Longitudinal associations between susceptibility to tobacco use and the onset of other substances among U.S. youth

Abstract

We examined whether tobacco susceptibility at Wave (W) 1 (2013–2014) predicts the onset of tobacco and other substances at W2 (2014–2015) among 5325 U.S. youth (12–17 years) never substance users at W1 in the Population Assessment of Tobacco and Health (PATH) Study. Tobacco susceptibility was based on curiosity, use intentions, and response to a best friend’s offer to use. Onset of use included past 12-month use of a specific substance or group of substances at W2 among those who had never used any substance at W1. Approximately, 31.3% of W1 youth were susceptible to tobacco use. W2 onset was 8.2% (SE = 0.4) for alcohol exclusively, 5.0% (SE = 0.4) for polysubstance including tobacco, 4.4% (SE = 0.3) for tobacco exclusively, 3.1% (SE = 0.3) for other drugs (misused prescription stimulants and painkillers, cocaine, other stimulants, heroin, inhalants, solvents and hallucinogens) exclusively, 1.4% (SE = 0.2) for polysubstance excluding tobacco, and 0.9% (SE = 0.1) for marijuana exclusively. Tobacco-susceptible compared with non-tobacco susceptible youth had higher odds of onset of exclusive tobacco use (AOR: 2.4; 95% CI: 1.7, 3.3), exclusive alcohol use (AOR: 1.5; 95% CI: 1.2, 1.8), and polysubstance use (AOR: 3.9; 95% CI: 2.8, 5.6 including tobacco and AOR: 1.8; 95% CI: 1.1, 3.0 excluding tobacco) compared with W2 never substance use. In this national study, tobacco susceptibility identified U.S. youth at risk for onset of tobacco and other substances, perhaps reflecting common etiology and clustering of substance use in youth. Identifying and preventing tobacco-susceptible youth from progressing to using addictive substances must remain a public health priority.

1. Introduction

Substance (tobacco, alcohol, and drug) use among U.S. youth (12–17 years) remains a major public health problem. For most substance users, initial use occurs early in adolescence (National Center for Chronic Disease Prevention and Health Promotion and Office on Smoking and Health, 2012; U.S. Department of Health and Human Services (HHS) and Office of the Surgeon General, 2016). This is concerning as youth who try substances are more likely to progress to regular use, dependence, and disorders in adulthood (Hall et al., 2016). Furthermore, early initiation, substance use, and substance use disorders are associated with several immediate and long-term negative physical, mental, social, and economic consequences (National Center for Chronic Disease Prevention and Health Promotion and Office on Smoking and Health, 2012; U.S. Department of Health and Human Services (HHS) and Office of the Surgeon General, 2016). Although tremendous progress has been made in the development of evidence-based prevention programs and policies for youth substance use (National Center for Chronic Disease Prevention and Health Promotion and Office on Smoking and Health, 2012; U.S. Department of Health and Human Services (HHS) and Office of the Surgeon General, 2016), sustained efforts are still needed to identify and prevent use across all substances among youth at risk.

Pierce et al.’s susceptibility to smoking index is a validated 3-item instrument that predicts the risk of cigarette smoking initiation several years before first use (Pierce et al., 1996). This index assessed both intentions to smoke and self-efficacy about refusing a cigarette and was expanded to include curiosity (Nodora et al., 2014; Strong et al., 2015). Pierce’s index has also been found to predict risk for tobacco product use other than cigarettes (Pierce et al., 2018). Given the co-occurring use of tobacco, alcohol, and drugs (Conway et al., 2018; Richter et al., 2017; Silveira et al., 2018; Silveira et al., 2019) possibly due to shared vulnerability to use these substances (Vanyukov et al., 2012) and clustering of problem behaviors (Jessor, 1991), youth susceptible to tobacco product use can also be at risk for initiating other substances.

A recent national study on U.S. youth, found that those who were susceptible to e-cigarettes had greater odds of onset of e-cigarettes, alcohol, and marijuana use at follow up (Nicksic and Barnes, 2019). However, this study focused on e-cigarette susceptibility only, excluded drugs besides marijuana, and did not examine polysubstance use. To investigate whether susceptibility to tobacco use can potentially be used to identify youth at risk for initiating substances other than tobacco, we conducted a prospective study of susceptibility to tobacco use among never substance users as a longitudinal predictor of substance use onset across tobacco products, alcohol, marijuana, and other drugs (misused prescription stimulants and painkillers, cocaine and other stimulants, heroin, inhalants, solvents and hallucinogens) in the Population Assessment of Tobacco and Health (PATH) Study, an ongoing, nationally-representative, longitudinal study of adults and youth in the U.S. Consistent with a framework of common underlying risk for substance use, we hypothesized that susceptibility to tobacco use will predict the onset of substance use generally and across specific substances.

2. Methods

2.1. Study population

The current study included data from Wave (W) 1 and W2 of the PATH Study. Of the 13,651 youth (12 to 17 years) who completed the youth interview at W1, 11,996 completed the W2 interview (average follow-up period: 52 weeks), including 10,081 youth (12 to 17 years at W2) and 1915 aged-up adults (18+ years at W2) who completed the W2 youth and adult interviews, respectively. Analyses were further restricted to 5325 W1 never substance users (defined as self-reported never users of tobacco products, alcohol, marijuana, and other drugs) with data on variables of interest.

