivators for CTC initiation.

Research has examined youths’ perceptions of e-cigarettes to identify potential usage motivators. Among many risk factors, attitudes and family or peer use of e-cigarettes have been identified as significant predictors of youth e-cigarette usage.^14,15 These risk variables also have been identified as strong predictors of marijuana initiation. Variables implicated in marijuana initiation, which may follow from e-cigarette use, involve parent and peer behaviors, and attitudes, expectancies, and willingness to engage in risky behavior. Parental behaviors, for example, have been identified as critical in the initiation or prevention of psychotropic substance use.¹⁶ E-cigarette use by adolescents’ family members may influence e-cigarette initiation, which in turn may encourage CTC and marijuana use.¹⁶ Similarly, normative pressures in adolescence have been shown to affect marijuana usage, with noteworthy consequences for adolescent health.^14,17–19 Attitudes toward marijuana and positive expectations of their effects also have been related strongly to usage.^16,20

Family Influence

The impact of family substance use on adolescent initiation of CTC or marijuana use is well documented.^21,22 Many studies have found substance use by immediate family members predicts uptake or concurrent use by adolescent family members.^22,23 However, the association between family members’ and peers’ e-cigarette use on abstinent adolescents’ e-cigarette use has attracted only limited examination.^23,24 Carey et al.²⁴ found adolescents who reported having a family member who used e-cigarettes were significantly more likely to initiate e-cigarettes than those reporting no family member use. Lozano et al.²⁵ examined this relationship in middle school students in Mexico. Their results indicated student susceptibility to e-cigarette use and self-reported e-cigarette use were higher in those reporting family members used e-cigarettes. However, after thoroughly examining the literature we have found no studies that have investigated the link between family e-cigarette use, self-reported e-cigarette use, and later use of CTCs and marijuana. The current study is designed to extend the literature on this linkage.

Overview and Hypotheses

Generalizing from the CTC–marijuana use literature, our study was designed to investigate to the role of self-reported e-cigarette use, peer and family e-cigarette use, and CTC expectations, attitudes, and willingness to enact risky behavior on adolescents’ susceptibility to subsequent CTC and marijuana use. Respondents were attendees of alternative high schools in California. These schools serve students unable to attend regular high schools for various reasons, including low school achievement, conduct problems, and substance use. As a group, these students are at considerably higher risk than those attending traditional high schools for a number of health-related problems, including use of psychoactive substances.²⁶ For example, El-Shahawy et al.²⁷ found self-reported past month CTC use among these students ranged between 36% and 42%, up to three times the average of Californian traditional high school students (13.6%). Our analysis was designed to extend this research by examining the longitudinal structural relations of e-cigarette exposure with CTC and marijuana use in this at-risk group. It examined predictive relationships at T1 between normative variables, and e-cigarette use with adolescent cigarette use in the second year of the study (T2) and marijuana use in study’s third year (T3).

Hypotheses

We hypothesize respondents with family members or peers reporting e-cigarette use will exhibit greater use of e-cigarettes (H1). Further, adolescents’ e-cigarette use will be associated with more positive expectancies toward CTCs (H2) and greater willingness to try them (H3). Willingness to try CTCs at T1 (H4) and positive cigarette expectancies at T1 (H5) will be positively associated with CTC use (H6) at T2. To examine and expand on previous findings on the direct link of e-cigarette use to later CTC and marijuana use, we hypothesize that adolescent e-cigarette use at T1 (H7) and adolescent CTC use at T2 (H8) will be directly and positively associated with marijuana use at T3.

Methods

Sampling

Data were collected for a longitudinal study examining the influence of point-of-sale advertising on the use of nicotine and tobacco products by alternative high school students in southern California.²⁸ Information from the California Department of Education was used to identify 183 eligible alternative high schools. Schools were considered eligible if they had least 100 students and were within 100 miles of the program offices in Claremont, CA. After obtaining approval from the Claremont Graduate University Institutional Review Board, all 183 schools were contacted and invited to enroll in the 3-year longitudinal panel study. Of the 183 schools contacted, 33 agreed to participate; however, four schools were dropped from the study due to low participation during the consent process, leaving a final sample of 29 sites (16%) successfully participating.

