Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2026 Aug 6.
Published in final edited form as: Addict Behav. 2025 Aug 6;170:108451. doi: 10.1016/j.addbeh.2025.108451

The gateway effect of cigarette, e-cigarette, cigar, and alcohol use vs. cannabis use

Yan Wang a,b, Patricia A Cavazos-Rehg c, Yuxian Cui a, Morgan Speer a, Cassidy R LoParco a, Darcey McCready a, Y Tony Yang b,d, Carla J Berg a,b
PMCID: PMC12372761  NIHMSID: NIHMS2104503  PMID: 40795599

Abstract

Background:

While the gateway hypothesis suggests that using tobacco and alcohol increases likelihood of initiating cannabis, cannabis use may precede and increase other substance use. We examined gateway effects of cigarettes, e-cigarettes, cigars, and alcohol on cannabis use, and reverse associations.

Methods:

We analyzed 2023 survey data from 4,031 US young adults (Mage=26.29, 60% female, 19% Hispanic, 14% Black, 14% Asian). Discrete-time survival analysis assessed hazards of initiating cannabis based on self-reported age of initiating other substances, and vice versa. Time(age)-lagged predictors indicated whether participants had initiated the other substances by one year younger, accounting for sociodemographics; state non-medical cannabis laws; lifetime depression, anxiety, or attention deficit disorder [ADD]) diagnoses; and personality characteristics.

Results:

Lifetime use was: 68% for cannabis, 45% cigarettes, 49% e-cigarettes, 31% cigars, and 85% alcohol. Past-year cigarette, e-cigarette, cigar, and alcohol initiation increased hazards of initiating cannabis (adjusted Hazard Ratio, aHR=3.78, 95%CI=3.39-4.22; aHR=2.17, 95%CI=1.86-2.53; aHR=2.90, 95%CI=2.45-3.43; aHR=3.41, 95%CI=3.11-3.75, respectively). Past-year cannabis initiation increased hazards of other substance initiation (cigarettes: aHR=3.51, 95%CI=3.11-3.96; e-cigarettes: aHR=3.73, 95%CI=3.34-4.17; cigars: aHR=3.66, 95%CI=3.20-4.18; alcohol: aHR=3.07, 95%CI=2.73-3.45). Associations were generally stronger when initiation occurred at ages 5-18 vs. >18. Depression predicted cannabis initiation; anxiety and ADD predicted e-cigarette initiation. Certain personality characteristics were protective against initiation (agreeableness and conscientiousness for each, openness for cigarettes and cigars, emotional stability for cannabis, cigarettes, and cigars); extraversion increased hazards of initiating cannabis and e-cigarettes.

Conclusions:

Interventions should target underlying mechanisms influencing the use of various substances, such as mental health and personality characteristics, especially among adolescents.

Keywords: Polysubstance use, Substance use, Young adults, Risk factors, Substance use initiation

1. INTRODUCTION

The gateway hypothesis posits that substance use initiation occurs in a sequential progression through different stages (Kandel, 2003; Kandel et al., 2006), usually beginning with alcohol and tobacco, followed by cannabis, then illegal/illicit substances (Barry et al., 2016; Keyes et al., 2015; Lemyre et al., 2019; Reed et al., 2022). While some research suggests these associations hold even when adjusting for certain individual (e.g., socio-economic status, personality traits) and contextual factors (e.g., family functioning) (Fergusson et al., 2006; Kandel et al., 1986; Lessem et al., 2006), other research indicates the gateway sequence of substance use may no longer hold when accounting for such factors.

One factor potentially impacting order of initiation involves shifts in a country’s product market and use prevalence (Degenhardt et al., 2010). In the US in 2023, lifetime use rates of cigarettes, e-cigarettes, cigars, alcohol, and cannabis among young adults (ages 18-25) were 34.6%, 50.1%, 23.1%, 74.7%, and 50.0% respectively, and past-month rates were 10.6%, 24.1%, 5.3%, 49.6%, and 25.2% (Substance Abuse and Mental Health Services Administration, 2024). Because of the higher use rates for cannabis and e-cigarettes than cigarettes and cigars, the traditional gateway hypothesis may not hold. Furthermore, these associations may be impacted by cohort effects, as cannabis and e-cigarette use has increased in the past decade (Substance Abuse and Mental Health Services Administration, 2024) coinciding with expansion of non-medical cannabis legalization (National Conference of State Legislators, 2024) and e-cigarette popularity since 2010 (Fadus et al., 2019). Notably, cannabis legalization may increase cannabis use but reduce tobacco and alcohol use (Dave et al., 2023; De & Sun, 2025).

Despite these important societal changes, limited research has assessed the gateway effects of cigars or newer substances like e-cigarettes to cannabis. Some studies show effects for e-cigarettes (Morean et al., 2015; Temourian et al., 2024; Temple et al., 2017; Unger et al., 2016) and cigars (Cornacchione Ross et al., 2020), while others show no or minimal effects or associations (e.g., e-cigarettes (Jorgensen & Wells, 2022), cigars (Cohn et al., 2018)). Furthermore, several studies show that some use cannabis before tobacco and/or alcohol or that cannabis use increases the probability of progressing to tobacco/alcohol use (Agrawal et al., 2006; Badiani et al., 2015; Cohn et al., 2018; Kokkevi et al., 2006; Mayet et al., 2016; Patton et al., 2005; Ramo et al., 2012; Reed et al., 2022; Scholes-Balog et al., 2016; Secades-Villa et al., 2015).

Among the explanations for these complex associations is the common liability model of vulnerability (Mayet et al., 2016; Selya, 2024; Vanyukov et al., 2012), which suggests that common genetic and individual factors, like mental health (e.g., depression (Moreno-Mansilla et al., 2021)), behavioral factors (e.g., sensation-seeking, impulse control (Wojciechowski, 2024)), or personality characteristics (Dash et al., 2023; Hokm Abadi et al., 2018; Kotov et al., 2010), increase risk of polysubstance use, regardless of order of initiation (Eaton et al., 2015; Moreno-Mansilla et al., 2021). Relatedly, one study (Melberg et al., 2010) documented the cannabis gateway effect only among youth who initiated substance use at younger ages or showed mental health or behavioral problems.

Given the complexity of previous findings, the range of explanatory models used, and the important implications of such findings for prevention efforts (Jorgensen & Wells, 2022; Melberg et al., 2010; Vanyukov et al., 2012), this topic warrants additional research. In particular, research is needed that considers: 1) the current context with increased e-cigarette and cannabis use in the more recent regulatory and marketing environment; 2) potential age or cohort effects that may manifest due to these shifts; and 3) the anticipated directions of substance use transitions suggested by the gateway theory (i.e., from tobacco/nicotine and alcohol use to cannabis use) and the reverse direction across this range of substances.

Thus, this study: 1) analyzed 2023 data among US young adults across all 50 states; 2) examined prior initiation of several legal substances – cigarettes, e-cigarettes, cigars, and alcohol – in relation to cannabis initiation, as well as the reverse relationships (i.e., prior cannabis initiation in relation to subsequent initiation of these other substances); 3) qualitatively compared strengths of these associations across substances in directionality and during adolescence vs. young adulthood; and 4) accounted for key contextual factors (i.e., cannabis policy, mental health, personality characteristics).

2. METHODS

2.1. Participants

We analyzed baseline (Wave 1 [W1]) data from a longitudinal study among 4,031 young adults in the Cannabis Regulation, Marketing & Appeal (CARMA) study, which assesses cannabis poicy, marketing, and impact (Berg et al., 2023). In June-November 2023, eligible individuals (English-speaking US residents ages 18-34) were recruited via Facebook ads. Individuals who clicked on ads were messaged via chatbot on Facebook Messenger; the chatbot provided a brief study overview and administered preliminary screening questions (e.g., age, US residence, sex, race/ethnicity), then directed those deemed preliminarily eligible to the study webpage to be officially consented, screened to confirm eligibility, and administered the W1 survey. We used purposive, quota-based sampling to obtain a sample comprising ~50% past-month cannabis use, ~50% per sex, and ~40% racial/ethnic minorities.

