Abstract
Objective:
Understanding transitions in nicotine and cannabis use has implications for prevention and efforts to reduce harmful use. Focusing on cross-substance associations, we examined how use of one substance was associated with year-to-year transitions in frequency of use of the other among young adults in the context of legalized non-medical cannabis.
Method:
A statewide sample from Washington (N = 4,039; ages 18–25 at baseline) provided up to 3 years of annual data on past-month cannabis use and nicotine use (tobacco cigarettes and e-cigarettes/vaping). Manifest Markov models examined how use of each substance was associated with transitions in the other across categories of past-month no use, occasional use (1–19 days), and frequent use (≥20 days).
Results:
Occasional and frequent nicotine use (vs. no use) predicted higher probability of transitioning from no cannabis use to occasional or frequent cannabis use and from occasional use to frequent use, whereas associations with cessation and de-escalation were inconsistent in direction, small in magnitude, and not statistically significant. Cannabis use positively predicted onset of nicotine use, and associations of cannabis use with escalation from occasional to frequent nicotine use, de-escalation in use, and cessation in use were small and inconsistent in direction.
Conclusions:
The findings corroborate prior research on cannabis and nicotine use as risk factors to address in prevention efforts. The findings do not provide strong support for prioritization of dual abstinence in efforts to encourage reductions in or cessation of cannabis or nicotine use among young adults.
Cannabis use and nicotine use are positively associated with one another (Chadi et al., 2019; Rabin & George, 2015; Strong et al., 2018; Tucker et al., 2019), and studies have found evidence of cross-substance longitudinal associations wherein use of one substance is linked to subsequent changes in use of the other (for review, see Agrawal et al., 2012; Lemyre et al., 2019). Identifying cross-substance associations related to onset or escalation may be particularly informative for prevention, helping identify individuals at heightened risk of increases in use. It may also be that use of one substance reduces the probability of de-escalation or cessation in use of the other (Agrawal et al., 2012; Lemyre et al., 2019), holding implications for treatment and efforts aimed to reduce or extinguish use of either substance (Agrawal et al., 2012; Lemyre et al., 2019; Rabin & George, 2015).
Studies using retrospective data on the sequence of initiation and escalation of tobacco and cannabis use have found evidence of both gateway and reverse gateway effects (e.g., Mayet et al., 2011). Also, some cross-sectional studies have found negative associations between use of one substance and quit attempts for the other (Strong et al., 2018; Weinberger et al., 2020b). In studies of annual longitudinal data on adults age 18 years and older in large U.S. nationwide samples, Weinberger et al. (2018, 2020a) found that cannabis use in one wave of data predicted increased risk of initiation, escalation, and relapse in cigarette smoking by the next wave and decreased probability of cessation of smoking tobacco. Becker et al. (2015) reported similar findings in a sample of male Swiss young adults, with cannabis use predicting progression to daily cigarette smoking 18 months later. Examining longitudinal associations in the other direction and using data on a nationwide sample of U.S. adults, Weinberger et al. (2021) found that cigarette use was positively associated with subsequent onset of cannabis use disorder (CUD) among those who did not use cannabis, but not escalation to CUD among those already using cannabis. Further, cigarette use was positively associated with persistence of CUD.
Some research on cross-substance effects on cessation has focused on treatment settings, where those in treatment often view cannabis and nicotine as substitutes for one another (McClure et al., 2019; Reboussin et al., 2021). Evidence from this research is mixed (Agrawal et al., 2012; Lemyre et al., 2019). For instance, Ford et al. (2002) and McClure et al. (2020) found that cannabis use was associated with lower likelihood of smoking cessation, whereas Britton et al. (2020) and Rabin et al. (2016) found little effect of cannabis use on tobacco smoking treatment outcomes.
The current study aimed to assess relationships between cannabis and nicotine use in terms of how use of one was associated with year-to-year transitions in use of the other. We were guided by the hypotheses that use of one substance would increase the probability of onset and escalation and decrease the probability of reductions in or cessation of use of the other. Similar to the studies by Weinberger et al. (2018, 2020a), we examined how cannabis use predicted transitions in nicotine use, although data for our study came from young adults in a state where nonmedical cannabis was legal for adults ages 21 and older and, instead of focusing solely on tobacco in the form of cigarettes, we examined a measure of nicotine use that included e-cigarettes, which have become widely used in the past decade (Patrick et al., 2022). Also, we examined how nicotine use predicted transitions in cannabis use, although we assessed transitions in frequency of use rather than transitions across the threshold of CUD. We used models that provide separate estimates for different transitions, including from non-use to use, escalation in use, de-escalation in use, and cessation of use.
Method
Study population and design
Data came from the Young Adult Health Survey (Kilmer et al., 2022), which used a cohort sequential design and online surveys. Cohorts of approximately 2,000 young adults in Washington State ages 18–25 enrolled each year starting in 2014, and then were followed annually. Of note, Washington voters approved nonmedical cannabis legalization in 2012, and the first nonmedical cannabis retail outlets opened in mid-2014. Two recruitment strategies were used to capture a wide range of the state's young adult population. First, the Washington Department of Licensing (DOL) provided address information from which a sample of 18- to 25-year-olds could be invited to participate. Second, to supplement the sample with residents who did not necessarily have a driver's license and stable housing, we recruited individuals from websites such as Facebook, Instagram, and Craigslist. Individuals recruited online completed an eligibility screening survey before being enrolled in the project. The current study used data from 2015–2017 cohorts. Source of recruitment was not tracked during 2015, but 52% and 58% of the 2016 and 2017 cohorts, respectively, responded to letters sent from the DOL list.
