Tobacco and cannabis co-use: Drug substitution, quit interest, and cessation preferences

Abstract

The co-use of tobacco and cannabis is a common practice worldwide and carries with it substantial public health burden. Few interventions exist that target both substances and little is known about quit interest, treatment preferences, and drug substitution during past cessation attempts, which is critical to guide the development of treatment strategies. The goal of this study was to provide descriptive information regarding quit interest, treatment preferences, and perceived drug substitution among adult (age 18+) cannabis-tobacco co-users. Participants (N=282) from two independent survey samples (recruited from Amazon Mechanical Turk) from across the United States were combined. Among all participants, 57% were female, 79% were White, and average age was 33.31 (SD=9.54) years old. Approximately 80% had tried to quit smoking cigarettes at least once, while 40% had tried to quit using cannabis at least once. Of those who tried to quit, 50% self-reported a perceived increase in their cannabis use during tobacco cessation and 62% self-reported a perceived increase in their tobacco use during cannabis cessation. Average quit interest (10-point scale) for cannabis was 2.39 (SD=2.35) and for tobacco was 7.07 (SD=2.90). Results of this study suggest that tobacco use should be addressed among cannabis-tobacco co-users, but interventions should consider lack of interest in cannabis cessation. Reduction-based strategies for cannabis use appear to be more acceptable to this non-treatment seeking, co-using population. Drug substitution during quit attempts for one substance should be further explored as an important treatment consideration.

Introduction

Cannabis and tobacco are frequently used together, either simultaneously (e.g., cigar wrappers filled with cannabis ‘blunts,’ or tobacco/cannabis rolled cigarettes ‘spliffs’), sequentially (‘chasing’ cannabis with tobacco), or in an asynchronous way (not within same use episode, but used by the same individual). The co-use of cannabis and tobacco continues to be a public health concern in the United States (US), even amid steadily declining tobacco use rates (Jamal et al., 2016). The co-use of these substances, which is typically defined as any past-month use of both substances (Schauer, Berg, Kegler, Donovan, & Windle, 2015; Schauer & Peters, 2018) is common (Agrawal, Budney, & Lynskey, 2012; Agrawal & Lynskey, 2009; Agrawal et al., 2008; Leatherdale, Ahmed, & Kaiserman, 2006; Leatherdale, Hammond, Kaiserman, & Ahmed, 2007; Richter et al., 2004; Tullis, Dupont, Frost-Pineda, & Gold, 2003). While overall rates of co-use in the US appear to have increased modestly from 2003 through 2012 (Schauer et al., 2015), the prevalence of daily co-use has doubled (Goodwin et al., 2018). Recent estimates of tobacco co-use among cannabis users in the US demonstrate that prevalence is still high (68.6% excluding blunts, and 78.3% including blunts) (Schauer, Berg, Kegler, Donovan, & Windle, 2016).

Dual use of tobacco and cannabis is associated with substantial public health burden, including greater prevalence of psychiatric and psychosocial problems (Peters, Schwartz, Wang, O’Grady, & Blanco, 2014; Ramo, Liu, & Prochaska, 2012) and additive health risk due to co-use (Meier & Hatsukami, 2016). The use of cannabis has been associated with an increased risk and greater levels of dependence on nicotine (Agrawal et al., 2008; Okoli, Richardson, Ratner, & Johnson, 2008; Wang, Ramo, Lisha, & Cataldo, 2016) and cigarette smoking has been associated with cannabis dependence (Hindocha et al., 2015). Co-users tend to have worse outcomes than cannabis-only users in human laboratory models of cannabis relapse (Haney et al., 2013) and more difficulty achieving abstinence than former and never tobacco users during treatment (Gray et al., 2017; E. A. McClure, Baker, & Gray, 2014; Moore & Budney, 2001; Peters et al., 2014). Co-users may also experience more difficulty in achieving abstinence from tobacco (El-Khoury Lesueur, Bolze, & Melchior, 2018; Schauer, King, & McAfee, 2017; Weinberger, Platt, Copeland, & Goodwin, 2018).

While preliminary work has been conducted on the evaluation of interventions to address cannabis and tobacco co-use (Becker, Haug, Sullivan, & Schaub, 2014; Beckham et al., 2018; Hill et al., 2013; Lee et al., 2015), little is known regarding quit interest and treatment preferences among co-users. Among young adults (ages 18–25), previous studies have found that the desire to quit smoking cigarettes is greater than for cannabis, the perceived importance of quitting is greater for cigarettes, but confidence in quitting smoking cigarettes is lower (Masters, Haardorfer, Windle, & Berg, 2018; Ramo, Delucchi, Liu, Hall, & Prochaska, 2014). Extending this work to adult co-users is an important contribution to the treatment literature. Adult co-users are likely to have a longer history of use, more entrenched use patterns and may have different reasons for use compared to young adults (e.g., managing withdrawal vs. socially motivated), which may impact their interest in quitting either positively or negatively. Given a longer history with these substances, they may also have had more past quit attempts to inform their perceptions of potential drug substitution during a quit attempt. Finally, adults may have encountered more negative consequences associated with use that also may impact their quit interest or may have different reasons for wanting to quit (e.g., health concerns). These data in adult co-users are necessary and formative for the development of acceptable treatment interventions and in promoting uptake and engagement leading to eventual dissemination.

