A Longitudinal Analysis of Nicotine Dependence and Transitions From Dual Use of Cigarettes and Electronic Cigarettes: Evidence From Waves 1–3 of the PATH Study

Abstract

Objective:

Electronic cigarettes (e-cigarettes) have gained popularity as a method to reduce conventional cigarette smoking, despite mixed evidence on their effectiveness. This study evaluates the relationship between overall and product-specific nicotine dependence and the transitions between dual use of cigarettes and e-cigarettes versus exclusive cigarette or e-cigarette use over time.

Method:

This study used data from Waves 1–3 (2013–2016) of the Population Assessment of Tobacco and Health (PATH) Study. Weighted logistic regressions with person-level random effects tested relationships between nicotine dependence and dual versus exclusive use over time. Dual use transitions were then compared with the characteristics of e-cigarette devices used and reasons to use them.

Results:

Higher tobacco dependence was associated with becoming or remaining a dual user rather than remaining or becoming an exclusive user of cigarettes or e-cigarettes (p < .05). Higher e-cigarette dependence was associated with remaining or becoming an exclusive e-cigarette user. The number of days smoking cigarettes or using e-cigarettes in the past 30 days was also associated with greater odds of remaining or transitioning to exclusive use of that product (p < .05). Exclusive e-cigarette users tended to invest more financially in their devices and were more likely to report owning modifiable devices.

Conclusions:

This study provides new evidence that established dual use and transitions to and from dual use are associated with higher tobacco dependence compared with remaining a cigarette- or e-cigarette-only user and that higher e-cigarette dependence is associated with becoming or remaining an exclusive user of e-cigarettes.

The prevalence of conventional cigarette smoking reached a historical low with 13.7% of U.S. adults smoking in 2018 (Creamer et al., 2019); however, use of electronic (e-) cigarettes has been increasing among both youth and adults (Cullen et al., 2018; Foxon & Selya, 2020), raising questions about whether e-cigarettes have the potential to help reduce the public health burden of combustible cigarettes. Current e-cigarette use among adults declined from 3.7% in 2014 to 2.8% in 2017, then increased to 3.2% in 2018 (Dai & Leventhal, 2019), and e-cigarette use among young adults (ages 18–24) increased from 4.7% in 2016 to 7.6% in 2018 (Dai & Leventhal, 2019). Further, use of cigarettes and e-cigarettes are closely related. The majority of current e-cigarette users (59%) are also past-30-day cigarette smokers, referred to as dual users, whereas 30% are former smokers who transitioned to exclusive e-cigarette use (Schoenborn & Gindi, 2015). A recent study of dual users found that after 1 year, 44% reverted to smoking alone while 49% remained dual users of both products (Piper et al., 2019). Dual use may maintain exposure to toxicants in cigarettes while adding the potential harms of e-cigarette use (National Academies of Sciences, Engineering, and Medicine, 2018; Reidel et al., 2018). The ability to identify and transition dual users to lower harm tobacco products or no tobacco products at all has substantial implications for tobacco control.

A recent longitudinal study revealed that 35% of adult smokers who made a quit attempt during the study period used e-cigarettes to try to reduce smoking (Caraballo et al., 2017). Although e-cigarettes have not been approved as a smoking cessation aid (U.S. Department of Health & Human Services, 2016), a recent study found that their use promoted a higher abstinence rate among smoking participants than nicotine replacement therapy (Hajek et al., 2019). Evidence that e-cigarettes can successfully help smokers maintain long-term abstinence remains mixed (Berry et al., 2019; Selya et al., 2018), and a recent Surgeon General’s report on cessation deems the evidence “suggestive not sufficient” that use of nicotine-containing e-cigarettes may be associated with increased cessation (U.S. Department of Health and Human Services, 2020). Furthermore, evidence suggests that use of e-cigarettes among those naive to conventional cigarette use increases the likelihood of initiating and maintaining smoking in the future (Leventhal et al., 2015; U.S. Department of Health & Human Services, 2016), although this may be attributable, in part, to shared risk factors for use of both products (Kim & Selya, 2020). Hence, transitions to and from exclusive and dual use are common and concerning.