The web-based surveys contained questions on substance use, related risk factors, and health behaviors. Participants spent approximately 20 minutes completing surveys and then received $10 e-gift cards (Kilmer et al., 2022). We used three waves of longitudinal data for each cohort. Thus, data span 2015–2017 for the first cohort, 2016–2018 for the second, and 2017–2019 for the third. Data collection periods in 2015–2018 ran from June to November, and the field period in 2019 was from July to December. The University of Washington Institutional Review Board approved all measures and procedures.
Of 6,509 individuals who enrolled between 2015 and 2017 and answered survey items concerning cannabis, cigarette, and e-cigarette use, 4,039 (62%) completed Wave 2 surveys and comprise the analytic sample. Of these individuals, 3,128 (77%) completed Wave 3 surveys. Descriptive information on sex assigned at birth, race/ethnicity, cohorts, and age at Wave 1 are shown in Table 1. Notably, a majority (75%) of the analytic sample was female (i.e., assigned at birth). The “non-Hispanic/Latinx (NH) other” category shown in Table 1 included NH Black (2% of the total sample), NH Native American (1%), NH Pacific Islander or Native Hawaiian (1%), and NH multiracial (7%). The racial/ethnic composition of the sample was similar to that of the Washington State young adult population, although the 2020 Census (U.S. Census Bureau, 2020) indicated that a smaller proportion of the state population were Asian (10%) and higher proportions were Hispanic/Latinx (14%) and Black (4%). At the first survey, 53% of the sample were in school (33% 4-year college, 11% 2-year college, 7% graduate or professional school, 1% high school, and 1% vocational school), and 37% were employed full-time. Most were still living in Washington State at their Wave 2 and 3 surveys (97% and 93%, respectively). Attrition was associated with some demographic variables (e.g., females were less likely than males to drop out). Attrition analyses are shown in supplemental materials (Table A). (Supplemental material appears as an online-only addendum to this article on the journal's website.) Missing data were addressed using robust maximum likelihood estimation.
Table 1.
Frequencies for cannabis and nicotine use at Waves 1–3 and descriptives for background covariates
| Variable | Wave 1 (N = 4,039) % | Wave 2 (N = 4,039) % | Wave 3 (n = 3,058/3,125a) % | Fixed characteristics (N = 4,039) % |
|---|---|---|---|---|
| Substance use | ||||
| Cannabis use | ||||
| No use | 72.8 | 69.9 | 68.8 | |
| Occasional use (1–19 days) | 19.3 | 20.3 | 21.2 | |
| Frequent use (≥20 days) | 7.9 | 9.9 | 10.0 | |
| Nicotine use | ||||
| No use | 87.2 | 87.3 | 89.3 | |
| Occasional use (1–19 days) | 8.1 | 7.7 | 5.9 | |
| Frequent use (≥20 days) | 4.6 | 5.1 | 4.8 | |
| Background covariates | ||||
| Female (assigned at birth) | 74.7 | |||
| Race/ethnicity | ||||
| NH White | 65.3 | |||
| NH Asian | 13.0 | |||
| Hispanic/Latinx | 11.3 | |||
| NH other | 10.3 | |||
| Cohort | ||||
| 2015 | 27.2 | |||
| 2016 | 39.0 | |||
| 2017 | 33.9 | |||
| M (SD) | ||||
| Age at Wave 1, in years | 21.74 (2.23) |
Notes: NH = non-Hispanic/Latinx.
Of the 3,128 individuals who completed the Wave 3 survey, 71 were missing data on nicotine use and 3 were missing data on cannabis use.
Measures
Substance use. Annual surveys included questions about past-month days of use of cannabis, cigarettes, and e-cigarettes. The survey did not have questions about use of tobacco in forms other than cigarettes. We coded past-month cannabis use (which included smoking, consuming edibles, vaping, and dabbing) into three categories: no use, occasional use (1–19 days in the past month), and frequent use (≥20 days). The ≥20 days cutoff is similar to the threshold for “daily or near-daily” use in Monitoring the Future reports (Patrick et al., 2022) and is the threshold for frequent use recommended in the International Cannabis Toolkit (Lorenzetti et al., 2022). We created a parallel measure of nicotine use based on whether respondents reported (a) no use of either cigarettes or e-cigarettes, (b) using either cigarettes or e-cigarettes, but neither for 20 or more days, and (c) using either cigarettes or e-cigarettes for 20 or more days. The e-cigarette item referred to “an electronic cigarette or e-cig to vaporize/vape nicotine.” Sensitivity analyses examined dichotomous measures of any cigarette and any e-cigarette use separately and a four-category nicotine measure (neither, only cigarettes, only e-cigarettes, or both).
Covariates. We included covariates we had found, in prior analyses of baseline cross-sectional data from this project (Fleming et al., 2022; Gilson et al., 2023; Kilmer et al., 2022), to be associated with substance use and substance use risk factors, and thus might be potential confounders. The covariates were Wave 1 age, sex assigned at birth (0 = male, 1 = female), cohort (coded 1–3), and race/ethnicity (indicator coding of four categories: NH White [reference], NH Asian, Hispanic/Latinx, NH other).
Analyses
We used manifest Markov chain models (Kaplan, 2008; Lanza et al., 2013) to examine transitions between observed categories of substance use across three waves of data collection. The models assessed how (a) nicotine use (indicator coded with no use as the reference) was associated with transitions between three categories of cannabis use (Figure 1) and (b) cannabis use (indicator coded with no use as the reference) was associated with transitions between three categories of nicotine use. The underlying statistical models involve multinomial logistic regression of categories of substance use on covariates. When predicting category of use at Waves 2 and 3, these covariates included category of use of that substance at the prior wave as well as the other covariates, with the effects of category of use of the other substance being of primary interest. From these models, model-expected probability of transitioning from one category to another (i.e., the probability of being in a given class at time t conditional on being in a given class at t − 1) conditional on covariates were derived.
Figure 1.