Respondents and Procedure

Between October 14, 2014 and May 18, 2015, research staff distributed interest forms to 6870 high school students who attended one of the 29 schools; 2726 students returned forms. Students returning forms were assigned a staff member who obtained written consent and provided a link to a web-based survey. Parental consent and youth assent were obtained for students under the age of 18. Students and their parent(s) or guardian(s) were told the study concerned health behaviors and that respondents would be asked about underage tobacco and drug use. All students were given until September 1, 2015 to complete the approximately 90-minute survey. A total of 1060 students took part in the initial (T1) assessment. Each respondent received a $45 gift card to compensate them for their time.

Respondents were tracked using established procedures.²⁸ One-year follow-up assessments (T2) were completed by 87.1% of the original sample (n = 923). Most of these respondents (96.6%) completed a web-based survey. Those without access to a web-enabled device (3.4%) completed a computer-assisted telephone interview. Among the 137 respondents who did not complete a follow-up assessment, 93.5% failed to respond to repeat contact attempts, 5.8% withdrew from the study, and 0.7% were incarcerated. Respondents received a $50 gift card for completing the 1-year follow-up.

Two-year follow-up assessments (T3) were completed by 81.0% of the original sample (n = 859). Most respondents (96.9%) completed a web-based survey, and some (3.1%) completed a computer-assisted telephone interview. Among the 201 respondents not completing an assessment, 93.0% failed to respond to repeated contact attempts, 4.5% withdrew from the study, 1.5% passed away, 0.5% were incarcerated, and 0.5% were deployed overseas after enrolling in the military. Respondents received a $100 gift card for completing the 2-year follow-up.

Measures

Demographics

Respondents reported age, sex, and ethnicity at T1.

Family E-Cigarette Use

Family e-cigarette use (T1) was assessed with one item: “Does anyone who lives with you now use electronic cigarettes, vaporizers, or vape pens?” Options were coded 0 (no) or 1 (yes). The measure was adapted from the Centers for Disease Control and Prevention’s National Youth Tobacco Survey (NYTS).²⁹

Peer E-Cigarette Use

Peer e-cigarette use (T1) was examined by asking, “How many of your four closest friends use electronic cigarettes, vaporizers, or vape pens?” Responses ranged from 0 (none) to 4 (four). The measure was adapted from the California Department of Public Health (CDPH) 2011–2012 California student tobacco survey (CSTS).³⁰

Positive Expectancies

Adolescents’ positive expectancies (T1; α = .97) about smoking CTCs were assessed using 12 items from the short form smoking consequences questionnaire (S-SCQ).³¹ Responses were assessed with 5-point Likert-type items whose options ranged from 1 (Extremely unlikely) to 5 (Extremely likely). Example items included, “Cigarettes taste good,” and “Cigarettes help reduce anxiety.” A mean score was computed—higher scores indicate more positive expectations.³¹

Willingness

Willingness to use CTCs (T1) was assessed by presenting respondents with the following item, “Suppose you were with a group of friends at a party and there were tobacco products there that you could try if you wanted. If it were offered by one of your friends, how willing would you be to try cigarettes?” Responses were based on a 4-point Likert-type item that ranged from 1 (Definitely No) to 4 (Definitely Yes).³²

Adolescent Substance Use

A validated drug use questionnaire³³ was adapted to query students on their use of CTCs, e-cigarettes, vaporizers, or vape pens, and marijuana in the past 30 days. Original response options ranged from 0 (0 times) to 10 (91 or more times) but were truncated to achieve a normal distribution. The final response scales ranged from 0 (0 times) to 3 (21 or more times). While the measure has not been validated for novel nicotine products such as vape pens, the measure has been used in prior studies comparing frequencies in at-risk groups reporting use of many addictive substances including CTCs and methamphetamine.³⁴

Eligibility Criteria and Data Preparation

This prevention investigation focused on adolescent substance use; thus, students between the ages of 14 and 18 during the first assessment (T1) were considered eligible for participation. Thirty-five respondents falling beyond the age range were removed from the sample (resultant N = 1025). Distribution of the data was examined using measures of skew and kurtosis (ie, skew < 3, kurtosis < 10).³⁵ Analysis revealed all variables within boundaries usually considered normal or acceptable. Multicollinearity was not an issue—the highest variance inflation factor was 2.33, which fell below the cutoff of 2.5.³⁶