Details regarding the PATH Study design and methods are published elsewhere (Hyland et al., 2017) and available at https://doi.org/10.3886/ICPSR36231.v19. Briefly, a four-stage stratified address-based area-probability sampling design was used at W1 that oversampled adult tobacco users, young adults (18 to 24 years), and African-American adults. At W1, the weighted response rate for the household screener was 54.0%. Among households that were screened, the overall youth weighted response rate was 78.4%. Conditional on W1 participation, weighted response rates for the youth and adult interviews were 87.3% and 83.2% at W2. The W1 weighting procedures adjusted for differential probabilities of selection and nonresponse allowing estimates to be representative of the U.S. civilian, noninstitutionalized population. At W2, after adjusting for nonresponse, additional adjustments were made to protect against potential bias from attrition. Consent was obtained from parents, emancipated youth, and adults and assent was obtained from youth. Audio computer-assisted self-Inter-views (ACASI) available in English and Spanish were used to collect self-reported information on tobacco-use patterns and associated health behaviors at W1 and W2. This study was conducted by Westat and approved by Westat’s Institutional Review Board.

2.2. Measures

2.2.1. Susceptibility to tobacco use

Susceptibility to tobacco use was assessed using Pierce et al.’s 3-item measure of susceptibility (Pierce et al., 2018; Pierce et al., 2017; Trinidad et al., 2017). At W1, never users of 12 tobacco products (cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, snus pouches, smokeless tobacco excluding snus pouches, dissolvable tobacco, bidis and kreteks) were asked: “Have you ever been curious about using < product >?”; “Do you think you might try using < product > soon?”; and “If one of your best friends were to offer you < product >, would you use it?”. The four-level response options for items ranged from “not at all curious” to “very curious” or from “definitely not” to “definitely yes”. The index was internally consistent among W1 youth in the PATH Study (Cronbach’s α = 0.82). Consistent with prior research (Pierce et al., 2018; Pierce et al., 2017; Trinidad et al., 2017), participants with the strongest rejection (i.e. “not at all curious” and “definitely not”) to all three questions for all products and those who had never heard of a product were classified as not susceptible. All others were considered susceptible to use ≥1 products, including those with missing responses (participants who endorsed don’t know or refused) based on the approach by Trinidad and colleagues, because their responses suggest that they have not endorsed a strong commitment not to use a tobacco product (Trinidad et al., 2017).

2.2.2. Onset of substance use

At W1, participants were asked whether they had ever used the following tobacco products: cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, snus pouches, smokeless tobacco excluding snus pouches, dissolvable tobacco, bidis and kreteks. At W2, past 12-month use of the same tobacco products was assessed in the youth (12–17) and adult (18+) interviews; except bidis and kreteks that were not assessed in the adult interview. As data for these tobacco products were not available for youth who turned 18 years at W2, W2 tobacco use was defined as past 12-month use of any tobacco product except bidis and kreteks. Similarly, participants reported W1 ever use and W2 past 12-month use of alcohol, marijuana (including blunts), and other drugs (i.e. non-prescribed use of Ritalin® or Adderall®, painkillers, sedatives, or tranquilizers, and cocaine or crack, other stimulants (i.e. methamphetamine or speed), heroin, inhalants, solvents, and halluci-nogens). Items were adapted from the National Epidemiologic Survey on Alcohol and Related Conditions (National Institutes of Health (NIH), 2004–2005) and the National Health and Nutrition Examination Survey (Centers for Disease Control and Prevention (CDC), 2011–2012).

Four onset of use variables were created for tobacco, alcohol, marijuana, and other drugs and defined as past 12-month use of that specific substance or group of substances at W2 among never substance users at W1. Further, to examine differences by exclusive and polysubstance use, a 7-category onset of substance use variable was sub-sequently created as follows: (1) never use, (2) exclusive tobacco use, (3) exclusive alcohol use, (4) exclusive marijuana (including blunts) use, (5) exclusive other drug use, (6) polysubstance use including tobacco, and (7) polysubstance use excluding tobacco. Polysubstance use including tobacco was defined as any past 12-month use of two or more of the following: tobacco, alcohol, marijuana, and other drugs; whereas polysubstance use excluding tobacco was defined as any past 12-month use of two or more of the following: alcohol, marijuana, and other drugs.

2.2.3. Covariates

At W1, youth self-reported their sex, age, and race/ethnicity. Tobacco product user in the household (yes, no) was derived from youths’ responses regarding whether a household member used cigarettes, smokeless tobacco, such as chewing tobacco, snuff, dip, or snus, cigars, cigarillos, or filtered cigars, or any tobacco product. Additionally, youth were asked about the number of hours during the past seven days they were around others who were smoking including time at home, in a car, at school, or outdoors. Youth exposure to any smoking (home, car, at school, or outdoors) was categorized as yes vs. no. Academic performance was based on parents’ responses regarding their child’s grades at school in the past 12 months and categorized as at or above average (C grades and higher) and below average (below Cs). Emancipated youth self-reported their academic performance. Parents also reported their personal educational attainment and marital status.

Receptivity to tobacco product advertising, associated with onset of tobacco use (Pierce et al., 2018), was also assessed. Participants were asked to select the brand of their favorite tobacco advertisement from a list of 20 ads randomly sampled from a near-census collection (n = 959) of print, direct mail, and television ads used in the period immediately preceding the wave. For each ad shown, participants were asked if they had seen the ad in the past 12 months and whether they liked, disliked, or were neutral to the ad. Receptivity to ads was categorized as any receptivity (recalled an ad but did not like or have a favorite ad, liked or indicated a favorite ad, or liked and indicated a favorite ad) versus no receptivity (did not recall, like, or indicate a favorite ad) (Pierce et al., 2018; Pierce et al., 2017).