Difficulty reducing or quitting cigarette smoking is attributable, in large part, to the nicotine in cigarettes sustaining addiction by enhancing mood and countering withdrawal symptoms through continued smoking (Benowitz, 2010). Nicotine dependence influences how smokers use cigarettes and e-cigarettes, including the number of cigarettes smoked per day (Buu et al., 2018). There is limited research on whether nicotine dependence predicts transitions to and from dual use over time, or transitions over time from dual use to potentially less harmful, exclusive use of e-cigarettes. In addition, validated scales have only recently been developed to assess dependence among e-cigarette users (Strong et al., 2017). Importantly, dual users can distinguish between their dependence on each product, and product-specific dependence has been associated with increased use (Morean et al., 2018).

Complicating the question of dual use transitions further, there has been significant innovation in the market for e-cigarettes including the proliferation of customizable devices, further influencing reasons to use e-cigarettes. Adult e-cigarette users have cited reasons to use these products, such as reducing or quitting smoking and using them in places where cigarettes are not allowed (Nicksic et al., 2019; Patel et al., 2016). Additionally, e-cigarette device characteristics such as size and ability to recharge batteries or refill e-liquid may be related to frequency of use (Buu et al., 2018; Coleman et al., 2019). With a variety of device features and possible motives for initiating e-cigarette use, it is likely that smokers transitioning to and from e-cigarette use may exhibit variation in device characteristics they find desirable.

Understanding the associations between nicotine dependence and dual use is necessary to tailor interventions to decrease harms associated with tobacco use. Using three waves of data from the Population Assessment of Tobacco and Health (PATH) Study, we extend the existing evidence by evaluating the relationship between nicotine dependence and transitions between three common choices facing smokers: remaining or becoming a cigarette-only smoker, remaining or becoming an established dual user, and remaining or becoming an e-cigarette-only user. We were unable to examine tobacco cessation as an outcome because of the limited number of participants achieving quitting in the analytic sample. However, we examined whether variation in e-cigarette device characteristics and reasons to use e-cigarettes were associated with transitions in dual use. Our hypotheses were (a) among established tobacco users, higher nicotine dependence would be associated with higher likelihood of remaining or becoming a dual user rather than an exclusive user of either product and (b) preferences regarding device characteristics would differ across individuals who shifted to exclusive use of e-cigarettes by Wave 3 rather than remaining dual users.

Method

Data source and sample

Data were obtained from Waves 1 (October 2013–December 2014), 2 (October 2014–October 2015), and 3 (October 2015–October 2016) of the Population Assessment of Tobacco and Health (PATH) Study, a longitudinal, nationally representative cohort study of 32,320 adult smokers and nonsmokers. The PATH sample was selected based on a multistage area probability sample design. The sample was designed to generalize to the noninstitutionalized U.S. population using survey weights. The weighted Wave 1 to Wave 3 response rate for adult surveys was 78% (Westat, 2019). Further information on the design and methods of the PATH Study is provided elsewhere (Hyland et al., 2016).

Measures

Dual use groups.

Our analytic sample consisted of smokers who were current, established users of either cigarettes, e-cigarettes, or both products during Wave 1 data collection and had smoked or used e-cigarettes within the past 30 days at follow-up.

Current, established cigarette smoking was defined in PATH as those who currently smoke cigarettes every day or some days, have smoked at least 100 cigarettes in their lifetime, and have smoked cigarettes “regularly,” in comparison to “experimental” use, defined in PATH as current use without having met the lifetime cigarette threshold for regular use.

Current, established e-cigarette use was defined in PATH as currently using e-cigarettes every day or some days and having regularly used e-cigarettes. There is no lifetime consumption threshold in PATH for defining established e-cigarette use, and “regularly” is not specifically defined for either product. Dual use was defined in our study as current, established use of both products, a definition that is consistent, although not universal, in the literature (Jorenby et al., 2017; Meltzer et al., 2017). We focused on a sample of established users of either or both products in Wave 1, to isolate regular product users rather than including those who might simply be experimenting with either product but had not progressed to the level of dependence observed among regular users.

Sample participants were defined as exclusively current, established cigarette users; exclusively current, established e-cigarette users; or current, established dual users of both products in Wave 1. In Wave 1 the study sample consisted of 6,946 participants (representing a U.S. population [N] = 33,956,399), of whom:

• (a) 5,922 (N = 29,086,352) were cigarette-only smokers (i.e., current, established cigarette smokers who were not current established, experimental, or past-30-day users of e-cigarettes);

• (b) 748 (N = 3,500,149) were current, established users of conventional and e-cigarettes; and

• (c) 276 (N = 1,369,898) were e-cigarette-only users (current, established e-cigarette users who were not current established, experimental, or past-30-day users of conventional cigarettes).