2.2. Measures

All data were from the W1 survey. Participants were shown a table of different substances (e.g., alcohol, cannabis, cigarettes, e-cigarettes) with descriptions and example photos. Cannabis was described as “marijuana (i.e., cannabis, pot, weed) including: dried herb, edibles, oils, hash, kief, concentrates, marijuana drinks, tinctures, lotions, etc. (Don’t include hemp-derived cannabinoids, like Delta-8.)” (Substance Abuse and Mental Health Services Administration, 2024).

Substance use.

Key variables were lifetime use of each substance (cannabis, cigarettes, e-cigarettes [for nicotine], large or little cigars, alcohol) and retrospective self-reported age at first use of each substance ever used: ≤5, 6, 7…33, or ≥34. Few participants reported first use at ≤5 (range 1 for cigarettes and cigars to 40 for alcohol) or ≥34 (range 0 for alcohol to 5 for e-cigarettes). Thus, we coded ≤5 as 5 and ≥34 as 34. Also, the few instances (n=22, 0.5%) where participants provided illogical responses, specifically older age at initiation than at the W1 survey (i.e., 3 for cannabis, 4 cigarettes, 8 e-cigarettes, 2 cigars, 5 alcohol), were treated as censored at age of survey.

Psychosocial factors.

Participants self-reported lifetime diagnosis of depressive disorder, anxiety disorder, or attention deficit disorder (ADD). Personality characteristics (extraversion, agreeableness, conscientiousness, emotional stability, openness to new experiences) were assessed via the 10-Item Personality Inventory (TIPI), with 2 items for each characteristic (1=strongly disagree to 7=strongly agree; mean subscale scores range 1-7; Cronbach’s alpha were .35-.72) (Gosling et al., 2003).

Sociodemographics.

Participants reported age, sex, sexual orientation, ethnicity, race, parental education, and community type (e.g., rural), and state of residence, coded to reflect whether non-medical cannabis was legalized in that state at the time of survey assessment (yes vs. no).

2.3. Data Analysis

Sample characteristics were assessed overall and by lifetime use status for each substance. Frequencies/percentages of age-period during which each substance was initiated were plotted.

Because events of interest (initiation of each substance) were reported discretely in years, we used discrete-time survival analysis (DTSA) (Singer & Willett, 2003) to assess hazards of initiating each substance across age-periods (ages 5-34 years), using logit link in STATA 18.0. Observations were reorganized as person-period data, with each row indicating the event of initiation within a specific age-period (age-year). For each substance, the variable was coded as 0 for age-periods at pre-initiation, 1 at age-periods of initiation, and excluded at age-periods post-initiation. For participants reporting no lifetime use, time to event (initiation) was censored; therefore, each variable was coded 0 across all age-years. Hazard was defined as the conditional probability of substance initiation at age-period t, given that it had not occurred in earlier age-periods (t–1). To predict the event (e.g., cannabis initiation) at a specific age-period t, we included prior use of another substance (e.g., cigarettes) as a time-variant and time-lagged predictor indicating whether the participant had used the substance (e.g., cigarette) by the age-period one year younger than this age-period (t–1). Time-invariant covariates included sociodemographic and psychosocial factors. The hazard functions over time were plotted by levels of age-predictors.

We then tested the proportional hazard (PH) assumption, specifically whether the hazard ratio (HR) of initiating a substance (e.g., cannabis) at age-period t in relation to the predictor (e.g., ever use of cigarette by age-period t–1) differed across age-periods. To do so, we compared model with predictors and baseline hazard model (with all the dummy variables of age-periods) to the model also including interactions between dummy age-period variables and the predictor using Likelihood Ratio Test (LRT), adjusting for covariates – with significance suggesting that the PH assumption did not hold across the age-periods. We further tested whether HRs differed between childhood/adolescence (5-18 years old) and adulthood (>18) based on LRT and stratified DTSA models by age-period categories (5-18 vs. >18). Finally, sensitivity analyses excluding age responses “5 or younger” and “34 or older” were conducted; results were similar.

3. RESULTS

3.1. Sample characteristics

At W1 survey assessment, participants were an average age of 26.29 (SD=4.81), 60% female, 28% sexual minority, 19% Hispanic, 65% White, 14% Black, 14% Asian, 48% urban, and 51% residing in states with legalized non-medical cannabis (Table 1).

Table 1. Selected sample characteristics by lifetime use of cannabis, cigarettes, e-cigarettes, cigars, and alcohol, separately.