Sensation seeking, a risk factor for substance use (Hoyle et al., 2002), was assessed via three modified items from the Brief Sensation Seeking Scale. Response options for each item were summed to create a mean score (Conway et al., 2018). The scale was internally consistent among W1 youth in the PATH Study (Cronbach’s α = 0.76). Lifetime mental health problems were assessed via the internalizing and externalizing subscales of the Global Appraisal of Individual Needs – Short Screener (GAIN-SS) (Dennis et al., 2006), modified for the PATH Study and categorized into no/low (0–2 symptoms), moderate (3–7 symptoms), or high (8–11 symptoms) severity based on prior research (Conway et al., 2018).

2.3. Statistical analysis

Distributions of susceptibility to tobacco use, onset of substance use, and covariates were examined. Sex, age, and race/ethnicity were imputed as described in the PATH Study Restricted Use Files User Guide (United States Department of Health and Human Services et al., 2020). Logistic regression was used to compute odds ratios (OR) and 95% confidence intervals (CI) separately for five onset of use outcomes: tobacco (yes, no), alcohol (yes, no), marijuana (yes, no), other drugs (yes, no), and the seven-category substance use outcome (never use, exclusive tobacco use, exclusive alcohol use, exclusive marijuana (including blunts) use, exclusive other drug use, polysubstance use including tobacco, and polysubstance use excluding tobacco). Covariates included factors previously found to be associated with substance use such as sociodemographic factors (age, sex, and race/ethnicity (Stone et al., 2012; Wellman et al., 2016)), parent’s marital status and educational attainment (Stone et al., 2012; Wellman et al., 2016), academic performance (National Center for Chronic Disease Prevention and Health Promotion and Office on Smoking and Health, 2012; Stone et al., 2012; Tomczyk et al., 2016; Wellman et al., 2016), sensation seeking (Hoyle et al., 2002; Wellman et al., 2016), exposure to tobacco use (tobacco product use in household (Wellman et al., 2016), exposure to any smoking (Wellman et al., 2016), receptivity to tobacco product advertising (National Center for Chronic Disease Prevention and Health Promotion and Office on Smoking and Health, 2012; Wellman et al., 2016)), and lifetime mental health problems at Wave 1 (Green et al., 2018; Stone et al., 2012). Parent’s marital status and educational attainment and youth’s exposure to any smoking were not significant in the adjusted models and therefore excluded from the final models. All covariates were analyzed as categorical variables except for sensation seeking which was analyzed continuously such that the ORs re-presented a one-unit increase in the mean score.

Estimates were weighted to represent the U.S. youth population; variances and CIs were estimated using the balanced repeated replication method (McCarthy, 1969) with Fay’s adjustment set to 0.3 to increase estimate stability (Judkins, 1990). W2 longitudinal weights were used in the analyses (United States Department of Health and Human Services et al., 2020). Estimates were flagged for low statistical precision if they were based on a denominator sample size of fewer than 50, or the coefficient of variation of the estimate or its complement was larger than 30% (Klein et al., 2002). All analyses were conducted on the Restricted use Files (United States Department of Health and Human Services et al., 2020) using SAS Survey Procedures, version 9.4 (SAS Institute Inc., Cary, NC).

3. Results

Demographic characteristics are presented in Table 1. Approximately one third (31.3%) of youth never substance users at W1 were susceptible to tobacco use. In unadjusted models, higher sensation seeking, older age, non-Hispanic Black and Hispanic race/ethnicity were positively associated with susceptibility to tobacco use. Female youth had lower odds of susceptibility to tobacco use compared to male youth. In addition, youth who had unmarried parents, had moderate and high severity lifetime mental health problems, had below average grades, lived with a tobacco product user, were exposed to any smoking, or had higher receptivity to tobacco advertising had higher odds of susceptibility to tobacco use in W1 compared to their counterparts.

Table 1.

Selected characteristics by susceptibility to tobacco use among W1 youth never substance users (n = 5325).