Cigarette and e-cigarette nicotine dependence.

Nicotine dependence was assessed in each wave using 12 items from the Wisconsin Inventory of Smoking Dependence Motives (WISDM; Strong et al., 2017). All participants who smoked cigarettes were asked to rate their dependence on tobacco (referred to as “tobacco dependence” from here forward), and participants who were dual users or e-cigarette-only users were asked to repeat the scale and provide answers specific to their dependence on e-cigarettes (“e-cigarette dependence”). All 12 variables were measured using a five-item Likert scale from not true of me at all to extremely true of me. A validation study of nicotine dependence scales among dual users indicated that the “e”-WISDM has high internal consistency and is highly correlated with validation criteria among this population (Piper et al., 2020).

Participants received a total tobacco and/or e-cigarette nicotine dependence score in each wave based on the average of their responses to all 12 items (range: 1–5). Participants with higher average scores were considered to have higher nicotine dependence. The Cronbach’s alpha score for reliability for the tobacco dependence scale was .93, and .86 for e-cigarette dependence items.

Cigarette and e-cigarette smoking behavior.

Cigarette smokers were asked the average number of cigarettes per day (CPD) among everyday smokers and for the past 30 days among some-days smokers in both waves. Based on the distribution of this measure, we used a log transformation of CPD as a control variable in our models predicting dual user groups (Oncken et al., 2020). Frequency of cigarette smoking was measured by asking some-days smokers how many days they smoked within the past 30 days and assigning 30 days to everyday smokers in all waves. Dual users and e-cigarette-only users answered similar questions about the number of e-cigarettes or e-cigarette cartridges they used on days they used e-cigarettes in the past 30 days, and the number of days they used e-cigarettes in the past 30 days. The number of times an e-cigarette was used was similarly transformed (Huang et al., 2015).

Covariates. Additional behaviors associated with variation in nicotine dependence (Buu et al., 2018) were measured in each wave: 30-day use of other tobacco products (cigars, cigarillos, pipe tobacco, hookah, smokeless products, snus, and dissolvable tobacco), and past-30-day use of alcohol and marijuana.

Sociodemographic covariates associated with variation in smoking and/or e-cigarette use behaviors (Du et al., 2019; U.S. Department of Health & Human Services, 2014) were sex, age (18–24, 25–34, 35–44, 45–54, 55–64, and ≥65), race/ethnicity (non-Hispanic White, non-Hispanic African American, Hispanic, and other), education (high school/General Educational Development [GED] credential or less, some college, and bachelor’s degree or higher), and annual household (HH) income (<$10,000, $10,000–$24,999, $25,000–$49,999, $50,000–$99,999, and ≥$100,000).

E-cigarette device characteristics and reasons to use.

The following questions (with yes/no responses unless noted) were added in Waves 2 and 3 to determine e-cigarette device characteristics and reasons to use e-cigarettes: owning a device, price paid (less than $10, $10–$20, $21–100, more than $100), device is rechargeable, uses a tank system, uses cartridges, able to refill device or cartridges with e-liquid, contain nicotine, can change voltage, to quit tobacco altogether, to cut down on cigarettes, and to use when could not smoke cigarettes in the past 30 days.

Statistical analysis

Sample characteristics were described using weighted univariate statistics. Weighted bivariate statistics were used to test for unadjusted differences between sample characteristics and dual use groups. Two different random-effects logistic regressions estimated adjusted odds ratios (AOR) to test for associations between nicotine dependence in each wave and dual use transitions. In both models, dual users were the comparison group, as likelihood of remaining a dual user versus transitioning from dual use to a potentially lower harm condition (exclusive use of e-cigarettes) is of paramount interest to inform policy. Model 1 focused on transitions between exclusive use of cigarettes and remaining or becoming a dual user of both products across waves, whereas Model 2 provided estimates describing transitions between exclusive use of e-cigarettes and becoming or remaining a dual user over time. As the sample size for individuals making transitions between waves remains low, a random-effects, rather than fixed-effects, model allowed us to use information available both within and between subjects for parameter estimates (Wooldridge, 2014). We assessed the degree of multicollinearity between tobacco use outcomes, nicotine dependence, and use behaviors (cigarettes/e-cigarettes per day, number of days used) using variation inflation factor (VIF) analysis and our measure scores fell below the threshold of concern (VIF: 10, Tolerance: 0.1; Wooldridge, 2014). Weighted frequencies and Pearson’s chi square tested bivariate associations between e-cigarette device characteristics in Wave 2, reasons to use e-cigarettes, and transitions to and from dual use between Waves 2 and 3. The probability of making a type I error of .05 was used for all statistical tests. Analyses were performed using Stata Version 15.0 (StataCorp LP, College Station, TX).