General form of Markov model for transitions in cannabis use, with covariate effects of nicotine use. The underlying statistical model involves multinomial logistic regressions with three categories of cannabis use at each time point regressed on covariates.
The general form of these models is shown in Figure 1 for the model predicting transitions in cannabis use. In addition to the stability/autoregressive effects of use of the same substance at the prior time point (paths a1 and a2 in Figure 1), the concurrent association of the two substances at Wave 1 (path b), and cross-substance main effects across intervals (paths c1 and c2), the models included moderation paths (d1 and d2) that allow for the cross-substance longitudinal paths to differ by category of use at the prior time point. These moderation paths allow for differences in the effects of covariates on transitioning out of one category compared with exiting out of another, and thereby allowed for different effects on different types of transitions. As an omnibus test of moderation, we compared fit of models with and without moderation paths specified using a likelihood ratio test (LRT), based on change in model log likelihood (LL), and comparison of values for the Bayesian Information Criterion (BIC) and Akaike's Information Criterion (AIC). Age, sex, and cohort were specified to have main effects on the outcomes at each time point. We assumed homogeneity of stationary chains wherein transition odds were equal across the two intervals and both main and moderation effects of covariates were equal across intervals (i.e., from Wave 1 to 2 and Wave 2 to 3). We conducted model fit tests (a) relaxing the stationarity constraints assumption and (b) adding moderation of effects of the three background covariates on Wave 2 and 3 substance use outcomes by prior wave category of use of the given outcome.
Robust maximum likelihood estimation allowed for inclusion of cases with partially missing data (i.e., individuals missing data at Wave 3), under the assumption that data were missing at random after taking nonmissing values on other model variables into account (Graham, 2012). We used Mplus 8.8 (Muthén & Muthén, 2017) to run all analyses and a statistical significance criterion of p < .05 to help organize the presentation of the findings.
Results
Descriptive information
Table 1 shows frequencies for categories of cannabis and nicotine use at each wave and descriptive information on background covariates. For both substances at all waves, a majority of the sample did not report use, ranging from 69% to 73% for cannabis and 87% to 89% for nicotine. Concurrent associations between cannabis and nicotine use were positive and strong (Supplement Table B). For instance, at Wave 1, 21% of those reporting no nicotine use indicated occasional or frequent cannabis use compared with 70% of those who did use nicotine. Generally, stability was more common than transitioning. Depending on category of use at the prior time point, 54%–87% of individuals stayed in the same category of cannabis use across adjacent intervals and 47%–97% stayed in the same category of nicotine use (Supplement Tables C and D).
Markov model estimates
LRTs comparing fit of models with and without moderation effects (d paths in Figure 1) indicated significantly better fit with moderation (model predicting cannabis use: ΔχLR2 = 40.54, Δdf = 8, p < .001, ΔBIC = 23, ΔAIC = -24; model predicting nicotine use: ΔχLR2 = 71.07, Δdf = 8, p < .001, ΔBIC = -5, ΔAIC = -55; Supplement Tables E and F). Models with these moderation effects are presented below. Tests of adding additional parameters by relaxing the stationarity assumption or adding moderation paths for background covariates supported more parsimonious models. Although changes in fit due to adding these parameters to the model predicting nicotine use were statistically significant according to LRTs, changes in fit for cannabis use models were not, and changes in AIC and BIC favored more parsimonious models in all cases (Supplement Tables E and F).
Tables 2 and 3 present model estimates of covariate effects on the log odds of occasional and frequent use relative to no use for the given substance. In both models, estimates of positive within-substance effects across waves (a paths in Figure 1) reflect within-person stability in use categories. There were concurrent positive main effects at Wave 1 of one type of substance use on the other (b paths), reflecting the strong concurrent association noted above. There were also positive main effects (c paths), such that, among those in the no-use category at the prior wave, being in the occasional- or frequent-use categories at Waves 2 and 3 were positively associated with prior wave occasional and frequent use of the other substance. All moderation effects (d paths) in both models were negative, indicating that the associations of one type of substance use with being in occasional or frequent use categories of the other type of substance use were lower for those reporting occasional or frequent use (vs. no use) of that substance at the prior wave. For the model predicting cannabis use, five out of eight moderation effects were statistically significant; for the model predicting nicotine use, six out of eight were statistically significant.
Table 2.