	Wave 1 susceptibility to tobacco use
	Total			No			Yes
	na	%b	SEb	na	%b	SEb	na	%b	SEb	ORc	95% CIc
Total				3616	68.7	0.6	1709	31.3	0.6
Sensation seekingd	N/A			N/A			N/A			2.1	1.9	2.3
Age (years)
12–14	3402	63.2	0.7	2392	71.1	0.8	1010	28.9	0.8		Referent
15–17	1923	36.8	0.7	1224	64.6	1.1	699	35.4	1.1	1.3	1.2	1.6
Gender
Male	2803	52.5	0.4	1862	67.0	1.0	941	33.0	1.0		Referent
Female	2522	47.5	0.4	1754	70.6	1.0	768	29.4	1.0	0.8	0.7	1.0
Race/ethnicity
Non-Hispanic, White	2423	51.5	0.6	1758	72.1	0.9	665	27.9	0.9		Referent
Non-Hispanic, Black	786	14.9	0.4	491	62.6	1.7	295	37.4	1.7	1.5	1.3	1.8
Non-Hispanic, othere	463	9.5	0.3	308	70.9	2.2	155	29.1	2.2	1.1	0.8	1.4
Hispanic	1653	24.1	0.4	1059	64.3	1.2	594	35.7	1.2	1.4	1.3	1.6
Parent educational attainment
Up to 8th grade	398	5.9	0.4	272	69.2	2.4	126	30.8	2.4		Referent
HS, no diploma	785	13.1	0.6	529	68.7	1.4	256	31.3	1.4	1.0	0.8	1.3
HSD or GED	970	17.6	0.7	629	65.8	1.6	341	34.2	1.6	1.2	0.9	1.5
Some college or more	3155	63.4	1.1	2174	69.4	0.8	981	30.6	0.8	1.0	0.8	1.3
Parent current marital status
Married	3469	65.5	0.9	2435	71.1	0.7	1034	28.9	0.7		Referent
Unmarried	1848	34.5	0.9	1178	64.4	1.1	670	35.6	1.1	1.4	1.2	1.5
Lifetime mental health problemsf
No/low severity	1774	33.5	0.9	1433	81.2	1.0	341	18.8	1.0		Referent
Moderate severity	2320	43.2	0.7	1560	68.2	1.0	760	31.8	1.0	2.0	1.7	2.4
High severity	1231	23.3	0.7	623	51.7	1.4	608	48.3	1.4	4.0	3.4	4.8
Academic performanceg
At or above average	4817	90.9	0.5	3321	69.6	0.6	1496	30.4	0.6		Referent
Below average	508	9.1	0.5	295	59.6	2.7	213	40.4	2.7	1.6	1.2	2.0
Tobacco product user in household
No	3828	72.9	1.1	2691	71.2	0.7	1137	28.8	0.7		Referent
Yes	1497	27.1	1.1	925	62.1	1.3	572	37.9	1.3	1.5	1.3	1.7
Exposed to any smoking (home, car, school, outdoors)
No	3620	69.7	0.9	2616	73.1	0.7	1004	26.9	0.7		Referent
Yes	1590	30.3	0.9	913	58.1	1.3	677	41.9	1.3	2.0	1.7	2.3
Receptivity to tobacco advertising
No	3116	59.3	0.9	2338	76.0	0.8	778	24.0	0.8		Referent
Yes	2209	40.7	0.9	1278	58.1	1.2	931	41.9	1.2	2.3	2.0	2.6

Statistically significant associations at p < .05 indicated in bold text.

^an represents unweighted sample sizes.

^bPercents (%) and standard errors (SEs) represent weighted estimates.

^cOdds ratios (ORs) and 95% confidence intervals (CIs) from unadjusted logistic regression models.

^dMeasured via the modified Brief Sensation Seeking Scale.

^eIncludes non-Hispanic American Indian/Alaska Natives, Asian/Native Hawaiian/Other Pacific Islanders, or multiple races.

^fAssessed via the Global Appraisal of Individual Needs-Short Screener (GAIN-SS) subscales, categorized into no/low (0–2 symptoms), moderate severity (3–7 symptoms) and high severity (8–11 symptoms).

^gAcademic performance was based on parents’ responses regarding their child’s (or self-reported in case of emancipated youth) grades at school in the past 12 months and categorized as at or above average (C grades and higher) and below average (below Cs).

Among W1 youth never substance users, 23% began using at least one substance (i.e. tobacco products, alcohol, marijuana, or other drugs) at W2 (Table 2). Onset of exclusive alcohol use was the highest at 8.2% (SE = 0.4), followed by 5.0% (SE = 0.4) for polysubstance use including tobacco, 4.4% (SE = 0.3) for exclusive tobacco use, 3.1% (SE = 0.3) for exclusive other drug use, 1.4% (SE = 0.2) for polysubstance use excluding tobacco, and 0.9% (SE = 0.1) for exclusive marijuana use. In unadjusted models, sensation seeking, age, sex, race/ethnicity, parent’s marital status, lifetime mental health problems, academic performance, tobacco product user in the household, exposure to smoking, and receptivity to tobacco advertising were each associated with onset of substance use compared to never use.

Table 2.

Selected characteristics by substance use onset among W1 youth never substance users (n = 5325).