Variables All
participants
N=4,031
(100%)		 Cannabis use Cigarette use E-cigarette use Cigar use Alcohol use
No
n=1,273
(32%)		 Yes
n=2,758
(68%)		 No
n=2,200
(55%)		 Yes
n=1,831
(45%)		 No
n=2,063
(51%)		 Yes
n=1,968
(49%)		 No
n=2,774
(69%)		 Yes
n=1,257
(31%)		 No
n=599
(15%)		 Yes
n=3,432
(85%)
n (%) or
M (SD)		 n (%) or
M (SD)		 n (%) or
M (SD)		 p n (%) or
M (SD)		 n (%) or
M (SD)		 p n (%) or
M (SD)		 n (%) or
M (SD)		 p n (%) or
M (SD)		 n (%) or
M (SD)		 p n (%) or
M (SD)		 n (%) or
M (SD)		 p
Sociodemographics
Age (at W1 survey)* 26.29 (4.81) 25.07 (5.00) 26.85 (4.62) <.001 23.11 (4.81) 27.69 (4.42) <.001 26.06 (4.98) 26.53 (4.63) .002 25.60 (4.86) 27.79 (4.34) <.001 23.94 (4.96) 26.69 (4.67) <.001
Sex .003 .747 .574 <.001 <.001
 Female 2,394 (60) 708 (30) 1,686 (70) 1,314 (55) 1,080 (45) 1,239 (52) 1,155 (48) 1,785 (75) 609 (25) 297 (12) 2,097 (88)
 Male 1,612 (40) 558 (35) 1,054 (65) 871 (54) 741 (46) 810 (50) 802 (50) 974 (60) 638 (40) 298 (18) 1,314 (82)
Sexual orientation <.001 <.001 <.001 .020 <.001
 Heterosexual 2,844 (72) 999 (35) 1,845 (65) 1,593 (56) 1,251 (44) 1,527 (53) 1,317 (46) 1,946 (68) 898 (32) 476 (17) 2,368 (83)
 Other orientation 1,105 (28) 232 (21) 873 (79) 551 (50) 554 (50) 480 (43) 625 (57) 760 (69) 345 (31) 104 (9) 1,001 (91)
Race <.001 <.001 <.001 <.001 <.001
 White 2,525 (65) 793 (31) 1,732 (69) 1,310 (52) 1,215 (48) 1,266 (50) 1,259 (50) 1,691 (67) 834 (33) 349 (14) 2,176 (86)
 Black 545 (14) 109 (20) 436 (80) 311 (57) 234 (43) 255 (47) 290 (53) 357 (66) 188 (34) 78 (14) 467 (86)
 Asian 549 (14) 273 (50) 276 (50) 374 (68) 175 (32) 346 (63) 203 (37) 455 (83) 94 (17) 125 (23) 424 (77)
 Other 269 (7) 53 (20) 216 (80) 115 (43) 154 (57) 113 (42) 156 (58) 171 (64) 98 (36) 23 (9) 246 (91)
Ethnicity .003 .036 .028 .145 .068
 Non-Hispanic 3,204 (81) 1,039 (32) 2,165 (68) 1,757 (55) 1,447 (45) 1,649 (51) 1,555 (49) 2,197 (69) 1007 (31) 479 (15) 2,725 (85)
 Hispanic 766 (19) 208 (27) 558 (73) 401 (52) 365 (48) 381 (50) 385 (50) 528 (69) 238 (31) 105 (14) 661 (86)
Parent education <.001 <.001 <.001 <.001 .054
 <bachelor’s 2,334 (59) 604 (26) 1,730 (74) 1,135 (49) 1,199 (51) 1,075 (46) 1,259 (54) 1,525 (65) 809 (35) 326 (14) 2,008 (86)
 ≥bachelor’s 1,636 (41) 640 (39) 996 (61) 1,019 (62) 617 (38) 950 (58) 686 (42) 1,198 (73) 438 (27) 259 (16) 1,377 (84)
Ruralit^ .097 .003 .028 .120 .030
 Rural 844 (21) 279 (33) 565 (67) 419 (50) 425 (50) 402 (48) 442 (52) 567 (67) 277 (33) 141 (17) 703 (83)
 Suburban 1,226 (31) 397 (32) 829 (68) 703 (57) 523 (43) 654 (53) 572 (47) 863 (70) 363 (30) 183 (15) 1,043 (85)
 Urban 1,939 (48) 586 (30) 1,353 (70) 1,063 (55) 876 (45) 992 (51) 947 (49) 1,325 (68) 614 (62) 268 (14) 1,671 (86)
State non-medical cannabis law .151 .027 .013 .002 .447
 Not legalized 1988 (49) 649 (33) 1339 (67) 1050 (53) 938 (47) 978 (49) 1010 (51) 1323 (67) 665 (33) 304 (15) 1684 (85)
 Legalized 2043 (51) 624 (31) 1419 (69) 1150 (56) 893 (44) 1085 (53) 958 (47) 1451 (71) 592 (29) 295 (14) 1748 (86)
Lifetime mental health diagnoses <.001
Depression <.001 <.001 <.001 <.001
 No 2,315 (57) 902 (39) 1,413 (61) 1,401 (61) 911 (39) 1,324 (57) 991 (43) 1,653 (71) 662 (29) 425 (18) 1,890 (82)
 Yes 1,716 (43) 371 (22) 1,345 (78) 796 (46) 920 (54) 739 (43) 977 (57) 1,121 (65) 595 (35) 174 (10) 1,542 (90)
Anxiety <.001 <.001 <.001 .001 <.001
 No 2,202 (55) 867 (39) 1,335 (61) 1,344 (61) 858 (39) 1,272 (58) 930 (42) 1,566 (71) 636 (29) 407 (18) 1,795 (82)
 Yes 1,829 (45) 406 (22) 1,423 (78) 856 (47) 973 (53) 791 (43) 1,038 (57) 1,208 (66) 621 (34) 192 (11) 1,637 (89)
ADD <.001 <.001 <.001 <.001 .001
 No 3,246 (81) 1,115 (34) 2,131 (66) 1,848 (57) 1,398 (43) 1,762 (54) 1,484( 46) 2,300 (71) 946 (29) 511 (16) 2,735 (84)
 Yes 785 (19) 158 (20) 627 (80) 352 (45) 433 (55) 301 (38) 484 (62) 474 (60) 311 (40) 88 (11) 697 (89)
Personality characteristics
 Extraversion 3.70 (1.56) 3.50 (1.53) 2.79 (1.56) <.001 3.57 (1.56) 3.86 (1.55) <.001 3.58 (1.56) 3.82 (1.55) <.001 3.60 (1.57) 3.91 (1.52) <.001 3.58 (1.50) 3.71 (1.57) .045
 Agreeableness 4.89 (1.18) 4.85 (1.18) 4.91 (1.17) 0.124 4.90 (1.18) 4.88 (1.17) 0.632 4.92 (1.19) 4.87 (1.16) 0.235 4.93 (1.18) 4.83 (1.16) .012 4.84 (1.22) 4.90 (1.17) .226
 Conscientiousness 5.10 (1.34) 5.22 (1.32) 5.04 (1.35) <.001 5.16 (1.36) 5.01 (1.31) <.001 5.19 (1.34) 5.00 (1.34) <.001 5.12 (1.35) 5.04 (1.32) .063 5.09 (1.39) 5.10 (1.33) .880
 Emotional stability 4.10 (1.45) 4.23 (1.43) 4.04 (1.45) <.001 4.16 (1.43) 4.03 (1.47) .005 4.16 (1.44) 4.03 (1.45) .004 4.07 (1.44) 4.16 (1.46) .095 4.27 (1.45) 4.07 (1.45) .002
 Openness to new experiences 5.14 (1.21) 4.90 (1.20) 5.25 (1.20) <.001 5.07 (1.20) 5.23 (1.23) <.001 5.07 (1.20) 5.22 (1.23) <.001 5.11 (1.20) 5.22 (1.25) .007 4.88 (1.22) 5.19 (1.21) <.001
Open in a new tab

Notes: *All n (%) except age and depressive symptoms, showing M (SD). Missing values: sex: other/refuse n=25; sexual orientation: refuse n=82; race: refuse n=143; ethnicity: don’t know/refuse n=61; rurality: other n=22; parental education: don’t know n=61. ^ Rural: <10,000 people; suburban/micropolitan: 10,000-49,999 people; urban: ≥50,000 people. The % reported is among those who responded; responses such as “prefer not to answer”, “don’t know”, etc. were excluded from the denominator when reporting the %.

3.2. Lifetime substance use

Lifetime use was 68% for cannabis (n=2,758), 45% for cigarettes (n=1,831), 49% for e-cigarettes (n=1,968), 31% for cigars (n=1,257), and 85% for alcohol (n=3,432). Age of initiation varied across substances (Figure 1). For cannabis, cigarettes, cigars, and alcohol, the largest percentages initiated at 13-18 years old (58%, 59%, 56%, 60%, separately), but for e-cigarettes, the largest percentage initiated at ages 19-25 (45%). About 8%, 15%, 2%, 6%, and 11% initiated cannabis, cigarettes, e-cigarettes, cigars, and alcohol at ages 5-12. Only 1-4% initiated any substance at ages 26-34. Mean age of initiation was 17.54 (SD=4.01) for cannabis, 16.47 (SD=3.86) for cigarettes, 20.16 (SD=4.43) for e-cigarettes, 18.13 (SD=3.91) for cigars, and 16.71 (SD=2.63) for alcohol. Correlates of lifetime substance use are shown in Table 1.

Figure 1. Percent initiating use of each substance within each age-period among the participants who reported lifetime use of the substance.

Figure 1.

Open in a new tab

3.3. Participants following gateway or reverse gateway sequence

Among the 4,031 participants, 451 did not initiate cannabis, cigarettes, e-cigarettes, cigars, or alcohol; 53 initiated cannabis but no other substance; 824 initiated cigarettes, e-cigarettes, cigars, or alcohol but not cannabis; and 702 reported the same age of initiating cigarettes, e-cigarettes, cigars, or alcohol (whichever was youngest) as initiating cannabis. Of the 2,001 participants remaining, 1,589 (79%) followed the gateway sequence (i.e., used another substance before cannabis) and 412 (21%) the reverse gateway (i.e., used cannabis before all other substances). Of the 1,212 reporting use of cigarettes and cannabis, 50% (n=609) reported using cigarettes first, and 50% (n=603) cannabis first. Of the 1,452 who used e-cigarettes and cannabis, 17% (n=250) used e-cigarettes first, and 83% (n=1,202) cannabis first. Of the 865 who used cigars and cannabis, 24% (n=208) used cigars first, and 76% (n=657) cannabis first. Of the 1,978 who used alcohol and cannabis, 68% (n=1,341) used alcohol first, and 32% (n=637) cannabis first.

3.4. Hazards of cannabis initiation based on prior use of other substances

Past-year cigarette, e-cigarette, cigar, and alcohol initiation significantly increased hazards of cannabis initiation (adjusted Hazard Ratio, aHR=3.78, 95%CI=3.39-4.22; aHR=2.17, 95%CI=1.86-2.53; aHR=2.90, 95%CI=2.45-3.43; aHR=3.41, 95%CI=3.11-3.75; Table 2). Other factors associated with higher hazards of cannabis initiation in all 4 models included being younger, female, Black or other race (vs. White), White (vs. Asian), with parents with <bachelor’s degree (vs. ≥bachelor’s degrees), prior depressive disorder diagnosis, higher extraversion, and lower agreeableness, conscientiousness, and emotional stability. Other model-specific correlates are shown in Table 2.