Analytic Plan

To examine potential clustering effects of school environment, an intraclass correlation was calculated in SPSS version 24, with marijuana use (T3) specified as the outcome variable. Findings revealed a minimal association between school and marijuana use (Intra-Class Correlations = .02), supporting inclusion of individuals as the unit of analysis.³⁷

R statistical software was used for the remaining analyses. Several variables had missing data, common in longitudinal designs. Patterns of missingness are shown in Table 1. To correct for missing data bias, multiple imputation using 100 imputations was implemented under the conditions of missing at random (MAR) using R’s Mice package. A path model was tested using R’s (version 3.3.1) lavaan package with 100 imputed datasets and maximum-likelihood estimation. Age, sex, and ethnicity were added as covariates in the model; that is, paths from initial (T1) substance use (CTC, e-cigarette, and marijuana) to each outcome variable (positive expectancies and willingness T1, CTC use T2, and marijuana use T3) were estimated to control for initial level of use.³⁸ Sex and ethnicity were dummy coded prior to model entry. The reference group for sex was female; non-Hispanic was the reference group for ethnicity. Prior use of e-cigarettes, CTCs, and marijuana at T1 also were entered to control for initial usage levels. Model fit was evaluated with three robust fit indices—the comparative fit index (CFI), Tucker–Lewis fit index (TFI), and root mean square error of approximation (RMSEA). For the CFI and TFI, scores close to 1.00 indicate good fit. The RMSEA indicates good fit if values are below .05.³⁸

Table 1.

Same Characteristics and Statistics Measure Summary Information (N = 1025)

Age	Range	14–18
	M (SD)/missing	17.40 (0.82)/n = 0
Sex	Male/female/missing	n = 522/ n = 499/n = 4
Ethnicity	Non-Hispanic/Hispanic/missing	n = 250/ n = 752/n = 23
Family e-cigarette use (T1)	No/yes/missing	n = 802/ n = 147/n = 76
Peer e-cigarette use (T1)	M (SD)/missing	1.25 (1.45)/n = 228
Positive expectancies (T1)	M (SD)/missing	1.92 (1.11)/n = 43
Willingness (T1)	M (SD)/missing	1.63 (1.01)/n = 32
Adolescent e-cigarette use (T1; past 30 days)	0 times	n = 780
	1–10 times	n = 103
	11–20 times	n = 30
	21 or more times	n = 59
	Missing	n = 53
Adolescent cigarette use (T1; past 30 days)	0 times	n = 818
	1–10 times	n = 78
	11–20 times	n = 27
	21 or more times	n = 52
	Missing	n = 50
Adolescent marijuana use (T1; past 30 days)	0 times	n = 598
	1–10 times	n = 133
	11–20 times	n = 44
	21 or more times	n = 171
	Missing	n = 79
Adolescent cigarette use (T2; past 30 days)	0 times	n = 780
	1–10 times	n = 103
	11–20 times	n = 30
	21 or more times	n = 59
	Missing	n = 53
Adolescent marijuana use (T3; past 30 days)	0 times	n = 530
	1–10 times	n = 69
	11–20 times	n = 29
	21 or more times	n = 127
	Missing	n = 270

In estimating the hypothesized model, paths from family e-cigarette use, peer e-cigarette use, and adolescent CTC use (T1) to adolescent e-cigarette use (T1), and from adolescent e-cigarette use (T1) to positive expectancies and willingness (T1) were specified. Paths from positive expectancies and willingness (T1) with CTC use (T2) also were estimated. Further, a direct path from CTC use (T2) to marijuana use (T3) was assessed. Based on the modification indices, one additional direct path was added from adolescent e-cigarette use (T1) to marijuana use (T3). Based on the findings of the modification indices, covariance between positive expectancies and willingness was estimated. Indirect effects of family and peer e-cigarette use (T1) on marijuana use (T3) via adolescent e-cigarette use (T1), willingness (T1), and CTC use (T2) also were estimated using the Delta (Sobel) method in the R lavaan package.^39–41

Results

For descriptive information on all key study variables, see Table 1. The model chi-square was not statistically significant, χ  ² (9) = 10.14, p = .34, indicating good fit of model with data. The other goodness of fit indices demonstrated good model fit (CFI = .99, TLI = .99, RMSEA = .01 [90% confidence interval: .00– .04]).