	Wave 2 onset of substance use
	Never usea			Exclusive tobacco useb						Exclusive alcohol use						Exclusive marijuana usec
Wave 1	ng	%h	SEh	ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi
Total	4090	77.0	0.7	246	4.4	0.3				422	8.2	0.4				53	0.9	0.1
Sensation seekingj		N/A			N/A		1.4	1.1	1.7				1.4	1.6					1.5
Age (years)
12–14	2722	80.3	0.8	138	3.8	0.3		Referent		231	7.0	0.5	Referent			31	0.8	0.1	Referent
15–17	1368	71.3	1.2	108	5.3	0.5	1.6	1.2	2.1	191	10.2	0.8	1.6	2.1		22	1.0	0.2	1.4
Gender
Male	2185	78.0	0.9	162	5.4	0.4		Referent		186	7.0	0.5	Referent			26	0.8	0.2	Referent
Female	1905	75.9	1.0	84	3.2	0.4	0.6	0.5	0.8	236	9.5	0.7	1.4	1.8		27	0.9	0.2	1.2
Race/ethnicity
Non-Hispanic, White	1874	77.5	1.0	103	4.0	0.4		Referent		228	9.5	0.7	Referent			10	0.4n	0.1	Referent
Non-Hispanic, Black	602	76.4	1.7	36	4.6	0.8	1.2	0.8	1.8	43	5.4	0.9	0.6	0.8		19	2.4	0.6	6.1
Non-Hispanic, otherk	350	78.6	2.3	18	2.6	0.7	0.6	0.4	1.1	41	8.4	1.4	0.9	1.3		8	1.0n	0.3	2.4
Hispanic	1264	75.7	1.3	89	5.6	0.7	1.4	1.0	2.1	110	7.1	0.6	0.8	1.0		16	0.9	0.2	2.2
Parent educational attainment
Up to 8th grade	314	78.5	2.3	19	5.0	1.1		Referent		23	6.2	1.4	Referent			5	1.1n	0.5	Referent
HS, no diploma	588	74.5	1.7	53	7.4	1.2	1.6	0.9	2.8	43	5.5	1.0	0.9	1.8		10	1.2n	0.4	1.2
HSD or GED	731	75.3	1.4	53	5.4	0.6	1.1	0.7	1.9	69	7.3	0.9	1.2	2.1		16	1.4	0.4	1.3
Some college or more	2442	77.8	0.9	121	3.4	0.3	0.7	0.4	1.2	287	9.2	0.7	1.5	2.6		22	0.6	0.1	0.6
Parental marital status
Married	2722	78.8	0.9	132	3.6	0.3		Referent		284	8.3	0.5	Referent			30	0.7	0.1	Referent
Unmarried	1363	73.7	1.1	113	5.7	0.5	1.7	1.3	2.2	138	8.1	0.8	1.0	1.3		23	1.1	0.2	1.7
Lifetime mental health problemsl
No/low severity	1470	82.5	0.9	64	3.6	0.5		Referent		99	5.8	0.6	Referent			20	0.9	0.2	Referent
Moderate severity	1747	75.8	1.0	98	3.9	0.4	1.2	0.8	1.7	219	9.7	0.7	1.8	2.3		20	0.7	0.2	0.8
High severity	873	71.4	1.5	84	6.4	0.7	2.0	1.4	2.9	104	8.8	1.0	1.8	2.4		13	1.0n	0.3	1.2
Academic performancem
At or above average	3736	77.8	0.8	198	3.8	0.3		Referent		393	8.4	0.5	Referent			43	0.8	0.1	Referent
Below average	354	68.9	2.1	48	10.1	1.4	3.0	2.1	4.3	29	5.8	1.1	0.8	1.2		10	1.8n	0.6	2.6
Tobacco product user in household
No	3015	79.2	0.9	141	3.4	0.3		Referent		303	8.1	0.5	Referent			36	0.8	0.1	Referent
Yes	1075	71.1	1.3	105	7.0	0.7	2.3	1.7	3.2	119	8.6	0.7	1.2	1.5		17	1.1	0.3	1.7
Exposed to any smoking (home, car, school, outdoors)
No	2873	79.6	0.8	135	3.4	0.3		Referent		277	7.9	0.5	Referent			30	0.7	0.1	Referent
Yes	1125	70.7	1.2	105	6.5	0.6	2.2	1.6	2.9	137	8.9	0.7	1.3	1.6		21	1.2	0.3	1.9
Receptivity to tobacco advertising
No	2498	80.6	0.8	118	3.5	0.3		Referent		225	7.4	0.5	Referent			28	0.7	0.2	Referent
Yes	1592	71.7	1.3	128	5.6	0.6	1.8	1.3	2.5	197	9.4	0.7	1.4	1.8		25	1.0	0.2	1.5