Table 2. Discrete-time survival analysis (DTSA) assessing initiation of cannabis use, based on prior use* of each substance, adjusting for sociodemographic, state non-medical cannabis laws, and psychosocial variables.

Variables Cigarettes E-cigarettes Cigars Alcohol
aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p
Prior use* (ref: no use) 3.78 (3.39-4.22) <.001 2.17 (1.86-2.53) <.001 2.90 (2.45-3.43) <.001 3.41 (3.11-3.75) <.001
Sociodemographics
Age (at W1 survey) 0.93 (0.92-0.94) <.001 0.94 (0.93-0.95) <.001 0.94 (0.93-0.94) <.001 0.94 (0.93-0.94) <.001
Male (ref: female) 0.85 (0.78-0.93) <.001 0.85 (0.78-0.93) <.001 0.84 (0.77-0.91) <.001 0.90 (0.82-0.98) .015
Other sexual orientation (ref: heterosexual) 1.06 (0.96-1.16) .244 1.10 (1.00-1.20) .047 1.10 (1.00-1.21) .044 0.99 (0.90-1.09) .857
Race (ref: White)
 Black 1.22 (1.09-1.37) .001 1.14 (1.02-1.28) .026 1.17 (1.04-1.31) .009 1.17 (1.04-1.31) .009
 Asian 0.50 (0.44-0.58) <.001 0.49 (0.43-0.56) <.001 0.51 (0.44-0.59) <.001 0.51 (0.44-0.59) <.001
 Other 1.20 (1.03-1.41) .020 1.17 (1.00-1.36) .049 1.23 (1.05-1.43) .010 1.23 (1.05-1.43) .010
Hispanic (ref: non-Hispanic) 0.90 (0.81-1.00) .056 0.90 (0.81-1.00) .052 0.91 (0.82-1.01) .085 0.91 (0.82-1.01) .085
Parent education ≥bachelor’s (ref: <bachelor’s) 0.69 (0.64-0.76) <.001 0.66 (0.61-0.72) <.001 0.66 (0.60-0.72) <.001 0.51 (0.44-0.59) <.001
Rurality (ref: Rural)
 Suburban/micropolitan 0.91 (0.81-1.01) .076 0.87 (0.78-0.97) .010 0.87 (0.78-0.97) .011 1.17 (1.04-1.31) .009
 Urban 1.04 (0.94-1.15) .461 1.00 (0.90-1.10) .948 1.01 (0.91-1.11) .921 1.17 (1.04-1.31) .009
State non-medical cannabis legalized (vs. not) 1.04 (0.96-1.13) .351 1.02 (0.94-1.11) .620 1.03 (0.95-1.12) .472 1.01 (0.94-1.10) .719
Lifetime mental health diagnoses
 Depressive disorder (vs. no) 1.11 (1.00-1.23) .047 1.14 (1.03-1.26) .015 1.14 (1.03-1.26) .011 1.12 (1.01-1.25) .030
 Anxiety disorder (vs. no) 1.11 (1.00-1.23) .048 1.07 (0.96-1.18) .204 1.10 (0.99-1.21) .080 1.10 (0.99-1.23) .065
 ADD (vs. no) 1.08 (0.97-1.20) .152 1.07 (0.96-1.19) .205 1.06 (0.95-1.17) .304 1.07 (0.96-1.18) .226
Personality characteristics
 Extraversion 1.05 (1.02-1.07) .001 1.05 (1.02-1.08) .001 1.05 (1.02-1.08) <.001 1.06 (1.03-1.09) <.001
 Agreeableness 0.89 (0.86-0.92) <.001 0.87 (0.84-0.90) <.001 0.88 (0.85-0.91) <.001 0.88 (0.85-0.91) <.001
 Conscientiousness 0.86 (0.84-0.89) <.001 0.85 (0.83-0.88) <.001 0.85 (0.83-0.88) <.001 0.86 (0.83-0.88) <.001
 Emotional stability 0.96 (0.93-0.99) .009 0.94 (0.91-0.97) <.001 0.94 (0.91-0.98) .001 0.95 (0.92-0.99) .005
 Openness to new experiences 0.99 (0.95-1.02) .478 1.00 (0.96-1.03) .792 1.00 (0.96-1.03) .937 0.98 (0.94-1.01) .164
Open in a new tab

Notes: * Prior use defined as ever having used the substance by the age one year earlier (time-lagged predictor). aHR: Adjusted Hazard Ratio adjusted for all the other predictors or covariates in the table.

3.5. Hazards of initiating other substances based on prior cannabis use

Past-year cannabis use increased hazards of initiating each other substance (cigarettes: aHR=3.51, 95%CI=3.11-3.96; e-cigarettes: aHR=3.73, 95%CI=3.34-4.17; cigars: aHR=3.66, 95%CI=3.20-4.18; alcohol: aHR=3.07, 95%CI=2.73-3.45; Table 3). Other factors associated with other substance iniation in all 4 models included lower hazards of initiation included younger age and lower agreeableness and conscientiousness. Other model-specific correlates are shown in Table 3.

Table 3. Discrete-time survival analysis (DTSA) assessing initiation of each substance, based on prior cannabis use*, adjusting for sociodemographic, state non-medical cannabis laws, and psychosocial variables.

Variables Cigarettes E-cigarettes Cigars Alcohol
aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p
Prior cannabis use* (ref: no use) 3.51 (3.11-3.96) <.001 3.73 (3.34-4.17) <.001 3.66 (3.20-4.18) <.001 3.07 (2.73-3.45) <.001
Sociodemographics
Age (at W1 survey) 0.99 (0.98-1.00) .003 0.88 (0.87-0.89) <.001 0.96 (0.95-0.97) <.001 0.95 (0.94-0.96) <.001
Male (ref: female) 0.91 (0.82-1.00) .062 1.18 (1.06-1.31) .002 1.63 (1.45-1.84) <.001 0.71 (0.66-0.77) <.001
Other sexual orientation (ref: heterosexual) 0.95 (0.85-1.07) .401 1.09 (0.98-1.22) .113 0.81 (0.71-0.93) .003 1.20 (1.10-1.32) <.001
Race (ref: White)
 Black 0.60 (0.52-0.70) <.001 0.88 (0.76-1.01) .069 0.85 (0.72-1.01) .058 0.69 (0.61-0.78) <.001
 Asian 0.60 (0.50-0.71) <.001 0.92 (0.78-1.09) .322 0.45 (0.36-0.56) <.001 0.59 (0.52-0.67) <.001
 Other 1.09 (0.91-1.30 ) .368 1.05 (0.88-1.25) .604 1.01 (0.81-1.25) 0.947 1.04 (0.89-1.22) .602
Hispanic (ref: non-Hispanic) 0.77 (0.68-0.88) <.001 0.93 (0.82-1.05) .245 0.69 (0.59-0.8) <.001 0.85 (0.76-0.94) .002
Parent education ≥bachelor’s (ref: <bachelor’s) 0.66 (0.60-0.74) <.001 0.85 (0.77-0.95) .003 0.8 (0.71-0.91) .001 0.93 (0.86-1.01) .078
Rurality (ref: Rural)
 Suburban/micropolitan 0.66 (0.58-0.75) <.001 0.93 (0.81-1.06) .260 0.77 (0.66-0.89) .001 0.92 (0.83-1.02) .110
 Urban 0.76 (0.67-0.85) <.001 1.01 (0.89-1.15) .825 0.88 (0.76-1.02) .086 1.02 (0.92-1.13) .682
State non-medical cannabis legalized (vs. not) 0.83 (0.75-0.91) <.001 0.95 (0.86-1.05) .309 0.76 (0.68-0.85) <.001 0.96 (0.89-1.04) .303
Lifetime mental health diagnoses
 Depressive disorder (vs. no) 1.03 (0.91-1.16) .673 1.11 (0.99-1.26) .082 0.95 (0.83-1.10) .518 1.01 (0.92-1.12) .780
 Anxiety disorder (vs. no) 1.01 (0.89-1.13) .911 1.25 (1.10-1.41) <.001 0.88 (0.77-1.02) .097 0.96 (0.87-1.06) .450
 ADD (vs. no) 1.01 (0.89-1.14) .907 1.18 (1.05-1.33) .006 1.12 (0.97-1.29) .123 1.00 (0.90-1.10) .960
Personality characteristics
 Extraversion 1.02 (0.99-1.05) .239 1.06 (1.02-1.09) .001 1.02 (0.98-1.06) .306 1.00 (0.97-1.02) .728
 Agreeableness 0.81 (0.78-0.85) <.001 0.94 (0.90-0.98) .005 0.79 (0.75-0.83) <.001 0.85 (0.82-0.88) <.001
 Conscientiousness 0.84 (0.81-0.87) <.001 0.95 (0.91-0.98) .005 0.88 (0.84-0.92) <.001 0.91 (0.88-0.93) <.001
 Emotional stability 0.93 (0.89-0.97) <.001 1.03 (0.99-1.07) .183 0.97 (0.92-1.01) .177 0.94 (0.91-0.97) <.001
 Openness to new experiences 0.94 (0.90-0.98) .004 1.04 (0.99-1.08) .089 0.91 (0.87-0.96) <.001 0.99 (0.96-1.02) .504
Open in a new tab