A relevant cessation concern among co-users is the possible occurrence of drug substitution (or compensatory use) of one substance during reduction or abstinence from the other. Patterns of substitution are important to determine as treatments for only one may be missing an unintended negative effect. However, little research exists to suggest whether drug substitution occurs reliably for co-users when trying to quit one substance. Of the available literature on substitution, results are mixed, with some evidence supporting compensatory tobacco use during cannabis abstinence (Copersino et al., 2006; Schaub, Gmel, Annaheim, Mueller, & Schwappach, 2010) and other studies finding no changes in use (Haney et al., 2010; E. A. McClure et al., 2014; Peters & Hughes, 2010).

Existing data to inform interventions for tobacco and cannabis co-use are limited, and thus the goal of this cross-sectional survey study was to address those gaps in the literature with the following aims; 1) assess self-reported substitution of either cannabis or tobacco during past cessation attempts, and 2) assess motivation to quit both cannabis and tobacco and cessation preferences among co-users, as well as predictors of motivation to quit.

Methods

Data were drawn from two independent samples of adult cannabis tobacco co-users in the US. Tobacco use refers to combustible cigarette smoking, and questions did not assess the use of other forms of tobacco/nicotine use. Cannabis use was not specific to only smoking or using plant material, though the majority of study participants smoked cannabis as their primary method of administration. The first sample was from a national (US) survey that recruited co-users (N=100). Data collection occurred in April 2016. The second US survey sample recruited heavy cannabis users from the Southeastern US and then asked about tobacco use status (tobacco use was not required for participation); only co-users are included in this report (N=182). Data collection for the Southeastern sample occurred from June 2017 until August 2017. The Southeastern focus within the latter sample was meant to control for and further explore regional differences, potentially due to differences in cannabis legislation, as well as tobacco control policies. For example, the Southeastern US tends to have relaxed tobacco control policies (e.g., low taxes), while also having more restrictive cannabis legislation compared to other regions of the US. Additionally, in a recent national cannabis cessation study (Gray et al., 2017), there were regional differences in tobacco co-use rates among cannabis users depending on their geographic location (E.A. McClure et al., In Press).

Participants

Across both studies, participants were recruited from Amazon Mechanical Turk (MTurk) survey listings (separately). MTurk is an online crowdsourcing service and marketplace and has been used successfully for recruitment of other substance use populations (Adkison, O’Connor, Chaiton, & Schwartz, 2015; Huhn, Tompkins, & Dunn, 2017; Peters, Rosenberry, Schauer, O’Grady, & Johnson, 2017; Rass, Pacek, Johnson, & Johnson, 2015; Strickland & Stoops, 2015).

Southeastern (SE) Sample.

The study opportunity was presented as a Human Intelligence Task (HIT) on MTurk. An initial eligibility survey and full survey (if applicable) were administered via Research Electronic Data Capture (REDCap) (Harris et al., 2009). The states included in this survey were based on the US Census Division 5 (South Atlantic) and Division 6 (East South Central), which includes the following states and territories: Alabama, Delaware, District of Columbia, Florida, Georgia, Kentucky, Maryland, Mississippi, North Carolina, South Carolina, Tennessee, Virginia and West Virginia.

MTurk enrollees (also known as workers) saw an advertisement that asked if they had ever used marijuana (cannabis) or other drugs. If eligible, workers were routed to the full survey. To be eligible, the participant had to meet the following criteria; 1) ≥18 years of age, 2) must reside in the Southeastern US, 3) self-reported use of cannabis on at least 20 of the past 30 days, and 4) must have an MTurk HIT approval rating of 85% or higher (based on the quality of previous tasks completed). The full questionnaire included 40–70 questions, depending on the worker’s use history and took approximately 20–30 minutes to complete. Participants received $1.00 for completing the full survey, and an additional $1.50 if workers responded correctly to attention check questions (e.g., answer “A” for the following question) dispersed throughout the survey, for total possible compensation of $2.50.

The eligibility screener was completed by 10,175 MTurk workers and 472 (4.6%) were eligible to complete the full survey. Fifty participants were excluded from analyses due to inconsistent or incomplete responses or failing attention checks. The final sample included 432 adult cannabis users, of which 182 (42%) were current tobacco co-users (any cigarettes smoked in past 30 days). The Institutional Review Board at the Medical University of South Carolina reviewed and approved all study procedures (Protocol number: PRO00067646; Study Title: Marijuana and Tobacco Use and Co-Use Prevalence and Patterns).

National (NL) Sample.

Participants from this sample were also recruited through MTurk (14 months earlier), using the following eligibility criteria: 1) current residence in the US, 2) ≥18 years of age, 3) a 95% HIT approval rating, and 4) self-reported use of cannabis on at least 20 of the past 30 days and daily tobacco (cigarette) smoking in the past 30 days. Study findings and procedures from this survey have been published elsewhere (Peters et al., 2017). In brief, a survey opportunity was advertised on MTurk as a “Marijuana Survey.” Individuals completed an anonymous screening on Survey Monkey to determine eligibility; those who met inclusion criteria then completed an electronic informed consent and were directed to complete the full survey. Compensation was identical to the SE survey. A total of 579 MTurk workers completed the eligibility screener and 105 (18%) completed the full survey. Survey advertising text was specifically directed to the co-using population for the NL survey, whereas the SE advertising text was more general, resulting in varying numbers of eligibility across samples. The Battelle Memorial Institute Institutional Review Board reviewed and approved the study (Protocol number: 0611–100073659 Rev 0.0; Study Title: A Laboratory Model of Increasing THC Potency on Cigarette Smoking (MJCIG) - Phase A).