Results

Sample description and unadjusted associations between dual use groups and participant characteristics in Wave 1

The analytic sample in Wave 1 was predominantly non-Hispanic White (69.3%), between ages 25 and 54 (63.5%), earned a high school education or less (54.2%), and had an annual HH income between $10,000 and $49,999 (52.9%; Table 1). Statistically significant differences were observed between dual user groups and age, race/ethnicity, education, and annual HH income such that dual users and exclusive e-cigarette users tended to be younger and White, with higher levels of educational attainment and annual income (p < .05, all). The sample average score for tobacco dependence was 2.76 and 1.79 for e-cigarette dependence. These both varied across dual user groups, where tobacco dependence was highest in dual users (M = 3.01) and e-cigarette dependence was highest among e-cigarette-only users (M = 2.15, both p < .01). In the analytic sample of user groups, the mean number of CPD was 10.5 in Wave 1, and the mean number of days smoked was 26.34. Compared with dual users, cigarette-only smokers smoked more days within the past 30 days (26.4 vs. 25.9, respectively, p = .02), whereas e-cigarette-only users reported using e-cigarettes almost 10 days more than dual users (24.3 vs. 14.7, respectively, p < .01).

Table 1.

Weighted descriptive characteristics by dual use groups in Wave 1 of the Population Assessment of Tobacco and Health (PATH) Study