Notes: * Prior cannabis use defined as ever having used cannabis by the age one year earlier (time-lagged predictor). aHR: Adjusted Hazard Ratio adjusted for all the other predictors or covariates in the table.

3.6. Moderating effect of age-periods

Initiating cannabis use based on prior use of other substances.

Figure 2a shows hazards of cannabis initiation by previous use of each substance. The PH assumption was not supported for any substance (LRT p’s <.05, see Table 4 footnote). LRT suggested that the relationships between prior use of any substance and later hazards of cannabis initiation differed between the 5-18 vs. >18 age-periods (LRT p’s<.01, Table 4 footnote).

Figure 2. Hazard of initiating cannabis use, based on prior use* of each substance, and of initiating each substance, based on prior cannabis use*.

Figure 2.

Open in a new tab

Notes: * Prior use defined as ever having used the substance by the age one year earlier (time-lagged predictor). Covariates in each model included: age, sex, sexual orientation, race, ethnicity, parental education, rurality, state non-medical cannabis law, lifetime mental health diagnosis including depressive disorder, anxiety disorder, and ADD, and personality characteristics including extraversion, agreeableness, conscientiousness, emotional stability, and openness to new experiences.

Table 4. Discrete-time survival analysis (DTSA) assessing initiation of cannabis use, based on prior use* of each substance, and assessing initiation of each substance, based on prior cannabis use*, adjusting for sociodemographic, state non-medical cannabis laws, and psychosocial variables, among 5-18 years old and 19-34 years old, separately.
Prior use* of each substance in relation to initiating cannabis use
Age-period Cigarettes E-cigarettes Cigars Alcohol
aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p
5-18 years old 5.41 (4.69-6.25) <.001 3.33 (2.61-4.25) <.001 6.22 (4.73-8.17) <.001 3.79 (3.39-4.25) <.001
19-34 years old 2.40 (2.03-2.84) <.001 1.67 (1.36-2.05) <.001 2.11 (1.68-2.64) <.001 2.51 (2.14-2.93) <.001
Prior use* of cannabis in relation to initiating use of each substance
Age-period Cigarettes E-cigarettes Cigars Alcohol
aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p aHR (95% CI) p
5-18 years old 4.10 (3.54-4.75) <.001 4.08 (3.45-4.82) <.001 4.30 (3.64-5.07) <.001 4.50 (3.91-5.19) <.001
19-34 years old 2.37 (1.95-2.88) <.001 3.60 (3.11-4.17) <.001 2.33 (1.91-2.84) <.001 1.51 (1.24-1.83) <.001
Open in a new tab

Notes: * Prior use defined as ever having used the substance by the age-period one year earlier (time-lagged predictor). Cigarette, e-cigarette, cigar, and alcohol use by the prior age-period (t–1) in relation to cannabis initiation in a specific age-period (t): LRTs on PH assumption across all time periods: LR=137.05, df=24, p<.001; LR=33.70, df=24, p=.014; LR=68.44, df=21, p<.001; LR=47.96, df=21, p<.001. LRTs on PH assumption across the 2 age-period categories (>18 vs. 5-18): cigarette, e-cigarette, cigar, and alcohol use by the prior age-period (t–1) in relation to cannabis initiation in a specific age-period (t): LR=76.05, df=1, p<.001; LR=17.26, df=1, p<.001; LR=57.78, df=21, p<.001; LR=9.53, df=21, p=.002. Cannabis use by the prior age-period (t–1) in relation to initiating cigarette, e-cigarette, cigar and alcohol use in a specific age-period (t): LRTs on PH assumption across all the time periods: LR=92.73, df=21, p<.001; LR=184.37, df=22, p<.001; LR=126.52, df=21, p<.001; LR=151.14, df=20, p<.001. LRTs on PH assumption across the 2 age-period categories (>18 vs. 5-18): cigarette, e-cigarette, cigar, and alcohol use by prior age-period (t–1) in relation to cannabis initiation in a specific age-period (t): LR=19.84, df=1, p<.001; LR=0.70, df=1, p=.404; LR=30.31, df=1, p<.001; LR=90.40, df=1, p<.001. Covariates in each model included: age, sex, sexual orientation, race, ethnicity, parental education, rurality, state non-medical cannabis laws, lifetime mental health diagnosis including depressive disorder, anxiety disorder, and ADD, and personality characteristics including extraversion, agreeableness, conscientiousness, emotional stability, and openness to new experiences (not shown in table).

Stratified analyses by age-period (>18 vs. 5-18, Table 4) showed that when participants were ages 5-18, HRs of cannabis initiation, based on prior use of each substance, were 5.41 (95%CI=4.69-6.25) for cigarettes, 3.33 (95%CI=2.61-4.25) for e-cigarettes, 6.22 (95%CI=4.73-8.17) for cigars, and 3.79 (95%CI=3.39-4.25) for alcohol. When participants were ages 19-34, HRs were lower: 2.40 (95%CI=2.03-2.84) for cigarettes, 1.67 (95%CI=1.36-2.05) for e-cigarettes, 2.11 (95%CI=1.68-2.64) for cigars, and 2.51 (95%CI=2.14-2.93) for alcohol, indicating that the relationships between prior substance use and hazards of cannabis initiation were weaker among those >18 vs. 5-18.

Initiating other substances based on prior cannabis use.

Figure 2b shows hazards of initiating each substance based on prior cannabis use. LRT showed no support for the PH assumption for any substance (LRT p’s<.001) and suggested that the relationships between prior cannabis use and later hazards of initiating cigarettes, cigars, and alcohol differed between the 5-18 vs. >18 age-periods (LRT p’s<.001), but not e-cigarettes (p=.404).

Stratified analyses by age-period (>18 vs. 5-18, Table 4) showed that when participants were ages 5-18, based on prior use of cannabis, HRs of initiation were 4.10 (95%CI=3.54-4.75) for cigarettes, 4.30 (95%CI=3.64-5.07) for cigars, and 4.50 (95%CI=3.91-5.19) for alcohol. When participants were ages 19-34, HRs were lower: 2.37 (95%CI=1.95-2.88) for cigarettes, 2.33 (95%CI=1.91-2.84) for cigars, and 1.51 (95%CI=1.24-1.83) for alcohol. No significant difference existed for e-cigarette initiation between those >18 vs. 5-18.

4. DISCUSSION

In this study, prior cigarette, e-cigarette, cigar, and alcohol use had significant effects on subsequent cannabis initiation and vice versa. Furthermore, a larger proportion (79%) of participants followed the gateway sequence (i.e., used some other substance before cannabis) vs. the reverse gateway sequence (21%), consistent with a study documenting that the reverse gateway sequence held among a minority of US adults (28%) and US adults ages 18-29 (33%) (Degenhardt et al., 2010). While this prior study did not report sequences for different specific substances (e.g., cigarettes, e-cigarettes, cigars, alcohol), the current study found that the gateway sequence was particularly applicable to alcohol and cannabis (i.e., over two-thirds used alcohol before cannabis), aligning with prior research (Barry et al., 2016; Keyes et al., 2015; Lemyre et al., 2019; Reed et al., 2022), while the reverse gateway sequence predominantly applied to cannabis, e-cigarette, and cigar use (i.e., >75% used cannabis before e-cigarettes or cigars, respectively). Moreover, effects were generally stronger when initiation occurred during adolescence vs. young adulthood. Collectively, these findings align with both the gateway hypothesis (Kandel, 2003; Kandel et al., 2006) and common liability model of vulnerability (Mayet et al., 2016; Vanyukov et al., 2012), underscoring the importance of prevention efforts addressing underlying mechanisms of various substance use (Kim & Hodgins, 2018), particularly for adolescents.