Covariate effects on the log odds of being in occasional and frequent categories of cannabis use compared with no use
| Variable | Occasional (vs. no use) | Frequent (vs. no use) | ||||
|---|---|---|---|---|---|---|
| B | SE | p | B | SE | p | |
| Wave 1 cannabis use on | ||||||
| Wave 1 nicotine use (ref. = no use) | ||||||
| Occasional | 2.07 | 0.14 | <.001 | 2.36 | 0.18 | <.001 |
| Frequent | 1.52 | 0.20 | <.001 | 2.96 | 0.20 | <.001 |
| Cohort | 0.07 | 0.06 | .241 | 0.06 | 0.08 | .485 |
| Female (assigned at birth) | -0.04 | 0.10 | .707 | -0.13 | 0.14 | .343 |
| Age | 0.04 | 0.02 | .050 | 0.07 | 0.03 | .011 |
| Race/ethnicity (ref. = NH White) | ||||||
| NH Asian | -0.63 | 0.15 | <.001 | -1.21 | 0.28 | <.001 |
| Hispanic/Latinx | -0.13 | 0.14 | .343 | -0.23 | 0.21 | .281 |
| NH other | 0.17 | 0.13 | .213 | -0.02 | 0.20 | .913 |
| Wave 2 and 3 cannabis use ona | ||||||
| Prior wave nicotine use (ref. = no use) | ||||||
| Occasional | 1.23 | 0.17 | <.001 | 1.35 | 0.44 | .002 |
| Frequent | 1.08 | 0.24 | <.001 | 2.24 | 0.39 | <.001 |
| Prior wave cannabis use (ref. = no use) | ||||||
| Occasional use | 2.58 | 0.09 | <.001 | 3.32 | 0.17 | <.001 |
| Frequent use | 2.99 | 0.24 | <.001 | 6.71 | 0.27 | <.001 |
| Prior wave occasional nicotine use by | ||||||
| Prior wave occasional cannabis use | -0.98 | 0.25 | <.001 | -0.75 | 0.49 | .127 |
| Prior wave frequent cannabis use | -0.98 | 0.36 | .007 | -1.35 | 0.52 | .010 |
| Prior wave frequent nicotine use by | ||||||
| Prior wave occasional cannabis use | -0.85 | 0.49 | .081 | -1.50 | 0.60 | .012 |
| Prior wave frequent cannabis use | -0.52 | 0.56 | .358 | -1.82 | 0.65 | .005 |
| Cohort | -0.05 | 0.04 | .243 | 0.05 | 0.07 | .485 |
| Female (assigned at birth) | -0.21 | 0.08 | .006 | 0.08 | 0.12 | .500 |
| Age | -0.06 | 0.02 | <.001 | -0.07 | 0.03 | .007 |
| Race/ethnicity (ref. = NH White) | ||||||
| NH Asian | -0.03 | 0.10 | .753 | -0.58 | 0.20 | .003 |
| Hispanic/Latinx | -0.15 | 0.11 | .173 | 0.03 | 0.18 | .888 |
| NH other | 0.14 | 0.11 | .209 | 0.25 | 0.17 | .147 |
Notes: Ref. = reference group; NH = non-Hispanic/Latinx.
Homogeneity constraints fix the covariate effects on Wave 2 and 3 substance use to be equal.
Table 3.
Covariate effects on the log odds of being in occasional and frequent categories of nicotine use compared with no use
| Variable | Occasional (vs. no use) | Frequent (vs. no use) | ||||
|---|---|---|---|---|---|---|
| B | SE | p | B | SE | p | |
| Wave 1 nicotine use on | ||||||
| Wave 1 cannabis use (ref. = no use) | ||||||
| Occasional | 2.07 | 0.14 | <.001 | 1.52 | 0.20 | <.001 |
| Frequent | 2.36 | 0.18 | <.001 | 2.96 | 0.20 | <.001 |
| Cohort | -0.24 | 0.08 | .002 | -0.08 | 0.10 | .426 |
| Female (assigned at birth) | -0.31 | 0.13 | .019 | -0.45 | 0.17 | .008 |
| Age | -0.06 | 0.03 | .043 | 0.04 | 0.04 | .307 |
| Race/ethnicity (ref. = NH White) | ||||||
| NH Asian | -0.47 | 0.23 | .037 | -0.65 | 0.34 | .059 |
| Hispanic/Latinx | -0.22 | 0.21 | .295 | -0.08 | 0.27 | .757 |
| NH other | -0.32 | 0.21 | .124 | 0.18 | 0.23 | .432 |
| Wave 2 and 3 nicotine use ona | ||||||
| Prior wave cannabis use (ref. = no use) | ||||||
| Occasional | 1.35 | 0.17 | <.001 | 1.46 | 0.41 | <.001 |
| Frequent | 1.64 | 0.23 | <.001 | 3.14 | 0.37 | <.001 |
| Prior wave nicotine use (ref. = no use) | ||||||
| Occasional use | 4.11 | 0.21 | <.001 | 4.80 | 0.38 | <.001 |
| Frequent use | 4.64 | 0.40 | <.001 | 7.93 | 0.48 | <.001 |
| Prior wave occasional cannabis use by | ||||||
| Prior wave occasional nicotine use | -1.32 | 0.28 | <.001 | -1.72 | 0.53 | .001 |
| Prior wave frequent nicotine use | -1.68 | 0.36 | <.001 | -3.03 | 0.53 | <.001 |
| Prior wave frequent cannabis use by | ||||||
| Prior wave occasional nicotine use | -1.25 | 0.65 | .055 | -0.90 | 0.71 | .207 |
| Prior wave frequent nicotine use | -1.81 | 0.58 | .002 | -3.08 | 0.65 | <.001 |
| Cohort | -0.15 | 0.07 | .032 | -0.08 | 0.10 | .408 |
| Female (assigned at birth) | -0.33 | 0.12 | .005 | -0.30 | 0.17 | .082 |
| Age | -0.09 | 0.03 | .001 | -0.10 | 0.04 | .007 |
| Race/ethnicity (ref. = NH White) | ||||||
| NH Asian | 0.09 | 0.18 | .596 | -0.02 | 0.32 | .954 |
| Hispanic/Latinx | 0.08 | 0.17 | .625 | -0.16 | 0.28 | .580 |
| NH other | -0.02 | 0.19 | .931 | 0.20 | 0.25 | .412 |
Notes: Ref. = reference group; NH = non-Hispanic/Latinx.
Homogeneity constraints fixed the covariate effects on Wave 2 and 3 substance use to be equal.
Table 4 translates these results into odds ratios (ORs) for associations of one type of substance use with odds of transitions into new use categories of the other substance (i.e., the effect of being in a given category conditional on category at the prior time point), and Figures 2 and 3 convert these estimated effects into transition probabilities. Supplement Tables G–J summarize results of models (a) stratifying by whether individuals were at least age 21 at enrollment and (b) using past-month substance use frequency categories of 0, 1–29, and 30 days.
Table 4.