MTurk worker IDs were compared for duplicates across the two samples. Of the five duplicates identified, two participants had inconsistent responses across surveys and were excluded from analyses. The remaining three duplicates were retained in the SE sample, as it was the more recent survey. Thus, of the 105 NL participants who completed the full survey, 100 were included in the present analysis. The final study sample included 282 adult cannabis-tobacco co-users (n=100 in the NL sample and n=182 in the SE sample).

Measures

The survey used for the SE sample was derived from the NL survey (Peters et al., 2017). Since the SE survey was conducted after the NL survey, additional items were included that were deemed appropriate and of interest to assess among co-users, but were not specific to region. Several items included in this report were only administered to the SE sample. Additional questions and modifications to existing questions for the SE questionnaire are noted below. Participants in the NL sample had the response option “Refuse to answer” for questions about their cannabis and tobacco use, which was not a response option in the SE survey.

Demographics.

Participants provided standard demographic information on age, gender, race, ethnicity, education, employment, marital status, and income. When necessary, response options between the samples were re-coded for consistency, and some categories (race, education, employment status, marital status, and income) were collapsed.

Cannabis and Tobacco Use and History.

Participants self-reported current and historic tobacco (i.e., cigarettes) and cannabis use characteristics. Any mention of cigarettes in the survey referred to tobacco products, not cannabis. Questions included the following: number of (tobacco) cigarettes smoked per day, smoking their first cigarette within 30 minutes of waking, age started smoking regularly (defined as daily or near daily), number of days that cannabis was used in the past 30, number of times cannabis was used per day on using days (on average), use of cannabis within 30 minutes of waking, and age that cannabis was first used. Participants were also asked (in their lifetime) which substance they started using first (cannabis, tobacco, or both at the same time). Wording of these questions was similar across study samples. All participants in the NL sample endorsed being daily cigarette smokers, while 74% of the SE sample were daily smokers and 26% were non-daily smokers. All tobacco variables were collapsed across daily and non-daily smokers in the SE sample.

Cannabis and Tobacco Cessation History.

Participants were asked if they had ever tried to quit cannabis completely, ever tried to quit tobacco completely and any past-year quit attempts for both substances (yes/no). Participants were also asked about perceived drug substitution (i.e., increases) during periods of attempted cessation (“When you have tried to quit using marijuana completely, has your use of tobacco increased?” and “When you have tried to quit smoking completely, has your use of marijuana increased?”). Response options were yes/no. Participants did not have the option to self-report on perceived decreases in other substance use during quit attempts. Participants were not instructed to respond to this question for a particular quit attempt, so there may have been variation in responding regarding their most recent quit attempt vs. an aggregate of all quit attempts in their lifetime.

Quit Interest.

Participants were asked about their interest in quitting cannabis and in a separate question, their interest in quitting smoking cigarettes. The NL survey used a scale from 1 to 10 (1=not at all interested, 10=extremely interested in quitting). The SE survey used a 100-point visual analog scale (0=not at all interested, 100=extremely interested in quitting). To transform responses from the SE sample into a 1–10 point scale, all responses were rounded to the nearest integer and then standardized using the following equation: [((X * 9) / 100) + 1)], in which X was the rounded response on the 0–100 scale. Participants were also asked if they were planning to quit using cannabis in the next 30 days and if they were planning to quit smoking cigarettes in the next 30 days (yes/no).

Cessation Preferences (SE survey only).

The SE survey included additional questions regarding cessation preferences and cannabis quit history. Participants were asked if they had ever tried to reduce their cannabis use in the past (yes/no) and if they had ever been to treatment for their cannabis use (yes/no). Participants were asked how likely they would be seek out cannabis treatment (5-point Likert scale), their readiness and confidence to quit cannabis and tobacco (100-point scale; 0=not at all; 100=extremely), as well as their interest in reducing their cannabis use (100-point scale). Finally, participants were asked about their interest in quitting both cannabis and tobacco (100-point scale), treatment preferences and how they see their dual use changing in the next year. Participants were asked about treatment strategies they would use to quit both. Since no pharmacotherapies are approved for cannabis use disorder, pharmacotherapies listed were for tobacco cessation, while other treatment strategies were kept general and included counseling, groups, apps, etc.

Statistical Analyses

Analyses were performed using SPSS Version 22.0 (IBM) and SAS Version 9.4 (SAS Institute). Descriptive statistics (i.e., mean, range, standard deviation for continuous variables, frequencies for categorical variables) were calculated for SE and NL samples. Independent-samples t-tests (continuous variables) and Chi-square tests (categorical variables) were used to compare responses between SE vs. NL samples, using a significance threshold of α = 0.05. Levene’s test for equality of variances were conducted for t-tests, and adjusted p-values reported when appropriate. Logistic regression was used to predict past tobacco/cannabis quit attempts, planned tobacco/cannabis quit attempts, and interest in and readiness for cannabis cessation from demographic and substance use variables (listed in Table 1). Linear regression was used to predict interest in tobacco cessation and cannabis reduction, in addition to readiness and confidence to quit both substances. Survey cohort (SE vs. NL sample) was included as a covariate in all regression models. Because interest in cannabis and tobacco cessation and readiness for cannabis cessation were highly skewed, cannabis quit interest and readiness were dichotomized into “0” representing no interest/readiness and “1” representing any interest (>1). Tobacco quit interest was transformed by raising it to the third power.

Table 1.

Demographics, cannabis and tobacco use characteristics for the overall sample (N=282) and separated by the Southeastern sample (n=182) and the National sample (n=100).