The more pronounced effects of prior cannabis use on subsequent e-cigarette and cigar use (vs. in the opposite direction) are noteworthy given the mixed findings from the limited research assessing gateway effects of e-cigarettes (Jorgensen & Wells, 2022; Morean et al., 2015; Temple et al., 2017; Unger et al., 2016) and cigars (Cohn et al., 2018; Cornacchione Ross et al., 2020) to cannabis use, and evidence that cannabis use may precede the use and/or increase the probability of subsequently using tobacco and alcohol (Agrawal et al., 2006; Badiani et al., 2015; Cohn et al., 2018; Kokkevi et al., 2006; Mayet et al., 2016; Patton et al., 2005; Ramo et al., 2012; Reed et al., 2022; Scholes-Balog et al., 2016; Secades-Villa et al., 2015). These results should be considered within the context of the cannabis, e-cigarette, and cigar markets in the US, given the expansion of non-medical cannabis legalization since 2014 (National Conference of State Legislators, 2024), the introduction of e-cigarettes in the US around 2010, and the increases in sales of e-cigarettes (Ali et al., 2022), cigars (Wang et al., 2021), and cannabis (Dilley et al., 2023). Since 2010, the use prevalence of both e-cigarettes and cannabis has increased among US young adults (Substance Abuse and Mental Health Services Administration, 2024), and cigar use, particularly small cigar use, has increased among certain populations (Jensen et al., 2025). Furthermore, cannabis is often used via vaping or blunts (Rubenstein et al., 2024; Simpson et al., 2021). Current results may indicate that initiating cannabis also introduces individuals to these other modes of use and promotes tobacco/nicotine use in these forms.

As noted above, the associations between prior use of other substances and initiation of each substance were generally stronger when initiation occurred during adolescence vs. young adulthood. However, this did not hold true for the relationship between prior e-cigarette use and later cannabis initiation. This is difficult to interpret, but it may be because initiating other substances was most commonly reported during the 13-18 age range, while e-cigarette initiation was most commonly reported during the 19-25 age range (45%) – likely due to when many participants were first exposed to e-cigarettes (Ali et al., 2022; Kaplan et al., 2023). Nonetheless, this finding warrants replication and further exploration regarding the mechanisms that might account for this finding.

As suggested by the common liability model of vulnerability (Mayet et al., 2016; Vanyukov et al., 2012), lifetime mental health diagnoses and personality characteristics were related to initiation in several of the models (Dash et al., 2023; Hokm Abadi et al., 2018; Kotov et al., 2010; Moreno-Mansilla et al., 2021). Specifically, cannabis initiation was predicted by lifetime depression diagnosis in the models assessing effects of previously using each other substance, and by lifetime anxiety diagnosis in the model accounting for prior cigarette use. Meanwhile, lifetime anxiety disorder or ADD diagnosis predicted e-cigarette initiation only, after accounting for prior cannabis use. Furthermore, certain personality characteristics were protective against substance initiation, including agreeableness and conscientiousness for each substance, openness to new experiences for cigarettes and cigars, and emotional stability for cannabis, cigarettes, and cigars. Meanwhile, extraversion was associated with increased hazards of initiating cannabis and e-cigarettes. These findings largely align with prior research highlighting certain personality risk profileds (Dash et al., 2023; Hokm Abadi et al., 2018; Kotov et al., 2010) that might help target those in need of intervention.

Notably, state non-medical cannabis laws were not associated with cannabis, e-cigarette, or alcohol initiation, but were with lower hazards of initiating cigarettes and cigars, which may underscore the general availability and appeal of cannabis, e-cigarettes, and alcohol (relative to cigarettes and cigars), regardless of regulatory context (Berg et al., 2015). Findings regarding sociodemographics also warrant consideration. Besides being younger, other groups at higher risk for substance use initiation in certain models included those identifying as non-Hispanic White (vs. Hispanic, Black, and Asian), in rural areas, and with parents <bachelor’s degree educated. Additionally, males were more likely to initiate e-cigarettes and cigars, whereas females were more likely to initiate cannabis and alcohol. Finally, sexual minorities were less likely to initiate cigars but more likely alcohol and cannabis. These findings may reflect different substance use contexts across different populations (Pokhrel et al., 2015) and the need to better understand mechanisms contributing to disparities in substance use.

This study has limitations. First, generalizability is limited due to participant recruitment via social media and purposive sampling (~50% reporting past-month cannabis use). Second, self-reported measures and retrospective reports (e.g., age of initiation) introduce potential bias. Third, measures included lifetime mental health diagnoses but not when diagnoses occurred, precluding the ability to account for sequencing of mental health diagnosis and substance use. Similarly, state non-medical cannabis legalization at the time of W1 assessment (but not prior periods or residences) was assessed; thus, we could not account for when individuals were exposed to legal non-medical cannabis. Along these lines, cohort effects may have impacted findings (e.g., some participants may not have been exposed to e-cigarettes during adolescence, some may not have been exposed to cannabis retail (Fadus et al., 2019)). Finally, the strength of the gateway effect vs. reverse gateway effect was not compared statistically.

Findings have implications for research and practice. Results generally supported gateway effects in both directions (tobacco/alcohol to cannabis, vice versa) as well as common liability theory, suggesting the need for preventive efforts targeting initiation of multiple substances in various sequences and that address common mechanisms, such as psychological (e.g., mental health, neuroticism, disagreeableness) (Moreno-Mansilla et al., 2021; Zuckerman, 2007) and socio-environmental factors (e.g., social norms) (Windle et al., 2017), especially early in life (e.g., in adolescence) (Melberg et al., 2010). Certain tobacco/nicotine products, particularly those used with cannabis or via similar modes as cannabis (e.g., e-cigarettes, cigars/blunts), may be especially mixed in use sequence (Kim & Hodgins, 2018). Future research should consider replicating these findings and apply this analytic approach to assessing other substance use patterns (e.g., derived cannabis products, hallucinogens) given the ongoing substance-related regulatory and market changes (Wojciechowski, 2024).

  • Prior tobacco/alcohol initiation increased hazard of cannabis initiation

  • Prior cannabis initiation increased hazard of tobacco/alcohol initiation

  • Relative strength between the two directions varied by the specific substance

  • Most of these effects were stronger for initiation during childhood/adolescence vs. young adulthood

  • Interventions should target polysubstance use initiation, especially in adolescence

Funding sources:

This work was supported by the National Institute on Drug Abuse (R01DA054751, MPIs: Berg, Cavazos-Rehg). Dr. Wang is also supported by other US federal funding (R01DA059840, R01MD019033, R01CA275066, 2U54MD007600-36, 1UG3MD018296-01, R01DA054751, 5U01DP006639, R01HD105356, P30AI117970). Dr. Berg is also supported by other US NIH funding (R01CA215155, PI: Berg; R01CA239178, MPIs: Berg, Levine; R01CA278229, MPIs: Berg, Kegler; R01CA275066, MPIs: Yang, Berg; R21CA261884, MPIs: Berg, Arem), Fogarty International Center (R01TW010664, MPIs: Berg, Kegler; D43TW012456, MPIs: Berg, Paichadze, Petrosyan), and the National Institute of Environmental Health Sciences/Fogarty (D43ES030927, MPIs: Berg, Caudle, Sturua). The funder had no role in study design or conduct; data collection, management, analysis, and interpretation; manuscript preparation, review, or approval; and decision to publish.

Footnotes

Author Agreement

All authors have seen and approved the final version of the manuscript being submitted.