Direct and Indirect Effects

The final model relationships presented in Figure 1 show statistically significant direct paths from peer and family e-cigarette use (T1) to adolescent e-cigarette use (T1). Adolescent CTC use (T1) also was linked directly to adolescent e-cigarette use, positive expectancies, willingness (all T1), and CTC use at T2. Adolescent CTC use at T2 was directly linked directly to marijuana use at T3, as were adolescent e-cigarette and marijuana use (both T1).

Figure 1.

Path analytic model of proposed study relationships. Age, sex, and ethnicity (all T1) were entered into the model as covariates but are not pictured to maintain conceptual clarity. T1 substance use also was controlled for. T1 substance use variables (e-cig, CTC, and marijuana) also were entered in the model. Paths between marijuana use (T1) with family e-cigarette use (T1), positive expectancies (T1), and adolescent cigarette use at (T2), and paths between e-cigarette use (T1) and cigarette use (T2) were included in the estimated model, but were not statistically significant, and are not pictured. Covariance between positive expectancies and willingness was estimated but is not pictured to provide diagrammatic simplicity. T = time. *p < .05, **p < .01, ***p < .001.

Furthermore, adolescent e-cigarette use was directly related to positive expectancies about CTC use and greater willingness to try CTCs (both T1), and CTC use at T2. Teens scoring higher in positive expectancies and willingness at T1 also were more likely to use CTC at T2.

E-cigarette use and marijuana use (both at T1) and CTC use at T2 were direct predictors of marijuana use at T3.

Indirect effects were estimated using the guidelines for analysis recommended in the lavaan package for R software (Delta/Sobel method).^39–41 No direct effect between family or peer e- cigarette use (T1) with marijuana use (T3) was proposed. Instead, we predicted these relationships would operate via the mediators (adolescent e-cigarette use, positive expectancies, and willingness [all T1] and CTC use [T2]). The indirect effects of both family e-cigarette use (T1; B = .17, p < .001) and peer e-cigarette use (T1; B = .14, p < .001) on marijuana usage (T3), as mediated by adolescent e-cigarette use (T1), willingness (T1), and CTC use (T2), were statistically significant. The same model was estimated treating marijuana usage at T3 as a binary outcome variable (0 = no; 1 = yes) using probit regression estimation. Findings from this alternative model revealed a similar pattern of results and significance levels

Discussion

Accumulating evidence suggests the effects of e-cigarette use for adults and adolescents.^41,42 The current research indicates that e-cigarette use and marijuana use may be linked, and as such, this association deserves close scrutiny. Analysis of a unique 3-year longitudinal sample of alternative high school students in California disclosed an unsettling pattern of results considering prior research on adolescents’ use of psychoactive substances. Based on a longitudinal panel survey involving 859 respondents, the results of this analysis support the plausibility of a well-established developmental sequence in which e-cigarette initiation is furthered by family and peer use, which may be associated with more positive expectations and openness to CTCs. These positive responses could give rise to subsequent CTC use, which prior research has related to marijuana initiation, which was demonstrated in the current research. We realize many factors have been established as strongly implicated in adolescents’ decisions to use marijuana, including peer use and positive predispositions toward the substance.^21–24 The current study indicates that e-cigarette use also may be considered to have an influence on marijuana use, a substance clearly not indicated for use by adolescents, and thus its potential effects, too, should be pointed out and monitored carefully.^43,44

Earlier studies across a host of psychotropic substances have implicated peer and family use as strong antecedents to adolescents’ own usage.^23,45,46 Our analysis demonstrated the association between peer and family use and adolescents’ own e-cigarette use at the first measurement session (T1). Adolescents’ e-cigarette use at T1 was associated with positive expectancies toward CTCs and willingness to use them, and this was realized behaviorally in the second year of the study. That these social psychological constructs anticipated CTC use at T2 could be predicted by Fishbein and Ajzen’s theory of reasoned action, which is popular in many studies bearing on public health.⁴⁷ In turn, CTC use at T2 significantly predicted marijuana use at T3. It also is noteworthy that adolescent e-cigarette use at T1 was directly and significantly associated with marijuana use 2 years later at T3.