Odds ratios (OR) for effects of nicotine use at prior wave on the odds of transitions from cannabis use category in prior wave (rows) into cannabis use categories in the subsequent wave (columns) and effects of cannabis use on nicotine transitions
| Variable | Cannabis or nicotine use category | ||
|---|---|---|---|
| No use OR [95% CI] | Occasional OR [95% CI] | Frequent OR [95% CI] | |
| Prior wave cannabis use | |||
| Occasional nicotine use (vs. no use) | |||
| No use | 1.00 | 3.42 [2.46, 4.81] | 3.84 [1.63, 9.09] |
| Occasional | 0.78 [0.56, 1.07] | 1.00 | 1.40 [0.95, 2.06] |
| Frequent | 1.16 [0.51, 2.63] | 1.70 [1.03, 2.81] | 1.00 |
| Frequent nicotine use (vs. no use) | |||
| No use | 1.00 | 2.93 [1.83, 4.71] | 9.38 [4.38, 20.09] |
| Occasional | 0.91 [0.55, 1.51] | 1.00 | 2.23 [1.25, 3.96] |
| Frequent | 0.66 [0.26, 1.67] | 1.15 [0.68, 1.93] | 1.00 |
| Prior wave nicotine use | |||
| Occasional cannabis use (vs. no use) | |||
| No use | 1.00 | 3.86 [2.78, 5.36] | 4.32 [1.92, 9.69] |
| Occasional | 0.97 [0.63, 1.49] | 1.00 | 0.75 [0.41, 1.37] |
| Frequent | 0.57 [0.20, 1.63] | 0.63 [0.29, 1.38] | 1.00 |
| Frequent cannabis use (vs. no use) | |||
| No use | 1.00 | 5.15 [3.26, 8.12] | 23.12 [11.15, 47.93] |
| Occasional | 1.04 [0.61, 1.77] | 1.00 | 1.16 [0.56, 2.39] |
| Frequent | 0.95 [0.38, 2.33] | 0.79 [0.39, 1.64] | 1.00 |
Notes: Bold indicates p < .05. CI = confidence interval. Estimates from models that include age, sex assigned at birth, and cohort as covariates.
Figure 2.
Marginal probabilities for cannabis use and cannabis use transitions for three patterns of nicotine use. Transition probabilities differ slightly between the two intervals because of changes in the proportions of the sample in frequency of use categories between Wave 1 and 2.
Figure 3.
Marginal probabilities for nicotine use and nicotine use transitions for three patterns of cannabis use. Transition probabilities differ slightly between the two intervals because of changes in the proportions of the sample in frequency of use categories between Wave 1 and 2.
The associations of both occasional and frequent (compared to no) nicotine use with the odds of both transitions involving onset of cannabis use were positive (OR range: 2.93–9.38; Table 4). For example, individuals reporting occasional nicotine use compared with no use at the prior wave had 3.42 times the odds of transitioning from no cannabis use to occasional cannabis use. This is also reflected in Figure 2: For the Wave 1 to 2 interval, the probability (not odds) of transitioning from no use to occasional cannabis use was .11 for those not using nicotine and .27 for those using nicotine occasionally (Figure 2). Both nicotine use levels were also positively associated with escalating from occasional to frequent cannabis use (occasional nicotine use effect: OR = 1.40; frequent nicotine use effect: OR = 2.23), although the effect of occasional nicotine use on this transition was not statistically significant. The estimated effects of nicotine use on transitions in cannabis use involving de-escalation in use or cessation were weaker and inconsistent in direction (OR range: 0.66–1.70). The only statistically significant association was occasional (vs. no) nicotine use being associated with greater odds (OR = 1.70) of de-escalation from frequent to occasional cannabis use.
Conversely, the estimated effects of occasional and frequent (vs. no) cannabis use on transitions from not using nicotine to using nicotine were positive and statistically significant (OR range: 3.86–23.12; Table 4). For example, reporting occasional cannabis use compared with no use at the prior wave was associated with 3.86 times the odds of transitioning to occasional nicotine use from no use. This is echoed in a difference in probabilities of this transitioning from Wave 1 to 2 of .02 versus .08 (Figure 3). The estimated effects of occasional and frequent (vs. no) cannabis use on escalating from occasional to frequent nicotine use were smaller and not statistically significant (OR range: 0.75–1.16). Most of the estimated effects on de-escalation and cessation were negative, but none were statistically significant (OR range: 0.57–1.04).
Sensitivity analyses
Analyses with different categories of nicotine use are shown in Supplement Tables K–Q. Of individuals who smoked cigarettes at Wave 1, 29% reported using e-cigarettes compared with 1% of individuals who did not smoke cigarettes; conversely, of individuals who used e-cigarettes at Wave 1, 74% reported cigarette use compared with 9% of those who did not use e-cigarettes (Table K). Not surprisingly given this overlap, separate models with cigarette and e-cigarette use yielded similar results, with slightly larger effects of e-cigarette use than cigarette use on cannabis transitions (Tables L–O).
When nicotine use was represented by four categories (neither cigarette nor e-cigarette use, cigarette only, e-cigarette only, both cigarette and e-cigarette use), the two categories involving e-cigarette use each comprised less than 3% of the sample at each wave. Both using cigarettes only and e-cigarettes only (compared with no nicotine use) had significant positive associations with transitions out of the no use cannabis category (OR range: 3.47–9.74), and the estimated effects of using both cigarettes and e-cigarettes (compared with no nicotine use) were small and nonsignificant (OR range: 0.97–1.08; Table P). For models predicting transitions among the four categories of nicotine use, both occasional and frequent cannabis use had positive, statistically significant effects on transitions out of the no nicotine use category into each of the three other categories (OR range: 3.42–7.00; Table Q). None of the effects of cannabis use on transitions between the categories involving some form of nicotine use were statistically significant, with large uncertainty because of few individuals making these transitions.