Conflict of interest disclosure. The authors report no conflicting interests.

Ethics approval. This study was approved by the George Washington University Institutional Review Board (NCR224124). All participants provided informed consent.

Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Data availability.

The dataset used is available from the corresponding author on reasonable request.

REFERENCES

  1. Agrawal A, Grant JD, Waldron M, Duncan AE, Scherrer JF, Lynskey MT, Madden PA, Bucholz KK, & Heath AC (2006). Risk for initiation of substance use as a function of age of onset of cigarette, alcohol and cannabis use: findings in a Midwestern female twin cohort. Preventive Medicine, 43(2), 125–128. [DOI] [PubMed] [Google Scholar]
  2. Ali FRM, Seaman EL, Crane E, Schillo B, & King BA (2022). Trends in US E-cigarette Sales and Prices by Nicotine Strength, Overall and by Product and Flavor Type, 2017–2022. Nicotine & Tobacco Research, 25(5), 1052–1056. 10.1093/ntr/ntac284 [DOI] [PMC free article] [PubMed] [Google Scholar]
  3. Badiani A, Boden JM, De Pirro S, Fergusson DM, Horwood LJ, & Harold GT (2015). Tobacco smoking and cannabis use in a longitudinal birth cohort: evidence of reciprocal causal relationships. Drug and Alcohol Dependence, 150, 69–76. [DOI] [PubMed] [Google Scholar]
  4. Barry AE, King J, Sears C, Harville C, Bondoc I, & Joseph K (2016). Prioritizing alcohol prevention: Establishing alcohol as the gateway drug and linking age of first drink with illicit drug use. Journal of School Health, 86(1), 31–38. [DOI] [PubMed] [Google Scholar]
  5. Berg CJ, Romm KF, Pannell A, Sridharan P, Sapra T, Rajamahanty A, Cui Y, Wang Y, Yang YT, & Cavazos-Rehg PA (2023). Cannabis retailer marketing strategies and regulatory compliance: A surveillance study of retailers in 5 US cities. Addictive Behaviors, 143, 107696. 10.1016/j.addbeh.2023.107696 [DOI] [PMC free article] [PubMed] [Google Scholar]
  6. Berg CJ, Stratton E, Schauer GL, Lewis M, Wang Y, Windle M, & Kegler M (2015). Perceived harm, addictiveness, and social acceptability of tobacco products and marijuana among young adults: marijuana, hookah, and electronic cigarettes win. Substand Use & Misuse, 50(1), 79–89. 10.3109/10826084.2014.958857 [DOI] [PMC free article] [PubMed] [Google Scholar]
  7. Cohn AM, Johnson AL, Fryer CS, & Villanti AC (2018). Marijuana use predicts onset of current little cigar use in a national sample of US young adults. Drug and Alcohol Dependence, 190, 235–241. 10.1016/j.drugalcdep.2018.01.020 [DOI] [PMC free article] [PubMed] [Google Scholar]
  8. Cornacchione Ross J, Sutfin EL, Suerken C, Walker S, Wolfson M, & Reboussin BA (2020). Longitudinal associations between marijuana and cigar use in young adults. Drug and Alcohol Dependence, 211, 107964. 10.1016/j.drugalcdep.2020.107964 [DOI] [PMC free article] [PubMed] [Google Scholar]
  9. Dash GF, Martin NG, & Slutske WS (2023). Big Five personality traits and illicit drug use: Specificity in trait–drug associations. Psychology of Addictive Behaviors, 37(2), 318. [DOI] [PMC free article] [PubMed] [Google Scholar]
  10. Dave D, Liang Y, Pesko MF, Phillips S, & Sabia JJ (2023). Have recreational marijuana laws undermined public health progress on adult tobacco use? Journal of Health Economics, 90, 102756. [DOI] [PubMed] [Google Scholar]
  11. De PK, & Sun R (2025). The effects of recreational cannabis laws on alcohol and tobacco use among US adults, 2012 to 2022. American Journal of Preventive Medicine. [DOI] [PubMed] [Google Scholar]
  12. Degenhardt L, Dierker L, Chiu WT, Medina-Mora ME, Neumark Y, Sampson N, Alonso J, Angermeyer M, Anthony JC, & Bruffaerts R (2010). Evaluating the drug use “gateway” theory using cross-national data: consistency and associations of the order of initiation of drug use among participants in the WHO World Mental Health Surveys. Drug and Alcohol Dependence, 108(1-2), 84–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
  13. Dilley JA, Johnson JK, Colby AM, Sheehy TJ, Muse EJ, Filley JR, Segawa MB, Schauer GL, & Kilmer B (2023). Cannabis retail market indicators in five legal states in the United States: a public health perspective. Clinical Therapeutics, 45(8), 778–786. [DOI] [PMC free article] [PubMed] [Google Scholar]
  14. Eaton NR, Rodriguez-Seijas C, Carragher N, & Krueger RF (2015). Transdiagnostic factors of psychopathology and substance use disorders: a review. Social Psychiatry and Psychiatric Epidemiology, 50, 171–182. [DOI] [PubMed] [Google Scholar]
  15. Fadus MC, Smith TT, & Squeglia LM (2019). The rise of e-cigarettes, pod mod devices, and JUUL among youth: Factors influencing use, health implications, and downstream effects. Drug and Alcohol Dependence, 201, 85–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
  16. Fergusson DM, Boden JM, & Horwood LJ (2006). Cannabis use and other illicit drug use: testing the cannabis gateway hypothesis. Addiction, 101(4), 556–569. [DOI] [PubMed] [Google Scholar]
  17. Gosling SD, Rentfrow PJ, & Swann WB (2003). A very brief measure of the Big-Five personality domains. Journal of Research in Personality, 37(6), 504–528. 10.1016/S0092-6566(03)00046-1 [DOI] [Google Scholar]
  18. Hokm Abadi ME, Bakhti M, Nazemi M, Sedighi S, & Mirzadeh Toroghi E (2018). The relationship between personality traits and drug type among Substance Abuse. Journal of Research and Health, 8(6), 531–540. [Google Scholar]
  19. Jensen JK, Ganz O, Tomaino M, Glasser AM, Sterling K, Delnevo CD, & Bover Manderski MT (2025). Patterns of blunt and cigar use in the United States, 2015–2019. Nicotine and Tobacco Research, 27(3), 511–516. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Jorgensen C, & Wells J (2022). Is marijuana really a gateway drug? A nationally representative test of the marijuana gateway hypothesis using a propensity score matching design. Journal of Experimental Criminology, 18(3), 497–514. [Google Scholar]
  21. Kandel DB (2003). Does marijuana use cause the use of other drugs? JAMA, 289(4), 482–483. [DOI] [PubMed] [Google Scholar]
  22. Kandel DB, Davies M, Karus D, & Yamaguchi K (1986). The Consequences in Young Adulthood of Adolescent Drug Involvement: An Overvie w. Archives of General Psychiatry, 43(8), 746–754. [DOI] [PubMed] [Google Scholar]
  23. Kandel DB, Yamaguchi K, & Klein LC (2006). Testing the gateway hypothesis. Addiction, 101(4). [DOI] [PubMed] [Google Scholar]
  24. Kaplan B, Hardesty JJ, Welding K, Breland AB, Eissenberg T, & Cohen JE (2023). Electronic Nicotine Delivery System flavor use over time by age group in the US: A longitudinal analysis. Tobacco Induced Diseases, 21, 67. 10.18332/tid/162365 [DOI] [PMC free article] [PubMed] [Google Scholar]
  25. Keyes KM, Vo T, Wall MM, Caetano R, Suglia SF, Martins SS, Galea S, & Hasin D (2015). Racial/ethnic differences in use of alcohol, tobacco, and marijuana: Is there a cross-over from adolescence to adulthood? Social Science & Medicine, 124, 132–141. 10.1016/j.socscimed.2014.11.035 [DOI] [PMC free article] [PubMed] [Google Scholar]
  26. Kim HS, & Hodgins DC (2018). Component model of addiction treatment: A pragmatic transdiagnostic treatment model of behavioral and substance addictions. Frontiers in Psychiatry, 9, 406. [DOI] [PMC free article] [PubMed] [Google Scholar]