Variable	Total sample (n = 6,946, Na = 33,956,399) % [95% CI]	Current, established cigarette smokers (n = 5,922, N = 29,086,352) % [95% CI]	Dual users (n = 748, N = 3,500,149) % [95% CI]	E-cigarette-only users (n = 276, N = 1,369,898) % [95% CI]
Tobacco nicotine dependenceb	2.76 [2.72, 2.79]	2.77 [2.74, 2.81]	3.01 [2.93, 3.09]	1.85 [1.75, 1.94]
E-cigarette dependenceb	1.79 [1.73, 1.85]	.–	1.76 [1.70, 1.83]	2.15 [2.03, 2.28]
Number of days smokedc	26.36 [26.08, 26.65]	26.42 [26.10, 26.73]	25.90 [25.16, 26.65]	–
Number of cigarettesc smoked	10.50 [10.20, 10.80]	10.49 [10.18, 10.79]	10.62 [9.79, 11.45]	–
Number of days used e-cigarettesc	17.41 [16.53, 18.30]	.–	14.72 [13.69, 15.75]	24.30 [22.90, 25.70]
E-cigarettes usedc	2.32 [1.77, 2.87]	–	2.56 [1.76, 3.36]	1.81 [1.51, 2.11]
Past-30-day other tobacco use
Yes	23.1 [21.9, 24.3]	22.0 [20.8, 23.3]	32.7 [29.5, 36.0]	20.3 [15.6, 26.0]
No	76.9 [75.7, 78.1]	78.0 [76.7, 79.2]	67.3 [64.0, 70.5]	79.7 [74.0, 84.4]
Past-30-day alcohol
Yes	57.9 [56.5, 59.3]	57.7 [56.3, 59.1]	60.9 [56.3, 65.2]	53.4 [46.5, 60.1]
No	42.1 [40.7, 43.5]	42.3 [40.9, 43.7]	39.1 [34.8, 43.7]	46.6 [39.9, 53.5]
Past-30-day marijuana
Yes	18.7 [17.5, 20.01]	18.5 [17.2, 19.8]	22.6 [19.8, 25.7]	13.2 [9.4, 18.2]
No	81.3 [80.0, 82.5]	81.5 [80.2, 82.8]	77.4 [7.4, 80.2]	86.8 [81.8, 90.7]
Male	0.6 [0.5, 0.6]	0.6 [0.5, 0.6]	0.5 [0.5, 0.6]	0.5 [0.4, 0.6]
Age
18–24	12.5 [11.8, 13.3]	11.5 [10.7, 12.4]	19.4 [16.8, 22.2]	16.3 [12.5, 20.9]
25–34	23.9 [22.9, 25.0]	23.1 [22.0, 24.2]	30.2 [26.5, 34.1]	26.3 [20.4, 33.2]
35–44	19.0 [17.9, 20.2]	19.1 [18.0, 20.2]	19.7 [16.1, 24.0]	16.0 [12.1, 21.0]
45–54	20.6 [19.6, 21.6]	21.2 [20.1, 22.3]	15.5 [12.7, 18.9]	20.6 [15.4, 27.1]
55–64	16.0 [15.0, 17.1]	16.7 [15.5, 18.0]	11.1 [8.6, 14.2]	13.0 [9.0, 18.4]
≥65	7.9 [7.1, 8.8]	8.4 [7.5, 9.3]	4.1 [2.4, 6.9]	7.7 [4.7, 12.6]
Race/ethnicity
White	69.3 [68.0, 70.6]	67.9 [66.4, 69.3]	78.4 [75.1, 81.3]	76.5 [70.1, 81.9]
African American	14.1 [13.3, 14.9]	15.4 [14.5, 16.3]	6.8 [5.1, 9.0]	5.5 [3.4, 8.9]
Hispanic	10.8 [10.1, 11.6]	11.2 [10.3, 12.1]	8.3 [6.4, 10.6]	10.2 [7.1, 14.5]
Other	5.8 [5.2, 6.4]	5.6 [4.9, 6.3]	6.6 [5.0, 8.7]	7.8 [4.7, 12.6]
Education
High school/GED or less	54.2 [52.7, 55.7]	56.3 [54.8, 57.8]	42.7 [38.7, 46.8]	39.0 [32.4, 46.1]
Some college	34.0 [32.7, 35.3]	32.6 [31.2, 33.9]	41.6 [37.3, 46.0]	44.7 [38.5, 51.0]
Bachelor’s degree or higher	11.8 [11.0, 12.7]	11.1 [10.3, 12.1]	15.7 [12.8, 19.2]	16.3 [11.8, 22.2]
Annual household income
<$10,000	20.5 [19.3, 21.7]	21.6 [20.3, 23.0]	15.3 [12.6, 18.5]	9.8 [6.6, 14.2]
$10,000–$24,999	27.5 [26.2, 28.9]	27.9 [26.4, 29.4]	26.7 [23.0, 30.8]	21.4 [15.4, 29.0]
$25,000–$49,999	25.4 [24.2, 26.6]	25.0 [23.8, 26.3]	26.6 [22.7, 30.9]	29.6 [23.4, 36.7]
$50,000–$99,999	19.4 [18.3, 20.6]	18.8 [17.5, 20.1]	21.7 [18.6, 25.2]	28.2 [22.7, 34.5]
≥$100,000	7.2 [6.5, 8.1]	6.8 [6.0, 7.6]	9.7 [7.2, 12.9]	11.0 [7.0, 16.7]

Notes: Pearson’s chi-squared tests and t tests used to determine p values. CI = confidence interval; GED = General Educational Development credential.

^aN represents the population of U.S. adults to which the sample generalizes;

^brepresents average Wave 1 nicotine dependence score;

^call smoking behavior measures refer to past-30-day use; cigarettes smoked and e-cigarettes used refer to average consumption on days smoking/using e-cigarettes in the past 30 days; weighted sample means and standard errors reported.

Adjusted associations between nicotine dependence, smoking behavior, and dual use

A higher tobacco nicotine dependence score was negatively associated with remaining/becoming a cigarette-only user (AOR = 0.76, p < .01) or e-cigarette-only user (AOR = 0.26, p < .01) compared with dual use, across waves, after adjusting for covariates (Table 2). In contrast, e-cigarette nicotine dependence was positively associated with remaining/becoming an e-cigarette-only user (AOR = 1.63, p = .02). The number of days smoking cigarettes and using e-cigarettes in the past 30 days were each positively associated with remaining/becoming an exclusive user of that product (AOR cigarette-only use = 1.03, p < .01; AOR e-cigarette-only = 1.05, p = .01). Past-30-day other tobacco use was negatively associated with remaining/becoming a cigarette-only smoker compared with a dual user (AOR = 0.44, p < .01), as were past-30-day marijuana use, higher levels of education, and increased annual HH income (p < .05, all).