Demographics	Overall Sample (N=282)	Southeastern Sample (n=182)	National Sample (n=100)
	n (%) or Mean(SD)	n (%) or Mean(SD)	n (%) or Mean(SD)	p-value
Age				0.020
Mean (SD)	33.31 (9.54)	34.29 (9.76)	31.54 (8.90)
Range (years)	19–67	19–67	19–57
Gender				0.496
Male	119 (42.20)	73 (40.11)	46 (46.00)
Female	162 (57.45)	108 (59.34)	54 (54.00)
Other	1 (0.35)	1 (0.55)	0 (0.00)
Race				0.259
White	222 (78.72)	145 (79.67)	77 (77.00)
Black	31 (10.99)	22 (12.09)	9 (9.00)
Other/Unknown	29 (10.28)	15 (8.24)	14 (14.00)
Ethnicity				0.068
Hispanic/Latinx	32 (11.35)	16 (8.79)	16 (16.00)
Education				0.435
HS degree/GED/Some College	169 (59.93)	106 (58.24)	63 (63.00)
College graduate/Post graduate	113 (40.07)	76 (41.76)	37 (37.00)
Employment Status				0.123
Full-time/Part-time/Student	232 (82.27)	145 (79.67)	87 (87.00)
Unemployed/Disabled/Retired/Other	50 (17.73)	37 (20.33)	13 (13.00)
Marital Status				0.010
Married/Domestic Partnership	143 (50.71)	82 (45.05)	61 (61.00)
Not Married (All other categories)	139 (49.29)	100 (54.95)	39 (39.00)
Income				0.582
Less than $40,000	171 (60.64)	113 (62.09)	58 (58.00)
More than $40,000	110 (39.01)	68 (37.36)	42 (42.00)
Not sure	1 (0.35)	1 (0.55)	0 (0.00)
Tobacco Use Characteristics
Age started smoking regularly Daily/near daily	18.55 (4.13)	18.36 (4.32)	18.90 (3.75)	0.295
Cigarettes smoked/day				0.987
Mean (SD)	11.35 (6.97)	11.35 (7.32)	11.36 (6.31)
Range	1–32	1–32	1–30
Time to first cigarette				0.064
≤30 min from waking	183 (64.89)	111 (60.99)	72 (72.00)
Cannabis Use Characteristics
Age of cannabis first use	16.50 (4.43)	16.31 (4.76)	16.85 (3.77)	0.327
Days of cannabis use (past 30)	27.16 (3.87)	27.51 (3.47)	26.53 (4.44)	0.059
Times per day cannabis used (past 30)	(n=280) 3.45 (2.84)	3.87 (3.24)	(n=98) 2.67 (1.67)	<0.001
Time to first cannabis use	(n=281)		(n=99)	0.002
≤30 min from waking	120 (42.70)	90 (49.45)	30 (30.30)

Results

Sample Characteristics

Demographics of the overall sample, and the SE and NL samples separately, are shown in Table 1. On average, participants were 33.31 (SD=9.54) years of age, most were female (57%), and White (79%). Approximately 40% had completed a Bachelor’s degree or higher, and the majority (82%) were employed full-time, part-time, or were students. The NL sample included participants from a total of 30 different states; California (18%) and Pennsylvania (11%) had the greatest representation, while only 16% of the NL sample resided in the 13 Southeastern US states targeted by the SE survey. The SE and NL samples differed in age (p=0.020) and marital status (p=0.010). The SE sample tended to be slightly older with lower rates of being married (or in a domestic partnership) and higher rates of being divorced, separated, widowed, or never married compared to the NL sample.

Tobacco and cannabis use characteristics are also shown in Table 1. Participants smoked an average of 11.35 (SD=6.97) cigarettes per day and 65% smoked a cigarette within 30 minutes of waking. All current tobacco use characteristics were similar across study samples. Participants reported using cannabis an average of 27.16 (SD=3.87) of the past 30 days, with no significant differences between groups (though numerically the SE group used cannabis one day per month more on average than the NL group). However, the SE group reported using more times/day than the NL group [3.87 (SD=3.24) vs. 2.67 (SD=1.67), respectively; p<0.001)], and a larger proportion of the SE group vs. NL group reported using within 30 minutes of waking (50% vs. 30%, respectively; p=0.002). The majority of participants from both samples reported initiating tobacco use before cannabis (60%) and 9% of the sample reported initiating use of both substances at the same time.

Quit History and Drug Substitution

Quit history, including self-reported cross-drug substitution during past quit attempts, is shown in Table 2. The majority of participants endorsed having tried to quit smoking cigarettes at least once in their lifetime (79%), while only 40% endorsed ever trying to quit using cannabis at least once. Within the past year, 55% of co-users had made a cigarette quit attempt, while 16% had tried to quit using cannabis. Among those who had ever made a quit attempt for tobacco (n=222), 50% responded that their cannabis use had increased while trying to quit using tobacco. The NL sample had a higher proportion of participants endorsing an increase in cannabis use compared to the SE sample (p=0.011). During a cannabis quit attempt (n=112), 62% responded that their tobacco use had increased.

Table 2.

Cannabis and tobacco quit history, drug substitution and quit interest ratings for the overall sample (N=282) and separated by the Southeastern sample (n=182) and the National sample (n=100).