	Wave 2 onset of substance use
	Exclusive marijuana usec		Exclusive other drug used						Poly-substance including tobacco usee						Poly-substance excluding tobacco usef
Wave 1		95% CIi	ng	%h	SEh	ORi		95% CIi	ng	%h	SEh	ORi		95% CIi	ng	%h	SEh	ORi		95% CIi
Total			174	3.1	0.3				258	5.0	0.4				82	1.4	0.2
Sensation seekingj	1.0	2.1				1.1	0.9	1.4				1.9	1.6	2.2				1.3	1.0	1.7
Age (years)
12–14		Referent	110	3.2	0.3		Referent		125	3.7	0.4		Referent		45	1.2	0.2		Referent
15–17	0.9	2.4	64	3.0	0.4	1.1	0.8	1.5	133	7.4	0.7	2.3	1.8	2.8	37	1.8	0.3	1.7	1.1	2.7
Gender
Male		Referent	77	2.8	0.3		Referent		134	5.0	0.5		Referent		33	1.1	0.2		Referent
Female	0.7	2.2	97	3.5	0.4	1.3	0.9	1.9	124	5.1	0.5	1.1	0.8	1.4	49	1.8	0.3	1.7	1.0	2.8
Race/ethnicity
Non-Hispanic, White		Referent	74	2.9	0.3		Referent		105	4.7	0.5		Referent		29	1.1	0.2		Referent
Non-Hispanic, Black	2.6	14.2	29	3.8	0.8	1.3	0.8	2.3	43	5.4	0.9	1.2	0.8	1.8	14	2.0	0.6	1.9	1.0	3.7
Non-Hispanic, otherk	1.0	5.6	21	3.7	0.8	1.3	0.8	2.1	18	3.9n	1.3	0.8	0.4	1.8	7	1.8n	0.9	1.6	0.5	5.5
Hispanic	0.9	5.3	50	3.1	0.4	1.1	0.8	1.6	92	6.0	0.7	1.3	0.9	1.9	32	1.7	0.3	1.6	0.9	2.9
Parent educational attainment
Up to 8th grade		Referent	6	1.5n	0.7		Referent		21	5.7	1.2		Referent		10	2.1n	0.7		Referent
HS, no diploma	0.4	3.9	35	4.0	0.7	2.8	0.9	8.5	41	5.5	0.9	1.0	0.6	1.8	15	1.8	0.5	0.9	0.4	2.1
HSD or GED	0.4	4.4	31	3.2	0.6	2.3	0.8	6.7	56	5.9	0.9	1.1	0.6	1.9	14	1.4	0.4	0.7	0.3	1.7
Some college or more	0.2	2.0	101	3.1	0.3	2.1	0.8	5.6	139	4.6	0.5	0.8	0.5	1.4	43	1.3	0.3	0.6	0.3	1.4
Parental marital status
Married		Referent	114	3.2	0.3		Referent		140	4.3	0.5		Referent		47	1.2	0.2		Referent
Unmarried	1.0	3.1	59	3.0	0.4	1.0	0.7	1.4	117	6.4	0.6	1.6	1.2	2.2	35	1.9	0.4	1.8	1.1	2.9
Lifetime mental health problemsj
No/low severity		Referent	42	2.4	0.4		Referent		54	3.4	0.5		Referent		25	1.4	0.3		Referent
Moderate severity	0.5	1.6	94	3.9	0.5	1.8	1.1	2.9	113	4.9	0.5	1.5	1.1	2.2	29	1.2	0.3	0.9	0.5	1.8
High severity	0.5	2.9	38	2.8	0.4	1.4	0.8	2.3	91	7.7	0.9	2.6	1.7	3.9	28	2.0	0.4	1.6	0.9	2.9
Academic performancem
At or above average		Referent	159	3.2	0.3		Referent		215	4.6	0.4		Referent		73	1.4	0.2		Referent
Below average	1.1	6.5	15	2.7	0.7	0.9	0.5	1.7	43	8.9	1.3	2.2	1.6	2.9	9	1.8n	0.6	1.5	0.7	3.0
Tobacco product user in household
No		Referent	132	3.3	0.3		Referent		152	4.1	0.4		Referent		49	1.2	0.2		Referent
Yes	0.9	3.0	42	2.8	0.4	1.0	0.7	1.3	106	7.5	0.8	2.0	1.5	2.6	33	1.9	0.4	1.7	1.0	3.0
Exposed to any smoking (home, car, school, outdoors)
No		Referent	112	3.0	0.3		Referent		144	4.2	0.4		Referent		31	1.2	0.2		Referent
Yes	1.0	3.4	58	3.4	0.4	1.3	0.9	1.8	113	7.4	0.7	2.0	1.6	2.6	49	1.9	0.4	1.8	1.1	3.0
Receptivity to tobacco advertising
No		Referent	92	2.8	0.3		Referent		115	3.8	0.4		Referent		40	1.2	0.2		Referent
Yes	0.8	2.9	82	3.7	0.4	1.5	1.0	2.2	143	6.8	0.7	2.0	1.5	2.7	42	1.7	0.3	1.6	1.0	2.5

Statistically significant associations at p < .05 indicated in bold text.

^aReference group.

^bRefers to cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, snus pouches, smokeless tobacco excluding snus pouches, or dissolvable tobacco.

^cIncludes use of cigars as blunts.

^dIncludes non-prescribed Ritalin/Adderall, non-prescribed painkillers/sedatives, cocaine and other stimulants, heroin, inhalants, solvents, and hallucinogens.

^eIncludes use of two or more substances (tobacco, alcohol, marijuana, and other drugs).

^fIncludes use of two or more substances (alcohol, marijuana, and other drugs).

^gn represents unweighted sample sizes.

^hPercents (%) and standard errors (SEs) represent weighted estimates.

ⁱOdds ratios (ORs) and 95% confidence intervals (CIs) from unadjusted logistic regression models.

^jMeasured via the modified Brief Sensation Seeking Scale.

^kIncludes non-Hispanic American Indian/Alaska Natives, Asian/Native Hawaiian/Other Pacific Islanders, or multiple races.

^lAssessed via the Global Appraisal of Individual Needs-Short Screener (GAIN-SS) subscales, categorized into no/low (0–2 symptoms), moderate severity (3–7 symptoms) and high severity (8–11 symptoms).

^mAcademic performance was based on parents’ responses regarding their child’s (or self-reported in case of emancipated youth) grades at school in the past 12 months and categorized as at or above average (C grades and higher) and below average (below Cs).

ⁿEstimate should be interpreted with caution because it has low statistical precision. It is based on a denominator sample size of < 50, or the coefficient of variation of the estimate or its complement is larger than 30%.

Table 3 presents the associations between susceptibility to tobacco use and onset of any tobacco, alcohol, marijuana, and other drug use. Susceptibility to tobacco use was significantly associated with higher odds of onset of tobacco use (OR = 3.0; 95% CI: 2.3, 3.8), alcohol use (OR = 1.8; 95% CI: 1.5, 2.2), and marijuana use (OR = 3.4; 95% CI: 2.4, 4.8) after adjusting for demographics, academic performance, sensation seeking, tobacco product user in the household, receptivity to tobacco product advertising, and lifetime mental health problems. No significant association was observed between susceptibility to tobacco use and onset of other drug use.

Table 3.

Susceptibility to tobacco use and onset of tobacco, alcohol, marijuana, and other drug use among W1 youth never substance users (n = 5325).