Table 2.

Associations between nicotine dependence and dual use group transitions across Waves 1, 2, and 3 of the Population Assessment of Tobacco and Health (PATH) Study

Variable	Cigarette-only smokinga AOR [95% CI] (n = 6,690)	E-cigarette-only useb AOR [95% CI] (n = 1,026)
Tobacco nicotine dependence	0.76 [0.69, 0.83]	0.26 [0.14, 0.48]
E-cigarette nicotine dependence	–	1.63 [1.09, 2.42]
Number of days smoked/used (past 30 days)c	1.03 [1.02, 1.04]	1.05 [1.01, 1.09]
Cigarettes/e-cigarettes used per day (past 30 days)d	0.96 [0.89, 1.05]	1.00 [0.72, 1.38]
Past-30-day other tobacco use	0.44 [0.36, 0.53]	0.81 [0.42, 1.54]
Past-30-day alcohol use	0.87 [0.82, 1.39]	0.91 [0.47, 1.75]
Past-30-day marijuana use	0.50 [0.42, 0.61]	0.58 [0.32, 1.08]
Male	1.23 [1.02, 1.47]	0.87 [0.48, 1.60]
Age
18–24	Ref.	Ref.
25–34	1.21 [0.95, 1.53]	1.41 [0.68, 2.94]
35–44	1.60 [1.22, 2.11]	1.95 [0.76, 4.98]
45–54	2.26 [1.68, 3.04]	4.44 [1.40, 14.09]
55–64	2.62 [1.88, 3.67]	1.86 [0.53, 6.56]
≥65	4.98 [2.71, 9.16]	2.12 [0.38, 11.89]
Race/ethnicity
White	Ref.	Ref.
African American	3.39 [2.48, 4.63]	1.20 [0.39, 3.71]
Hispanic	1.30 [0.97, 1.72]	1.51 [0.63, 3.63]
Other	0.97 [0.70, 1.34]	0.66 [0.23, 1.86]
Education
High school/GED or less	Ref	Ref
Some college	0.74 [0.61, 0.90]	1.44 [0.75, 2.80]
Bachelor’s degree or higher	0.69 [0.51, 0.93]	0.64 [0.24, 1.67]
Annual household income
Less than $10,000	Ref	Ref
$10,000–$24,999	0.72 [0.57, 0.91]	0.90 [0.37, 2.16]
$25,000–$49,999	0.76 [0.59, 0.98]	1.07 [0.43, 2.64]
$50,000–$99,999	0.73 [0.54, 0.97]	0.88 [0.33, 2.33]
≥$100,000	0.72 [0.48, 1.07]	1.67 [0.51, 5.47]

Notes: Bolded p values indicate statistical significance < .05. AOR = adjusted odds ratio; CI = confidence interval; GED = General Educational Development credential; ref. = reference.

^aA random effects logistic regression tested associations between cigarette-only smokers and dual use with dual users acting as the reference group;

^ba random effects logistic regression tested associations between e-cigarette-only users and dual use with dual users acting as the reference group;

^cnumber of days smoked/used refers to the number of days a participant either smoked cigarettes or used e-cigarettes in the past 30 days;

^dcigarettes per day (CPD) or e-cigarettes per day on days used in the past 30 days is the natural log value of these two measures.

Relative to participants in the 18–24 age group, older age was associated with higher odds of remaining/becoming a cigarette-only smoker compared with a dual user, an association that showed a trend with increasing age. Compared with non-Hispanic Whites, being African American was positively associated with remaining/becoming a cigarette-only smoker.