Discussion
Our study examined associations between nicotine and cannabis use among young adults in the context of legalized nonmedical cannabis and during a period when e-cigarettes became widely available and vaping became an increasingly common mode of cannabis use (Knapp et al., 2019; Patrick et al., 2022). As observed elsewhere (e.g., Soneji et al., 2017), use of tobacco cigarettes and e-cigarettes strongly overlapped in our sample. Nicotine use was associated with increased probability of onset and escalation of cannabis use, and cannabis use was associated with increased probability of onset of nicotine use, corresponding with findings from prior research (Agrawal et al., 2012; Becker et al., 2015; Mayer et al., 2020; Mayet et al., 2011; Strong et al., 2018; Weinberger et al., 2018, 2020a, 2021). Less evidence was found for cross-substance associations with reductions in or cessation of either nicotine or cannabis use.
The findings regarding associations with transitions from no use to use support the theory that use of one potentially harmful substance may open a gateway to initiating use of another and suggest priorities for prevention. For instance, placing nicotine prevention messaging in cannabis retail shops and engaging cannabis retailers in nicotine prevention may hold promise. Conversely, our results support focusing on young adults who have begun using cigarettes or e-cigarettes for prevention of cannabis use. More broadly, the findings suggest that further population-wide decreases in cigarette use may mitigate the potential effects of cannabis legalization on cannabis use and misuse among young adults. On the other hand, increases in cannabis use after legalization, such as those documented in our earlier research with this sample (Kilmer et al., 2022), may hinder further reductions in nicotine initiation.
Findings from prior research on cross-substance effects on cutting down or quitting cannabis and nicotine have been mixed (Agrawal et al., 2012; Lemyre et al., 2019). There is also a history of allowing individuals to continue tobacco use while in treatment for other substance use disorders (Goodwin et al., 2021). We found that use of one substance was linked to lower odds of cessation of the other, but these associations were generally small and not statistically significant. There was also a positive association of occasional nicotine use with decreasing cannabis use from frequent to occasional, perhaps pointing to this transition being easier for individuals who substitute nicotine for cannabis. This contrasts with findings in prior research with an adult sample finding that cigarette use predicted persistence of CUD (Weinberger et al., 2021). Differences in results may be attributable to measures of cannabis use (frequent use vs. CUD) or the age range of the samples (young adults vs. age ≥18). Our findings suggest that dual cessation might not need to be prioritized over other treatment or harm reduction strategies, but more research is needed. Of note, our study used a general population sample, as opposed to specifically individuals seeking to reduce the frequency of their cannabis or nicotine use. Therefore, our findings may not reflect those that would be observed in a treatment setting.
Limitations and future research
Trends in substance use in Washington have been similar to trends in other states (Substance Abuse and Mental Health Services Administration, 2020), but relationships between cannabis and nicotine use may differ depending on state-specific cannabis and nicotine legislation. Future studies could examine these associations in states with different cannabis laws. Where both substances are legal, the role of reaching the legal age for purchase and use of cannabis and nicotine could be further explored. Typical of web-based studies attempting to garner broad and diverse samples, three quarters of the analytic sample was female (assigned at birth). Our sample also differs from the U.S. population in its racial/ethnic composition. A more diverse and larger sample might offer the opportunity to examine differences in cross-substance associations by identity factors. Some research indicates that patterns of nicotine risk factors and use differ by race or ethnicity (Chen et al., 2020; Nicksic et al., 2022), including differential exposure to tobacco advertising. The years of data collection in our study were characterized by shifting trends in young adult substance use, including increases in cannabis and e-cigarette use, increases in vaping cannabis, and decreases in cigarette use (Fleming et al., 2022; Kilmer et al., 2022; Patrick et al., 2022). Examining more recent data and focusing specifically on the interplay of e-cigarette and cannabis use will be informative. A sample with higher frequency of use might allow for further examining transitions across different cut points including true daily use (reported by <5% of the study's sample for both cannabis and nicotine; Supplement Table I). Finally, our study relied on a limited measure of nicotine use that did not include smokeless tobacco or information on nicotine concentration of e-cigarette products, which can vary (DeVito & Krishnan-Sarin, 2018) and have increased over time (Romberg et al., 2019).
Conclusions
In this study of young adults, we found that the strength of cross-substance associations differed by type of transition. Generally, cross-substance associations with transitions were stronger for individuals beginning in the no use category for the given substance use outcome. The findings have implications for prevention, but less clear implications for efforts to decrease population-wide occasional or frequent use of cannabis and nicotine. In the general young adult population, the findings support cannabis and nicotine use as risk factors with potential spillover of use of either type of substance to onset and escalation of the other.
Footnotes
This research was supported by National Institute on Drug Abuse Grants R01DA047996 and R01DA057705 (principal investigator: Katarina Guttmannova). The data were collected with support from the Washington State Health Care Authority's Division of Behavioral Health & Recovery (principal investigator: Jason R. Kilmer). Miranda L. M. Delawalla's contribution was supported by Grant T32AA007455 (principal investigator: Mary Larimer). Partial support for this research also came from a Eunice Kennedy Shriver National Institute of Child Health and Human Development research infrastructure grant, P2C HD042828, to the Center for Studies in Demography & Ecology at the University of Washington. The content of this manuscript is solely the responsibility of the authors and does not necessarily represent the official views of the National Institute on Drug Abuse, National Institutes of Health, University of Washington, or the Washington State Health Care Authority.