The analysis showed that e-cigarette use was associated with positive expectancies regarding CTC use, along with a willingness to try them. It supported the proposed hypotheses with strong associations, raising concerns about the potential short- and long-term effects of e-cigarettes for adolescents, whose use is growing at a rapid rate despite increasing evidence of the negative health implications raised in response to their use.^9,12,13 For vulnerable populations like the alternative high school students examined in this study, the pressures exerted by peers and even some parents may lead to CTC use by increasing positive expectations and to a greater willingness to experiment with e-cigarettes, which we have shown to have statistically significant associations with subsequent marijuana use, even after accounting for earlier marijuana use.

Limitations

The reported results should be considered in light of their limitations. Although the sample was large, a strength, it consisted of students attending alternative high schools in California. Such high school students are not representative of the general population of students in the state, and thus, the results may be viewed as restricted to a nonrepresentative group whose generalization to the population at large is uncertain. However, the sample was of the same age as most high school students in the state. Further, research suggests alternative students are at greater risk of substance use than students attending more usual (ie, nonalternative) high schools; however, the results do not suggest that the developmental pattern of psychoactive substance use implicated in this study is materially different between these two groups. The developmental patterns of our study are not unusual, and the bulk of prior literature is based on responses of students attending more standard high schools, not alternative schools.⁴⁷

Another possible limitation to generalizability is the low volunteer rate to the initial call for participants. Reasons for students/parents’ refusing to volunteer cannot be known, insofar as refusals occurred before any other data were collected. A reasonable hypothesis is that heavy substance users might have been reluctant to participate in a study in which they would be asked to report on their usage, and a large refusal rate might have rendered a school ineligible to participate. However, because analyses controlled for initial usage at T1, along with age, sex, and ethnicity, the nonvolunteer rate, although not desirable, was not seen as overly problematic.

A third objection might be that the research was based on nonexperimental methods, and so inferences of causation are not supported unambiguously. However, path analytic techniques making use of longitudinal panel data are generally accepted as the best examples of this analytic form. Even so, though we do not make strong causal claims, we do suggest the current pattern of findings fits well with prior research, supporting the plausibility of the possibilities raised herein.

Conclusion

The country is faced with some obvious facts relevant to e-cigarette use by adolescents. The first of these is that the growth of e-cigarette use in this population is nothing short of phenomenal.⁴⁸ The second is that CTC use is dangerous, especially for youth, whose developmental trajectories render them particularly prone to nicotine addiction. A third fact is that CTC use is a major threat to public health, a source of death and disability.⁴⁸ The Surgeon General’s report of 2019 stated the danger succinctly: “Every day, more than 1,200 people in this country die due to smoking. For each of those deaths, at least two youth or young adults become regular smokers each day. Almost 90% of those replacement smokers smoke their first cigarette by age 18.” The fourth fact is that some of our best scientists have concluded that marijuana, if started early and used intensively, arrests the brain development of young people.⁵

Now let us leave the land of facts for the land of conjecture. Suppose the developmental findings reported in this analysis are the result of the predicted causal sequence. The patterns suggested in the path analysis are far from controversial. Parent and peer influences on adolescent substance use are well established. CTC use has been established as dangerous to health and has been linked to marijuana uptake in many studies. The new ingredient added to this standard recipe, as both a direct and indirect factor, is the electronic nicotine delivery system (ENDS). It fits well with all the established patterns and suggests that we strongly consider ENDS as an ominous addition to the long list of threats to healthy development of our youth. If this warning is taken seriously by policy makers, parents, and the scientific community, this threat should be minimized or eliminated.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Funding

Research reported in this publication was supported by the National Institute of Child Health and Human Development and the Food and Drug Administration Center for Tobacco Products (R01HD077560). The content is solely the responsibility of the authors and does not necessarily represent the views of the National Institutes of Health or the Food and Drug Administration.

Declaration of Interests

None declared.