E-cigarette device characteristics and reasons to use among Wave 2 dual users and Wave 3 user groups

To determine use behaviors of dual users 1 year later, we analyzed weighted proportions and bivariate associations between reasons to use e-cigarettes among dual users in Wave 2 and user groups (remained dual users or transitioned to either e-cigarettes or cigarettes) in Wave 3. Among dual users in Wave 2, the majority owned a device (68.8%), had a rechargeable device (84.9%), used a device with a tank system (93.3%), used an e-cigarette that contains nicotine (91.0%), and used e-cigarettes to cut down on cigarettes (75.6%) (Table 3). Dual users in Wave 2 who became cigarette-only users in Wave 3 had significantly lower prevalence of owning a device, ability to refill device or cartridge with e-liquid, changing voltage, and using an e-cigarette to quit tobacco use and in the past 30 days when they could not smoke compared with dual users in Wave 3 (p < .05, each). In addition, cigarette-only users in Wave 3 had a higher prevalence of paying less than $10 for an e-cigarette device and using a device that uses cartridges (p < .001).

Table 3.

Weighted proportions and bivariate associations of Wave 2 device characteristics and reasons to use e-cigarettes and Wave 3 e-cigarette and cigarette user groups among Wave 2 dual users in the Population Assessment of Tobacco and Health (PATH) Study (n = 2,239; N = 10,073,264)

Variable	Total	Dual user in Wave 3 % (n = 1,026)	Cigarette user only in Wave 3 % (n = 1,078)	E-cigarette user only in Wave 3 % (n = 135)	%p a
Own device	68.8	73.8	63.5	73.5	<.001
Price paid					<.001
< $10	27.4	24.5	32.9	10.8
$10–$20	23.9	23.2	25.4	19.5
$21–$100	41.1	41.5	38.2	56.9
> $100	7.6	10.8	3.5	12.8
Rechargeable	84.9	84.0	84.4	94.3	.04
Uses a tank system	93.3	92.3	94.5	94.5	.60
Uses cartridges	46.9	43.0	54.7	26.4	<.001
Able to refill e-liquid	64.7	67.8	57.9	87.2	<.001
Contain nicotine	91.0	90.9	92.0	87.8	.56
Can change voltage	40.4	44.0	33.4	55.5	<.01
Used to quit tobacco altogether	60.8	62.4	57.0	73.9	<.01
Used to cut down cigarettes	75.6	77.6	72.8	80.7	.09
Used when could not smoke cigarettes in past 30 days	65.3	72.6	58.1	72.8	<.001

Notes: Bolded p values indicate statistical significance <.05.

^aPearson’s chi-squared tests used to determine p values.

Changes in e-cigarette device characteristics from Wave 2 to Wave 3

To further determine what e-cigarette device characteristics dual users in Wave 2 have 1 year later, we analyzed any change in device characteristic with those who remained dual users or transitioned to e-cigarette-only use in Wave 3. The most prevalent change made in device characteristics was price paid (47.7%), followed by using cartridges (31.7%), and can change voltage (28.1%; Table 4). Compared with those who remained dual users in Wave 3, dual users who transitioned to e-cigarette use only in Wave 3 had lower prevalence of changing whether one owns a device (p = .04), having a rechargeable device (p < .01), and whether they could refill e-liquid (p = .02).

Table 4.

Weighted proportions and bivariate associations of changes in e-cigarette device characteristics between Waves 2 and 3 and dual use versus exclusive e-cigarette use in the PATH Study (n = 1,161; N = 5,181,232)^a

Variable	Total %	Dual user in Wave 3 % (n = 1,026)	E-cigarette user only in Wave 3 % (n = 135)	pb
Own device	26.2	27.3	18.5	.04
Price paid	47.7	45.1	46.3	.78
Rechargeable	13.8	15.1	4.8	<.01
Uses a tank system	15.2	15.4	14.3	.88
Uses cartridges	31.7	32.7	25.3	.31
Able to refill e-liquid	16.1	17.2	7.5	.02
Contain nicotine	12.0	11.4	15.4	.67
Can change voltage	28.1	28.0	28.4	.99

Notes: Bolded p values indicate statistical significance <.05.

^aChange in e-cigarette device was measured as a change in response from no to yes or yes to no (with exception to any change in price categories) from Wave 2 to Wave 3;

^bPearson’s chi-squared tests used to determine p values.

Discussion

In contrast to declining trends in cigarette use, e-cigarette use has been increasing (Creamer et al., 2019), particularly among current adult smokers, including those trying to quit smoking (Hu et al., 2019; Mirbolouk et al., 2018).