References
- Agrawal A., Budney A. J., Lynskey M. T. The co-occurring use and misuse of cannabis and tobacco: A review. Addiction. 2012;107(7):1221–1233. doi: 10.1111/j.1360-0443.2012.03837.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Becker J., Schaub M. P., Gmel G., Haug S. Cannabis use and other predictors of the onset of daily cigarette use in young men: What matters most? Results from a longitudinal study. BMC Public Health. 2015;15(1):843. doi: 10.1186/s12889-015-2194-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Britton M., Haddad S., Derrick J. L. Prospective and daily effects of cannabis use on smoking outcomes during a self-guided quit attempt. Nicotine & Tobacco Research. 2020;22(8):1399–1403. doi: 10.1093/ntr/ntaa083. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chadi N., Schroeder R., Jensen J. W., Levy S. Association between electronic cigarette use and marijuana use among adolescents and young adults: A systematic review and meta-analysis. JAMA Pediatrics. 2019;173(10):e192574. doi: 10.1001/jamapediatrics.2019.2574. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen B., Sterling K. L., Bluestein M. A., Kuk A. E., Harrell M. B., Perry C. L., Pérez A. Age of initiation of cigarillos, filtered cigars and/or traditional cigars among youth: Findings from the Population Assessment of Tobacco and Health (PATH) study, 2013–2017. PLoS One. 2020;15(12):e0243372. doi: 10.1371/journal.pone.0243372. [DOI] [PMC free article] [PubMed] [Google Scholar]
- DeVito E. E., Krishnan-Sarin S. E-cigarettes: Impact of e-liquid components and device characteristics on nicotine exposure. Current Neuropharmacology. 2018;16(4):438–459. doi: 10.2174/1570159X15666171016164430. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Fleming C. B., Ramirez J. J., Rhew I. C., Hultgren B. A., Hanson K. G., Larimer M. E., Dilley J. A., Kilmer J. R., Guttmannova K. Trends in alcohol, cigarette, e-cigarette, and nonprescribed pain reliever use among young adults in Washington State after legalization of non-medical cannabis. Journal of Adolescent Health. 2022;71(1):47–54. doi: 10.1016/j.jadohealth.2022.03.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Ford D. E., Vu H. T., Anthony J. C. Marijuana use and cessation of tobacco smoking in adults from a community sample. Drug and Alcohol Dependence. 2002;67(3):243–248. doi: 10.1016/S0376-8716(02)00066-2. [DOI] [PubMed] [Google Scholar]
- Gilson M. S., Kilmer J. R., Fleming C. B., Rhew I. C., Calhoun B. H., Guttmannova K. Substance-specific risk factors for cannabis and alcohol use among young adults following implementation of non-medical cannabis legalization. Prevention Science. 2023;24(6):1047–1057. doi: 10.1007/s11121-022-01435-8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Goodwin R. D., Wu M., Davidson L. An empirical perspective on cigarette use in substance use recovery. Psychological Medicine. 2021;51(14):2299–2306. doi: 10.1017/S0033291721002300. [DOI] [PubMed] [Google Scholar]
- Graham J. W. Springer Science & Business Media; 2012. Missing data: Analysis and design. [Google Scholar]
- Kaplan D. An overview of Markov chain methods for the study of stage-sequential developmental processes. Developmental Psychology. 2008;44(2):457–467. doi: 10.1037/0012-1649.44.2.457. [DOI] [PubMed] [Google Scholar]
- Kilmer J. R., Rhew I. C., Guttmannova K., Fleming C. B., Hultgren B. A., Gilson M. S., Cooper R. L., Dilley J., Larimer M. E. Cannabis use among young adults in Washington State after legalization of nonmedical cannabis. American Journal of Public Health. 2022;112(4):638–645. doi: 10.2105/AJPH.2021.306641. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Knapp A. A., Lee D. C., Borodovsky J. T., Auty S. G., Gabrielli J., Budney A. J. Emerging trends in cannabis administration among adolescent cannabis users. Journal of Adolescent Health. 2019;64(4):487–493. doi: 10.1016/j.jadohealth.2018.07.012. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Lanza S. T., Bray B. C., Collins L. M. An introduction to latent class and latent transition analysis. In: Schinka J. A., Velicer W. F., Weiner I. B., editors. Handbook of psychology: Research methods in psychology. John Wiley & Sons, Inc; 2013. pp. 691–716. [Google Scholar]
- Lemyre A., Poliakova N., Bélanger R. E. The relationship between tobacco and cannabis use: A review. Substance Use & Misuse. 2019;54(1):130–145. doi: 10.1080/10826084.2018.1512623. [DOI] [PubMed] [Google Scholar]
- Lorenzetti V., Hindocha C., Petrilli K., Griffiths P., Brown J., Castillo-Carniglia Á., Caulkins J. P., Englund A., ElSohly M., gage S. H., Groshkova T., Gual A., Hammond D., Lawn W., López-Pelayo H., Manthey J., Mokrysz C., Pacula R. L., van Laar M., Freeman T. P. The International Cannabis Toolkit (iCannToolkit): A multidisciplinary expert consensus on minimum standards for measuring cannabis use. Addiction. 2022;117(6):1510–1517. doi: 10.1111/add.15702. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mayer M. E., Kong G., Barrington-Trimis J. L., McConnell R., Leventhal A. M., Krishnan-Sarin S. Blunt and non-blunt cannabis use and risk of subsequent combustible tobacco product use among adolescents. Nicotine & Tobacco Research. 2020;22(8):1409–1413. doi: 10.1093/ntr/ntz225. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Mayet A., Legleye S., Chau N., Falissard B. Transitions between tobacco and cannabis use among adolescents: A multi-state modeling of progression from onset to daily use. Addictive Behaviors. 2011;36(11):1101–1105. doi: 10.1016/j.addbeh.2011.06.009. [DOI] [PubMed] [Google Scholar]