	Overall Sample (N=282)	Southeastern Sample (n=182)	National Sample (n=100)
Cannabis and Tobacco Quit History	n (%)	n (%)	n (%)	p-value
Quit attempts for cigarettes (ever)				0.981
Yes	223 (79.08)	144 (79.12)	79 (79.00)
Quit attempts for cigarettes (past 12 months)				0.357
Yes	156 (55.32)	97 (53.30)	59 (59.00)
Quit attempts for cannabis (ever)	n=281		n=99	0.378
Yes	112 (39.86)	76 (41.76)	36 (36.36)
Quit attempts for cannabis (past 12 months)				0.836
Yes	44 (15.60)	29 (15.93)	15 (15.00)
Self-Reported Drug Substitution	n (%)	n (%)	n (%)	p-value
Cigarette quit attempt – increase in cannabis use (only participants with any quit attempts)	n=222	n=144	n=78	0.011
Yes	111 (50.00)	63 (43.75)	48 (61.54)
Cannabis quit attempt – increase in cigarette smoking (only participants with any quit attempts)	n=112	n=76	n=36	0.365
Yes	69 (61.61)	49 (64.47)	20 (55.56)
Cannabis and Tobacco Quit Interest	Mean (SD)	Mean (SD)	Mean (SD)	p-value
Interest in quitting smoking cigarettes (1–10 scale)	7.07 (2.90)	7.39 (3.03)	6.49 (2.54)	0.012
Interest in quitting cannabis (1–10 scale)	2.39 (2.35)	2.43 (2.43)	2.30 (2.20)	0.647
Planning on quitting smoking cigarettes (next 30 days)	n (%)n=278	n (%)	n (%)n=96	0.039
Yes	101 (36.33)	74 (40.66)	27 (28.13)
Planning on quitting using cannabis (next 30 days)	n=281		n=99	0.914
Yes	32 (11.39)	21 (11.54)	11 (11.11)

Note. When the number of responses for an item does not equal 100 for the National sample, participants preferred not to answer that question. Percentages include only those participants who answered that question.

Cannabis and Tobacco Quit Interest

Interest in quitting both cannabis and tobacco (assessed via separate questions) are shown in Table 2. Interest in quitting smoking cigarettes was moderately high for both samples (7.07 ± 2.90), while interest in quitting cannabis was generally low (2.39 ± 2.35). When asked about quit intentions in the next 30 days, 36% endorsed plans to quit smoking cigarettes in that time, while 11% endorsed plans to quit using cannabis. Quit intentions varied by survey sample with a greater proportion of participants from the SE sample having plans to quit smoking cigarettes in the next 30 days (p=0.039), and having greater interest in quitting smoking cigarettes compared to the NL sample (p=0.015).

Cessation Preferences

Additional questions regarding cannabis quit history and cessation preferences (SE survey only) are shown in Table 3. Consistent with low reported cannabis quit interest, many participants (78%) responded that they were “extremely unlikely” to seek treatment for cannabis. When asked about readiness and confidence to quit smoking cigarettes and using cannabis, participants had higher levels of readiness (100-point scale) to quit smoking cigarettes compared to quitting cannabis (58.69 vs. 12.98), though they were less confident in their ability to quit smoking cigarettes compared to quitting cannabis (51.37 vs. 65.84). Interest in treatment for both substances was 27.70 ± 31.31, though interest in reducing cannabis use yielded slightly higher interest ratings (34.69 ± 33.31). The majority of participants (71%) had the preference to quit using tobacco first and then quit cannabis, while only 16% had the preference for simultaneous cessation. When asked about how co-users saw their dual use changing in the next year, 59% of the co-using sample said they plan to quit smoking, but still use cannabis, while 29% had no plans to change use of either substance. Finally, interest in treatment methods for both substances was assessed, which was generally low and most participants endorsed they would try to quit on their own (i.e., cold turkey; Figure 1).

Table 3.

Cannabis quit history and cessation preferences for the SE sample only (n=182).

Quit History	N	%
Ever been in treatment for cannabis use (Yes)	8	4.40
Mandatory treatment (vs. voluntary) (n=8) (Yes)	6	75.00
Ever tried to reduce cannabis use (Yes)	103	56.59
Quit Interest	N	%
How likely would you be to seek out treatment for CAN use?
Extremely Unlikely	142	78.02
Somewhat Unlikely	12	6.59
Neutral	16	8.79
Somewhat likely	7	3.85
Extremely likely	5	2.75
(0–100 scales used for the following questions)	Mean	SD
Readiness to quit using cannabis	12.98	22.16
Confidence to quit using cannabis	65.84	34.72
Interest in REDUCING cannabis use	34.69	33.31
Readiness to quit smoking cigarettes	58.69	35.87
Confidence in quitting smoking cigarettes	51.37	33.94
Interest in treatment for BOTH tobacco and cannabis	27.70	31.31
Cessation Preferences
Preferred cessation order for both substances	N	%
Tobacco, then cannabis	129	70.88
Cannabis, then tobacco	24	13.19
Both at the same time	29	15.93
In one year, how do you see your substance use changing?
No change	53	29.12
Stop cigs, still use cannabis	107	58.79
Stop cannabis, still smoke cigs	12	6.59
Quit both	7	3.85
Don’t know	3	1.65

Figure 1.

Percentage of participants endorsing interest in treatment methods for both cannabis and tobacco cessation (n=182). “What treatment methods would you be willing to try if you were quitting both substances?”

Predictors of Quit Attempts and Quit Interest

Demographic, tobacco and cannabis use predictors of past quit attempts and current quit metrics are shown in Table 4. In a multivariate logistic regression model, no demographic or substance use characteristics (as listed in Table 1) were significant predictors of lifetime cannabis quit attempts (full sample) or reduction attempts (SE sample). For past year cannabis quit attempts, only racial minority status (Black/African American and Other) was predictive of a cannabis quit attempt, with racial minorities more likely to have attempted to quit than White respondents (b=1.21, p<0.01). No tobacco use characteristics were predictive of past year cannabis quit attempts.