	Wave 2
	Onset of tobacco usea						Onset of alcohol use						Onset of marijuana useb						Onset of other drug usec
Wave 1	nd	%e	SEe	ORf	95% CIf		nd	%e	SEe	ORf	95% CIf		nd	%e	SEe	ORf	95% CIf		nd	%e	SEe	ORf	95% CIf
Total	504	9.4	0.5				668	12.9	0.6				229	4.1	0.3				290	5.3	0.3
	Unadjusted model
Susceptibility to tobacco useg
No	202	5.4	0.4	Referent			352	10.0	0.6	Referent			79	2.0	0.2	Referent			187	5.0	0.4	Referent
Yes	302	18.1	1.1	3.8	3.0	4.9	316	19.2	1.2	2.1	1.8	2.6	150	8.8	0.8	4.7	3.5	6.3	103	6.0	0.7	1.2	0.9	1.6
	Adjusted modelh
Susceptibility to tobacco useg
No	202	5.4	0.4	Referent			352	10.0	0.6	Referent			79	2.0	0.2	Referent			187	5.0	0.4	Referent
Yes	302	18.1	1.1	3.0	2.3	3.8	316	19.2	1.2	1.8	1.5	2.2	150	8.8	0.8	3.4	2.4	4.8	103	6.0	0.7	1.0	0.7	1.4

Statistically significant associations at p < .05 indicated in bold text.

^aRefers to cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, snus pouches, smokeless tobacco excluding snus pouches, or dissolvable tobacco.

^bIncludes use of cigars as blunts.

^cIncludes non-prescribed Ritalin/Adderall, non-prescribed painkillers/sedatives, cocaine and other stimulants, heroin, inhalants, solvents, and hallucinogens.

^dn represents unweighted sample sizes.

^ePercents (%) and standard errors (SEs) represent weighted estimates.

^fOdds ratios (ORs) and 95% confidence intervals (CIs) from logistic regression models.

^gDefined as the strongest rejection to questions regarding curiosity about tobacco products, intention not to use tobacco products, and resistance of an offer to use tobacco products from a best friend.

^hAdjusted for age, sex, race/ethnicity, academic performance, sensation seeking, tobacco product user in household, receptivity to tobacco product advertising, and lifetime mental health problems at Wave 1.

Table 4 presents the association between susceptibility to tobacco use and the 7-category onset of substance use variable. In adjusted models (bottom portion of Table 3), tobacco-susceptible youth at W1 compared with those not susceptible to tobacco use, had higher odds of onset of exclusive tobacco use (AOR = 2.4; 95% CI: 1.7, 3.3), exclusive alcohol use (AOR = 1.5; 95% CI: 1.2, 1.8), and polysubstance (including [AOR = 3.9; 95% CI: 2.8, 5.6] and excluding [AOR = 1.8; 95% CI: 1.1, 3.0] tobacco) use at W2 compared with W2 never substance use. Results from the unadjusted models (top portion of Table 4) were largely consistent with the adjusted models, with two differences. First, and not surprisingly, the magnitude of the odds ratios is larger in the unadjusted models. Second, the association between susceptibility to tobacco use and the onset of exclusive marijuana use was significant only in the unadjusted model.

Table 4.

Susceptibility to tobacco use and substance use onset among W1 youth never substance users (n = 5325).

	Wave 2
	Onset of substance use
	Never usea			Exclusive tobacco useb						Exclusive alcohol use						Exclusive marijuana usec
Wave 1	ng	%h	SEh	ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi
Total	4090	77.0	0.7	246	4.4	0.3				422	8.2	0.4				53	0.9	0.1
	Unadjusted model
Susceptibility to tobacco usej
No	2956	81.9	0.8	113	3.0	0.3		Referent		255	7.3	0.5	Referent			29	0.7	0.1	Referent
Yes	1134	66.2	1.3	133	7.4	0.6	3.1	2.3	4.1	167	10.1	0.8	1.7	2.1		24	1.2	0.3	2.1
	Adjusted modelk
Susceptibility to tobacco usej
No	2956	81.9	0.8	113	3.0	0.3		Referent		255	7.3	0.5	Referent			29	0.7	0.1	Referent
Yes	1134	66.2	1.3	133	7.4	0.6	2.4	1.7	3.3	167	10.1	0.8	1.5	1.8		24	1.2	0.3	1.6

	Wave 2
	Onset of substance use
	Exclusive marijuana usec		Exclusive other drug used						Poly-substance including tobacco usee						Poly-substance excluding tobacco usef
Wave 1	95% CIi		ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi	95% CIi		ng	%h	SEh	ORi	95% CIi
Total			174	3.1	0.3				258	5.0	0.4				82	1.4	0.2
	Unadjusted model
Susceptibility to tobacco usej
No		Referent	131	3.5	0.3		Referent		89	2.5	0.3		Referent		43	1.2	0.2		Referent
Yes	1.1	4.0	43	2.4	0.4	0.9	0.6	1.3	169	10.7	1.1	5.4	3.9	7.5	39	2.0	0.4	2.2	1.3	3.6
	Adjusted modelk
Susceptibility to tobacco usej
No		Referent	131	3.5	0.3		Referent		89	2.5	0.3		Referent		43	1.2	0.2		Referent
No
Yes	0.8	3.1	43	2.4	0.4	0.8	0.5	1.2	169	10.7	1.1	3.9	2.8	5.6	39	2.0	0.4	1.8	1.1	3.0

Statistically significant associations at p < .05 indicated in bold text.

^aReference group.

^bRefers to cigarettes, e-cigarettes, traditional cigars, cigarillos, filtered cigars, pipe, hookah, snus pouches, smokeless tobacco excluding snus pouches, or dissolvable tobacco.

^cIncludes use of cigars as blunts.

^dIncludes non-prescribed Ritalin/Adderall, non-prescribed painkillers/sedatives, cocaine and other stimulants, heroin, inhalants, solvents, and hallucinogens.