- McClure E. A., Baker N. L., Hood C. O., Tomko R. L., Squeglia L. M., Flanagan J. C., Carpenter M. J., Gray K. M. Cannabis and alcohol co-use in a smoking cessation pharmacotherapy trial for adolescents and emerging adults. Nicotine & Tobacco Research. 2020;22(8):1374–1382. doi: 10.1093/ntr/ntz170. [DOI] [PMC free article] [PubMed] [Google Scholar]
- McClure E. A., Tomko R. L., Salazar C. A., Akbar S. A., Squeglia L. M., Herrmann E., Carpenter M. J., Peters E. N. Tobacco and cannabis co-use: Drug substitution, quit interest, and cessation preferences. Experimental and Clinical Psychopharmacology. 2019;27(3):265–275. doi: 10.1037/pha0000244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Muthén L. K., Muthén B. Mplus user's guide: Statistical analysis with latent variables, user's guide. 2017 Authors. [Google Scholar]
- Nicksic N. E., Bono R. S., Rudy A. K., Cobb C. O., Barnes A. J. Smoking status and racial/ethnic disparities in youth exposure to tobacco advertising. Journal of Ethnicity in Substance Abuse. 2022;21(3):959–974. doi: 10.1080/15332640.2020.1815113. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Patrick M., Schulenberg J., Miech R., Johnston L., O'Malley P., Bachman J. Ann Arbor, MI: University of Michigan Institute for Social Research; 2022. Monitoring the Future Panel Study annual report: National data on substance use among adults ages 19 to 60, 1976–2021. Monitoring the Future Monograph Series. [Google Scholar]
- Rabin R. A., Ashare R. L., Schnoll R. A., Cinciripini P. M., Hawk L. W., Jr., Lerman C., Tyndale R. F., George T. P. Does cannabis use moderate smoking cessation outcomes in treatment-seeking tobacco smokers? Analysis from a large multi-center trial. The American Journal on Addictions. 2016;25(4):291–296. doi: 10.1111/ajad.12382. [DOI] [PubMed] [Google Scholar]
- Rabin R. A., George T. P. A review of co-morbid tobacco and cannabis use disorders: Possible mechanisms to explain high rates of co-use. The American Journal on Addictions. 2015;24(2):105–116. doi: 10.1111/ajad.12186. [DOI] [PubMed] [Google Scholar]
- Reboussin B. A., Wagoner K. G., Ross J. C., Suerken C. K., Sutfin E. L. Tobacco and marijuana co-use in a cohort of young adults: Patterns, correlates and reasons for co-use. Drug and Alcohol Dependence. 2021;227:109000. doi: 10.1016/j.drugalcdep.2021.109000. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Romberg A. R., Miller Lo E. J., Cuccia A. F., Willett J. G., Xiao H., Hair E. C., Vallone D. M., Marynak K., King B. A. Patterns of nicotine concentrations in electronic cigarettes sold in the United States, 2013-2018. Drug and Alcohol Dependence. 2019;203:1–7. doi: 10.1016/j.drugalcdep.2019.05.029. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Soneji S., Barrington-Trimis J. L., Wills T. A., Leventhal A. M., Unger J. B., Gibson L. A., Yang J. W., Primack B. A., Andrews J. A., Miech R. A., Spindle T. R., Dick D. M., Eissenberg T., Hornik R. C., Dang R., Sargent J. D. Association between initial use of e-cigarettes and subsequent cigarette smoking among adolescents and young adults: A systematic review and meta-analysis. JAMA Pediatrics. 2017;171(8):788–797. doi: 10.1001/jamapediatrics.2017.1488. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Strong D. R., Myers M. G., Pulvers K., Noble M., Brikmanis K., Doran N. Marijuana use among US tobacco users: Findings from wave 1 of the Population Assessment of Tobacco Health (PATH) study. Drug and Alcohol Dependence. 2018;186:16–22. doi: 10.1016/j.drugalcdep.2017.12.044. [DOI] [PubMed] [Google Scholar]
- Substance Abuse and Mental Health Services Administration. Key substance use and mental health indicators in the United States: Results from the 2019 National Survey on Drug Use and Health. 2020 HHS Publication No. PEP20-07-01-001, NSDUH Series H-55. [PubMed] [Google Scholar]
- Tucker J. S., Pedersen E. R., Seelam R., Dunbar M. S., Shih R. A., D'Amico E. J. Types of cannabis and tobacco/nicotine co-use and associated outcomes in young adulthood. Psychology of Addictive Behaviors. 2019;33(4):401–411. doi: 10.1037/adb0000464. [DOI] [PMC free article] [PubMed] [Google Scholar]
- U.S. Census Bureau. Washington population change between census decade. 2020 https://www.census.gov/library/stories/state-by-state/washington-population-change-between-census-decade.html [Google Scholar]
- Weinberger A. H., Delnevo C. D., Wyka K., Gbedemah M., Lee J., Copeland J., Goodwin R. D. Cannabis use is associated with increased risk of cigarette smoking initiation, persistence, and relapse among adults in the United States. Nicotine & Tobacco Research. 2020a;22(8):1404–1408. doi: 10.1093/ntr/ntz085. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weinberger A. H., Pacek L. R., Wall M. M., Gbedemah M., Lee J., Goodwin R. D. Cigarette smoking quit ratios among adults in the USA with cannabis use and cannabis use disorders, 2002-2016. Tobacco Control. 2020b;29(1):74–80. doi: 10.1136/tobaccocontrol-2018-054590. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weinberger A. H., Platt J., Copeland J., Goodwin R. D. Is cannabis use associated with increased risk of cigarette smoking initiation, persistence, and relapse? Longitudinal data from a representative sample of US adults. Journal of Clinical Psychiatry. 2018;79(2):2254. doi: 10.4088/JCP.17m11522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Weinberger A. H., Platt J., Zhu J., Levin J., Ganz O., Goodwin R. D. Cigarette use and cannabis use disorder onset, persistence, and relapse: Longitudinal data from a representative sample of US adults. Journal of Clinical Psychiatry. 2021;82(4):34976. doi: 10.4088/JCP.20m13713. [DOI] [PMC free article] [PubMed] [Google Scholar]