Table 4.

Unstandardized beta estimates and standard errors for past cannabis and tobacco quit attempts and current quit metrics based on demographic, tobacco, and cannabis use characteristics.

	Demographics							Tobacco			Cannabis
Dependent Variable	Age	Gender	Race	Ethnicity	Marital Status	Education	Employment	Age of regular use	Smoke ≤30 min of waking	Cigarettes per day	Age at first use	Days of use in past month	Times used per day	Smoke ≤30 min of waking
Past Quit Attempts
Cannabis (lifetime) b	0.01 (0.02)	0.50 (0.26)	−0.02 (0.34)	−0.20 (0.43)	−0.15 (0.28)	−0.29 (0.28)	0.29 (0.35)	−0.02 (0.04)	0.37 (0.31)	−0.02 (0.02)	−0.02 (0.03)	0.05 (0.04)	−0.05 (0.05)	0.47 (0.29)
Cannabis (past year) b	0.00 (0.02)	0.01 (0.36)	1.21 (0.41)	−0.10 (0.55)	−0.32 (0.39)	0.53 (0.39)	0.73 (0.58)	0.00 (0.05)	0.75 (0.44)	−0.03 (0.03)	0.00 (0.04)	0.04 (0.05)	−0.25 (0.13)	0.31 (0.40)
Tobacco (lifetime) b	0.03 (0.02)	−0.06 (0.32)	0.38 (0.41)	−1.00 (0.46)	0.10 (0.35)	0.46 (0.35)	0.56 (0.42)	−0.10 (0.04)	0.79 (0.38)	−0.03 (0.03)	−0.06 (0.04)	0.01 (0.04)	−0.07 (0.06)	−0.23 (0.35)
Tobacco (past year) b	0.00 (0.02)	−0.05 (0.26)	0.50 (0.35)	−1.06 (0.42)	−0.08 (0.28)	0.63 (0.28)	0.59 (0.35)	−0.03 (0.04)	0.22 (0.31)	−0.06 (0.02)	−0.03 (0.03)	−0.03 (0.05)	0.01 (0.05)	0.01 (0.29)
Current Quit Metrics
Cannabis
Interest b	−0.05 (0.02)	0.18 (0.27)	0.07 (0.34)	0.24 (0.42)	−0.60 (0.30)	0.21 (0.30)	0.52 (0.39)	0.07 (0.04)	0.90 (0.34)	−0.01 (0.02)	0.01 (0.03)	0.04 (0.04)	−0.15 (0.07)	−0.03 (0.31)
Readiness b†	−0.02 (0.02)	−0.16 (0.34)	−0.59 (0.43)	0.11 (0.60)	−0.15 (0.35)	0.27 (0.35)	−0.23 (0.35)	0.01 (0.04)	0.48 (0.40)	−0.02 (0.03)	−0.02 (0.04)	0.07 (0.05)	−0.18 (0.07)	−0.40 (0.37)
Confidence c†	−0.43 (0.30)	1.42 (5.41)	−16.32 (6.93)	9.57 (9.63)	6.02 (5.51)	−0.26 (5.60)	3.73 (6.62)	0.64 (0.70)	−9.37 (6.29)	0.48 (0.43)	0.88 (0.61)	−0.77 (0.78)	−0.84 (0.87)	−5.60 (5.73)
Tobacco
Interest c	3.10 (2.63)	−45.45 (46.75)	12.68 (60.41)	−35.65 (74.59)	−0.84 (49.61)	20.19 (49.69)	129.64 (61.95)	−3.59 (6.49)	98.13 (55.69)	−4.68 (3.88)	−2.74 (5.85)	7.51 (6.22)	3.58 (9.00)	−77.69 (51.10)
Readiness c†	−0.08 (0.31)	−1.10 (5.60)	2.15 (7.16)	−4.47 (9.96)	6.19 (5.70)	8.44 (5.78)	25.38 (6.84)	0.14 (0.72)	3.65 (6.50)	−0.29 (0.44)	−0.31 (0.63)	0.23 (0.81)	0.28 (0.90)	−12.31 (5.92)
Confidence c†	0.16 (0.24)	4.44 (4.34)	15.33 (5.55)	16.10 (7.71)	1.19 (4.42)	5.00 (4.48)	16.77 (5.30)	0.80 (0.56)	−11.30 (5.04)	−1.36 (0.34)	0.39 (0.49)	−0.35 (0.63)	0.16 (0.70)	−2.07 (4.59)

Note. Significant beta estimates (p<0.05) are shown in bold. All beta estimates are unstandardized for both binary and continuous outcomes, denoted by ^b or ^c, respectively. Standard errors are shown in parentheses following each unstandardized beta estimate. Race (Racial Minorities=1, White=0); Ethnicity (Hispanic/Latinx=1, Not Hispanic/Latinx=0). Education (College or greater=1, Other=0). Employment/student status (Employed/student=1, Unemployed=0). Models controlled for survey sample (if relevant) and income.

^†Southeastern sample only.