^eIncludes use of two or more substances (tobacco, alcohol, marijuana, and other drugs).

^fIncludes use of two or more substances (alcohol, marijuana, and other drugs).

^gn represents unweighted sample sizes.

^hPercents (%) and standard errors (SEs) represent weighted estimates.

ⁱOdds ratios (ORs) and 95% confidence intervals (CIs) from multinomial logistic regression models.

^jDefined as the strongest rejection to questions regarding curiosity about tobacco products, intention not to use tobacco products, and resistance of an offer to use tobacco products from a best friend.

^kAdjusted for age, sex, race/ethnicity, academic performance, sensation seeking, tobacco product user in household, receptivity to tobacco product advertising, and lifetime mental health problems at Wave 1.

4. Discussion

Consistent with prior studies that focused on specific tobacco products (Nicksic and Barnes, 2019; Nodora et al., 2014; Pierce et al., 1996; Pierce et al., 2018), susceptibility to tobacco use was associated with the onset of tobacco use both for exclusive tobacco use and for polysubstance use including tobacco. Importantly, our findings extend susceptibility to tobacco use as a predictor of the onset of use of substances other than tobacco products. A prior PATH Study analysis showed significant associations between e-cigarette susceptibility and onset of alcohol and marijuana use a year later (Nicksic and Barnes, 2019), and our study expanded these findings to susceptibility to any tobacco use beyond e-cigarettes. Moreover, because youth who use alcohol and drugs tend to also use tobacco products, we examined onset of substance use as seven mutually exclusive categories and found that susceptibility to tobacco use predicted the onset of exclusive tobacco use, exclusive alcohol use, and polysubstance use (including and excluding tobacco).

Susceptibility to tobacco use was not significantly associated with the onset of exclusive marijuana use; this suggests that the association between susceptibility to tobacco use and onset of marijuana use may be driven by polysubstance use in youth marijuana users. However, the low incidence of exclusive marijuana use limited our ability to detect an association with a small effect, though it is consistent with previous research finding relative rarity of exclusive marijuana use in youth (Han et al., 2017). Future studies can examine this heterogeneity in youth marijuana users based on their use of other substances including tobacco products. Additionally, susceptibility to tobacco use was not significantly associated with exclusive other drug use (i.e. non-prescribed use of Ritalin® or Adderall®, painkillers, sedatives, or tranquilizers, and cocaine or crack, other stimulants [i.e. methamphetamine or speed], heroin, inhalants, solvents, and hallucinogens); this again implicates other risk factors for youth who initiate exclusively with these substances. Although relatively few youth (3%) who had never previously used substances exclusively initiated other drug use, it is worth noting that non-prescribed painkillers were the predominant substance used by this group (used by > 90%). Importantly, lifetime mental health problems and receptivity to tobacco advertising were the only covariates associated with onset of exclusive other drug use relative to never substance use. That other traditional risk factors for substance use onset (e.g. age, sex, race/ethnicity, and academic performance) were not associated with this group warrants further exploration of this unique group of youth substance users and highlights the need for tailored intervention approaches. Finally, that susceptibility to tobacco use predicted the onset of exclusive alcohol use and polysubstance use with or without tobacco use supports our hypothesis that susceptibility to tobacco use can be used to predict substance use in general and may be a marker of vulnerability to any substance use.

Our findings have important implications for substance use prevention. The propensity of overall substance use is likely driven by shared susceptibility to tobacco, alcohol, and drug use possibly originating from genetic vulnerabilities or environmental factors such as access to substances and opportunities to use that govern cognitions, intentions, and peer influence on substance use onset (Vanyukov and Ridenour, 2012; Vanyukov et al., 2012). Our findings were robust to adjustment for several risk factors for tobacco and other substance use such as sensation seeking, tobacco product user in household, receptivity to tobacco product advertising, and lifetime mental health problems. While each of these covariates is an indicator for some source of common variance, future studies can examine shared genetic or environmental factors driving the propensity of overall substance use. Identifying and targeting youth at risk for substance use onset early in this critical developmental period can prevent progression from susceptibility to experimentation. Although various substance use prevention models exist that incorporate relevant risk factors associated with substance use onset (Botvin, 2000; Corbett, 2001; Griffin and Botvin, 2010; Griffin et al., 2003), integrating susceptibility to tobacco use as a potential screening tool to identify youth at risk of substance use may strengthen these prevention measures.

4.1. Study limitations and strengths

Our findings should be interpreted within the context of a few limitations. The PATH Study did not collect information on susceptibility per se to alcohol, marijuana, and other drug use, and therefore its effect on our results is unknown. Information on social norms and peer and family influences on tobacco and other substance use was also not available in the PATH Study. Additionally, the self-reported data are subject to misclassification. Nevertheless, this study is the first to prospectively show that susceptibility to tobacco use can be extended to predict the onset of substance use besides tobacco in a nationally representative sample of U.S. youth never users of any substance. Thus, as an early risk indicator, the short susceptibility scale of Pierce et al. might show even greater promise for substance use prevention.

5. Conclusions

Susceptibility to tobacco use significantly predicts tobacco use onset as well as onset of exclusive alcohol and polysubstance use among American youth. To maximize success, substance use prevention programs should target youth prior to them becoming susceptible and develop strategies based on the shared cognitions, intentions, and peer influences that prevail in tobacco-susceptible youth. The simple 3-question index may be one way to quickly identify youth at risk for initiating the use of addictive substances.