  27. Kokkevi A, Gabhainn SN, Spyropoulou M, & HBSC RBFG o. t. (2006). Early initiation of cannabis use: a cross-national European perspective. Journal of Adolescent Health, 39(5), 712–719. [DOI] [PubMed] [Google Scholar]
  28. Kotov R, Gamez W, Schmidt F, & Watson D (2010). Linking “big” personality traits to anxiety, depressive, and substance use disorders: a meta-analysis. Psychological Bulletin, 136(5), 768. [DOI] [PubMed] [Google Scholar]
  29. Lemyre A, Poliakova N, & Bélanger RE (2019). The relationship between tobacco and cannabis use: a review. Substance Use & Misuse, 54(1), 130–145. [DOI] [PubMed] [Google Scholar]
  30. Lessem JM, Hopfer CJ, Haberstick BC, Timberlake D, Ehringer MA, Smolen A, & Hewitt JK (2006). Relationship between adolescent marijuana use and young adult illicit drug use. Behavior Genetics, 36, 498–506. [DOI] [PubMed] [Google Scholar]
  31. Mayet A, Legleye S, Beck F, Falissard B, & Chau N (2016). The gateway hypothesis, common liability to addictions or the route of administration model? A modelling process linking the three theories. European addiction research, 22(2), 107–117. [DOI] [PubMed] [Google Scholar]
  32. Melberg HO, Jones AM, & Bretteville-Jensen AL (2010). Is cannabis a gateway to hard drugs? Empirical Economics, 38, 583–603. [Google Scholar]
  33. Morean ME, Kong G, Camenga DR, Cavallo DA, & Krishnan-Sarin S (2015). High School Students' Use of Electronic Cigarettes to Vaporize Cannabis. Pediatrics, 136(4), 611–616. 10.1542/peds.2015-1727 [DOI] [PMC free article] [PubMed] [Google Scholar]
  34. Moreno-Mansilla S, Ricarte JJ, & Hallford DJ (2021). Cannabis use among early adolescents and transdiagnostic mental health risk factors. Clinical Child Psychology and Psychiatry, 26(2), 531–543. [DOI] [PubMed] [Google Scholar]
  35. National Conference of State Legislators. (2024). State regulated cannabis programs: https://www.ncsl.org/health/state-medical-cannabis-laws.
  36. Patton GC, Coffey C, Carlin JB, Sawyer SM, & Lynskey M (2005). Reverse gateways? Frequent cannabis use as a predictor of tobacco initiation and nicotine dependence. Addiction, 100(10), 1518–1525. [DOI] [PubMed] [Google Scholar]
  37. Pokhrel P, Herzog TA, Muranaka N, & Fagan P (2015). Young adult e-cigarette users’ reasons for liking and not liking e-cigarettes: a qualitative study. Psychology & Health, 30(12), 1450–1469. [DOI] [PMC free article] [PubMed] [Google Scholar]
  38. Ramo DE, Liu H, & Prochaska JJ (2012). Tobacco and marijuana use among adolescents and young adults: a systematic review of their co-use. Clinical Psychology Review, 32(2), 105–121. [DOI] [PMC free article] [PubMed] [Google Scholar]
  39. Reed ZE, Wootton RE, & Munafò MR (2022). Using Mendelian randomization to explore the gateway hypothesis: possible causal effects of smoking initiation and alcohol consumption on substance use outcomes. Addiction, 117(3), 741–750. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Rubenstein D, McClernon FJ, & Pacek LR (2024). Trends in cannabis and tobacco co-use in the United States, 2002–2021. Addictive Behaviors, 158, 108129. [DOI] [PMC free article] [PubMed] [Google Scholar]
  41. Scholes-Balog KE, Hemphill SA, Evans-Whipp TJ, Toumbourou JW, & Patton GC (2016). Developmental trajectories of adolescent cannabis use and their relationship to young adult social and behavioural adjustment: A longitudinal study of Australian youth. Addictive Behaviors, 53, 11–18. [DOI] [PubMed] [Google Scholar]
  42. Secades-Villa R, Garcia-Rodríguez O, Jin CJ, Wang S, & Blanco C (2015). Probability and predictors of the cannabis gateway effect: a national study. International Journal of Drug Policy, 26(2), 135–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
  43. Selya A (2024). The “Gateway” hypothesis: evaluation of evidence and alternative explanations. Harm Reduction Journal, 21(1), 113. 10.1186/s12954-024-01034-6 [DOI] [PMC free article] [PubMed] [Google Scholar]
  44. Simpson KA, Cho J, & Barrington-Trimis JL (2021). The association of type of cannabis product used and frequency of use with problematic cannabis use in a sample of young adult cannabis users. Drug and Alcohol Dependence, 226, 108865. 10.1016/j.drugalcdep.2021.108865 [DOI] [PMC free article] [PubMed] [Google Scholar]
  45. Singer JD, & Willett JB (2003). Applied longitudinal data analysis: Modeling change and event occurrence. Oxford university press. [Google Scholar]
  46. Substance Abuse and Mental Health Services Administration. (2024). Key substance use and mental health indicators in the United States: Results from the 2023 National Survey on Drug Use and Health. Rockville, MD: Center for Behavioral Health Statistics and Quality, Substance Abuse and Mental Health Services Administration. Retrieved from: https://www.samhsa.gov/data/release/2023-national-survey-drug-use-and-health-nsduh-releases#detailed-tables. [Google Scholar]
  47. Temourian AA, Halliday DM, Yan Y, Chan-Golston AM, & Song AV (2024). Marijuana and E-cigarette Initiation Among Adolescents: A Survival Analysis. Journal of Adolescent Health, 74(4), 747–754. 10.1016/j.jadohealth.2023.11.009 [DOI] [PubMed] [Google Scholar]
  48. Temple JR, Shorey RC, Lu Y, Torres E, Stuart GL, & Le VD (2017). E-cigarette use of young adults motivations and associations with combustible cigarette alcohol, marijuana, and other illicit drugs. Am J Addict, 26(4), 343–348. 10.1111/ajad.12530 [DOI] [PMC free article] [PubMed] [Google Scholar]
  49. Unger JB, Soto DW, & Leventhal A (2016). E-cigarette use and subsequent cigarette and marijuana use among Hispanic young adults. Drug and Alcohol Dependence, 163, 261–264. 10.1016/j.drugalcdep.2016.04.027 [DOI] [PMC free article] [PubMed] [Google Scholar]
  50. Vanyukov MM, Tarter RE, Kirillova GP, Kirisci L, Reynolds MD, Kreek MJ, Conway KP, Maher BS, Iacono WG, & Bierut L (2012). Common liability to addiction and “gateway hypothesis”: theoretical, empirical and evolutionary perspective. Drug and Alcohol Dependence, 123, S3–S17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  51. Wang X, Kim Y, Borowiecki M, Tynan MA, Emery S, & King BA (2021). Trends in Cigar Sales and Prices, by Product and Flavor Type—the United States, 2016–2020. Nicotine & Tobacco Research, 24(4), 606–611. 10.1093/ntr/ntab238 [DOI] [PMC free article] [PubMed] [Google Scholar]
  52. Windle M, Haardorfer R, Lloyd SA, Foster B, & Berg CJ (2017). Social Influences on College Student Use of Tobacco Products, Alcohol, and Marijuana. Substance Use & Misuse, 52(9), 1111–1119. 10.1080/10826084.2017.1290116 [DOI] [PMC free article] [PubMed] [Google Scholar]
  53. Wojciechowski T (2024). Understanding marijuana use initiation vs. frequency of use on risk for hallucinogen use: testing for mediating mechanisms underlying the gateway hypothesis. Criminal Justice Studies, 37(2), 171–191. 10.1080/1478601X.2024.2317188 [DOI] [Google Scholar]
  54. Zuckerman M (2007). Sensation seeking and risky behavior (Vol. 72). American Psychological Association; Washington, DC. [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Data Availability Statement

The dataset used is available from the corresponding author on reasonable request.

RESOURCES