For lifetime tobacco quit attempts (full sample), participants who started smoking cigarettes at a younger age were more likely to have made a tobacco cessation attempt (b=−0.10, p=0.02), as were individuals who smoked a cigarette within 30 minutes of waking (b=0.79, p =0.04). Hispanic or Latinx co-users were less likely to have ever made a tobacco quit attempt (b=−1.00, p =0.03). No cannabis use characteristics were associated with lifetime tobacco quit attempts. When examining predictors of past year tobacco quit attempts (full sample), individuals who were college-educated (b=0.63, p =0.03) and smoked fewer cigarettes per day (b=−0.06, p <0.01) were more likely to have made a tobacco cessation attempt. Smokers who identified as Hispanic or Latinx were less likely to have made a tobacco cessation attempt (b=−1.06, p =0.01). No cannabis use characteristics were associated with past year tobacco quit attempts.

In a multivariate logistic regression model, older individuals reported less interest in cannabis cessation (b= −0.05, p <0.01) and individuals who smoked a cigarette within 30 minutes of waking reported greater interest in cannabis cessation (b=0.90, p =0.01) (full sample). In a multivariate logistic regression model, number of times cannabis was used per day was negatively associated with readiness to quit cannabis (b=−0.18, p =0.01) (SE sample). Racial minority status was associated with lower confidence to quit cannabis, with minorities reporting less confidence in their ability to quit cannabis (b=−16.32, p =0.02) compared to White respondents (SE sample). No other demographic or substance use characteristics were predictive of confidence to quit cannabis.

When the same characteristics were used to predict metrics of tobacco quit interest, only employment status predicted interest in tobacco cessation with those who were employed (including students) more interested in tobacco cessation (b=129.64, p =0.04) (full sample). Individuals who were employed (or students) also reported greater readiness to quit tobacco (b=25.38, p <0.01), while individuals who used cannabis within 30 minutes of waking (b=−12.31, p =0.04) reported lower tobacco quit readiness in a multivariate linear regression model (SE sample). Individuals who identified as Hispanic/Latinx (b=16.10, p =0.04), who were employed/students (b=16.77, p <0.01), and identified as racial minorities (b=15.33, p <0.01) were more confident in their ability to quit smoking cigarettes, while those who used tobacco within 30 minutes of waking (b=−11.30, p =0.03) and smoked more cigarettes per day (b=−1.36, p <0.01) were less confident in their ability to quit smoking cigarettes.

Discussion

This is among the first studies in adult cannabis-tobacco co-users to explore quitting intentions and interest, as well as self-reported drug substitution. Quit interest and confidence in quitting was similar to previous studies that have been conducted among young adult co-users (Masters et al., 2018; Ramo et al., 2014). Important findings from the current study include: (1) while nearly all of the sample of co-users were interested in and had attempted to quit smoking cigarettes at least once, a much smaller proportion were interested in or had ever tried to quit using cannabis, and (2) approximately 50–60% of participants perceived increases in their use of the substance they were not trying to quit (cannabis or cigarettes, respectively) during past quit attempts. It may be the case that adult co-users experience a substitution effect, either purposefully (possibly to assist with their cessation) or inadvertently, though this conclusion must be tempered based on the wording of the question used here (i.e., did your use of X increase? Yes or No). A greater proportion of participants reported increases in cigarette smoking during cannabis cessation attempts (compared to cannabis use during tobacco cessation), but fewer participants reported a cannabis cessation attempt at all. While substitution data were self-reported and retrospective, this is a potentially concerning finding from a treatment perspective as interventions targeting only one substance in co-users may have negative unanticipated effects on the other substance. Additional work should be conducted to determine how prevalent drug substitution is, under what conditions it occurs, how long it persists, and whether it increases the risk of relapse.

Among all participants, interest in quitting cannabis was generally low. Multivariate analyses found that individuals who smoked tobacco within 30 minutes of waking (indicative of greater nicotine dependence) were slightly more likely to report interest in cannabis cessation. Lower ratings of interest in cannabis cessation were found among older co-users. Among the SE sample only, very few respondents had ever been in treatment for their cannabis use and many reported being unlikely to seek treatment for cannabis. Results showed that interest in reducing cannabis use was viewed more favorably than quitting and many participants had tried to reduce their cannabis use in the past. Interest in dual cessation treatments was low, which is not surprising given low quit interest for cannabis. Readiness and confidence to quit cannabis were also lower among Hispanic/Latinx respondents and those using cannabis more frequently during the day. However, tobacco quit interest and intention were high in this sample. Indicators of more severe tobacco use (i.e., earlier age of onset, smoking within 30 minutes of waking) were associated with a greater likelihood of past quit attempts, but lower confidence in one’s ability to quit. Individuals who identified as Hispanic/Latinx were less likely to have made a tobacco cessation attempt, but more confident in their ability to abstain from tobacco. Education and employment appeared to be protective, predicting quit attempts and interest in tobacco cessation, respectively. It is important to note that no cannabis use characteristics predicted lower rates of lifetime or past year tobacco quit attempts. Cannabis use within 30 minutes of waking was negatively associated with readiness to quit tobacco. These results suggest that tobacco quit interest and intention are high and may only be slightly influenced by more severe cannabis use characteristics, which should be taken into consideration when implemented tobacco cessation treatment.

One reason for the lack of interest in quitting cannabis may be the perception that cannabis is low-risk, and even beneficial, compared with perceptions that tobacco is harmful. In both study samples, approximately 10% of participants reported that they were using cannabis for medical reasons, while another 42% reported using it for both recreational and medical purposes (Salazar, Tomko, Akbar, Squeglia, & McClure, Under Review) (SE survey data only; NL data unpublished). These study results are consistent with trends in the US toward reduced perception of harm and expanded accessibility of cannabis (Johnston, O’Malley, Meiech, Bachman, & Schulenberg, 2015; Pacek, Mauro, & Martins, 2015), which may be contributing to increasing rates of use (Hasin et al., 2016). This shift is occurring at the same time that tobacco use rates continue to decline among adults (Jamal et al., 2016), but while electronic products are becoming more common and sophisticated. As policies legalizing recreational and medical cannabis become more widespread, more research is warranted to better understand actual and perceived harms from cannabis, and what motivates cannabis users to attempt to quit. The impact of cannabis legislation is still largely unknown, but it may contribute to an increase in heavy use and cannabis-related harms (Hall & Lynskey, 2016; Pacula & Lundberg, 2014) and needed treatment resources for those interested in quitting or reducing their cannabis use.

The product landscape and commercial availability for both cannabis and tobacco/nicotine products is rapidly changing and also likely influencing the co-use relationship. There is massive variation in cannabis products and methods of administration. Edibles, concentrates (e.g., wax, dabs), and vaporizers allow for cannabis use through administration methods that do not involve combustion and inhalation, potentially distinguishing them from tobacco. The popularity of electronic cigarettes and vaporizers for nicotine delivery may also be affecting the co-use relationship. Specific to this survey, it is possible that survey respondents had switched to using electronic cigarettes or used multiple tobacco/nicotine products, but did not consider themselves a smoker.

The current report is focused on historic and current quit metrics among cannabis and tobacco co-users. Not included in this report is detailed information about patterns and reasons for co-use among this sample. These data also have important treatment implications for co-users. As part of the SE survey, additional questions were asked regarding patterns and reasons for co-use, and those data have been published elsewhere (Akbar, Tomko, Salazar, Squeglia, & McClure, Under Review) to allow for sufficient presentation of data and interpretations. Along with the information presented in the current report, patterns and reasons for co-use are important to understand in the development of treatment strategies, specifically when there are individual differences in use patterns that may affect treatment and successful cessation.

Limitations

This study has several limitations. First, both surveys were conducted through an online crowdsourcing program (Amazon MTurk), which may not accurately represent the adult tobacco cannabis co-using population and may not generalize to other studies of co-users. Though sample representativeness is a limitation, MTurk has been demonstrated to be a valuable platform for data collection with adequate representation for some populations, and may even provide superior sample representation compared to convenience samples (Berinsky, Huber, & Lenz, 2012; M. Buhrmester, Kwang, & Gosling, 2011; M. D. Buhrmester, Talaifar, & Gosling, 2018; Dworkin, Hessel, Gliske, & Rudi, 2016). The participants recruited for this survey represent a population that is not currently engaged in treatment for cannabis, and are unlikely to seek treatment in the near future, which makes this recruitment source of value in answering these research questions. Second, these studies were cross-sectional, anonymous surveys and no biochemical verification of cannabis or tobacco use was conducted to ensure that participants were co-users. Also, while many questions captured current quit interest and preferences, many questions were retrospective in nature (e.g., drug substitution) and may be subject to recall bias. Third, there were slight variations across the survey study designs, items, response options, and inclusion criteria. Some items were included in the SE survey as a result of inspecting the data from the NL survey and forming additional research questions. Though some differences existed between these two samples, many responses were similar across survey samples and these variations were unlikely to have significantly impacted study results. Finally, the two surveys used in this report were administered over one year apart (April 2016 vs. June-August 2017). This delay in survey administration may have contributed to differences between samples, possibly due to rapidly shifting perceptions of harm associated with cannabis and tobacco use, rather than regional differences.

Conclusions

While interest in tobacco cessation was high in this sample, interest and readiness to quit using cannabis was low and interest in dual cessation interventions was also low. Drug substitution when quitting one substance is a relevant treatment concern and while substitution data from the current study are limited, this is an area that warrants further investigation. Additionally, participants endorsed a greater preference to quit smoking cigarettes first and then quit using cannabis, rather than simultaneous cessation. Tobacco use should be addressed among this population, but interventions may need to consider the lack of interest in cannabis cessation in some co-users. Motivational enhancement strategies and discussion of increased risk of relapse due to co-use are also worth exploring as a component of treating co-users, in addition to reduction-based or harm reduction strategies for cannabis use. Harm reduction strategies for cannabis may result in improved functioning (e.g., improvements in depression, anxiety, interpersonal relationships, etc.) and preliminary data exist to support the relationship between reductions in cannabis use and improved functional outcomes (Hser et al., 2017). It may be worthwhile for cannabis treatment to focus on the most harmful and problematic behavior and/or consequences of cannabis use, encourage lower-risk use practices (Fischer et al., 2017), and consider non-abstinence outcomes as a possible alternative (Sahlem, Tomko, Sherman, Gray, & McRae-Clark, 2018). Additionally, what constitutes a meaningful outcome for substance use disorder treatment is a topic of current debate in the field (Kiluk, Fitzmaurice, Strain, & Weiss, 2018; Kiluk, Frankforter, Cusumano, Nich, & Carroll, 2018). Specific to a tobacco-cannabis co-using population, special consideration should be given to cessation preferences and quit interest for each substance. Harm reduction strategies and defining meaningful outcomes of success for co-users may be an important area to further explore.

Public Significance Statement:

The co-use of cannabis and tobacco is common practice, yet little is known about quit interest, treatment preferences, and drug substitution during past cessation attempts among adult co-users. Understanding these factors are critical to guide the development of efficacious and acceptable treatment strategies for co-users.