Cigarette smokers who begin using e-cigarettes to reduce or quit cigarettes, as well as those who use e-cigarettes to substitute for cigarettes when it is more convenient to do so, might be dual users of both products for brief or extended periods and might transition back and forth between dual use and exclusive cigarette use (Malas et al., 2016; Pokhrel et al., 2015). Nicotine dependence is likely a critical predictor of patterns of dual versus exclusive use of cigarettes or e-cigarettes (Jorenby et al., 2017; Rostron et al., 2016). Yet, whether nicotine dependence predicts transitions to dual use and continued dual use over time has thus far been poorly understood. In addition, the variety of e-cigarette devices now available in the U.S. market requires further study regarding the relationship between characteristics of e-cigarette devices and use patterns over time.

This study tested whether tobacco and/or e-cigarette nicotine dependence were each associated with changes in dual use group membership in Waves 1–3. Individuals with higher levels of tobacco dependence were more likely to remain or become dual users than to transition to exclusive use of cigarettes or e-cigarettes over time. This evidence is supported by experimental laboratory data that dual users have higher levels of nicotine in their blood (Jorenby et al., 2017) and by nationally representative data on daily dual users who reported greater craving and withdrawal than cigarette-only smokers (Rostron et al., 2016). Furthermore, individuals reporting higher e-cigarette nicotine dependence were more likely to remain or become e-cigarette-only users over time rather than dual users. Frequency of cigarette use (number of days used within the past 30 days)—another measure of dependence—was also significantly associated with remaining or becoming a cigarette-only user rather than a dual user, and, likewise, frequency of e-cigarette use was associated with remaining or becoming an e-cigarette-only user rather than a dual user. These findings reinforce previous cross-sectional evidence regarding product dependence and patterns of use among dual users (Morean et al., 2018) and lend support to previous results from a longitudinal study that dual users who became “intensive” users of e-cigarettes, that is, used them daily for at least 1 month, were more likely to become exclusive e-cigarette users (Biener & Hargraves, 2015). However, dual users in our study also reported higher rates of other tobacco use in the past 30 days, emphasizing the importance of comprehensively categorizing tobacco product use when evaluating tobacco product transitions over time.

To explore potential mechanisms related to transitions in dual use groups, we investigated whether Wave 2 reasons to use e-cigarettes and e-cigarette device characteristics were associated with group membership in Wave 3. E-cigarette-only users in Wave 3 were more likely to report in Wave 2 that they used e-cigarettes to attempt to quit smoking, compared with cigarette-only users or dual users. In addition, nearly half (47%) of our participants who were exclusive users of e-cigarettes at Wave 1 were former, established smokers who had made their most recent, successful attempt at quitting smoking within the previous 12 months. This finding supports previous evidence suggesting that e-cigarette use may hold promise in assisting some highly dependent smokers hoping to quit (Hajek et al., 2019) and that many smokers use e-cigarettes to do so (Caraballo et al., 2017). However, our findings, supported by those found in previous work in this area (Buu et al., 2018), also show that frequency of use and dependence are important predictors of whether someone might shift exclusively to e-cigarettes.

We found significant variation in e-cigarette device characteristics across user groups. Of note, e-cigarette-only users in Wave 3 reported paying more for their device, using rechargeable systems, being able to refill e-liquid, and changing the voltage on their devices. This indicates that the more e-cigarette users invested in their devices, whether through finances or the ability to modify their device, the more likely they were to remain or become e-cigarette users. Interestingly, between Waves 2 and 3, dual users were more likely than exclusive e-cigarette users to change whether they owned a device or had the ability to recharge it, which could indicate that dual users were more likely to acquire and transition to a newer generation of e-cigarette products than were exclusive e-cigarette users, who likely had already adopted and continued to use such products. Our findings align with previous evidence that device characteristics differ across types of e-cigarette users and these users vary in reasons for e-cigarette use (Coleman et al., 2019; Soule et al., 2018). Limitations in the data make it difficult to draw a causal conclusion that dual users in our sample who successfully transitioned to exclusive e-cigarette use did so to quit smoking and that the number of days they used e-cigarettes contributed to that success. We were unable to observe among these participants, for example, variation in other important device characteristics that may have contributed to reduction in conventional cigarette use, or use of other cessation aids, number and duration of attempts to reduce cigarette use, motivation and preparedness to reduce smoking, and the influence of attitudes toward conventional and e-cigarettes in each participant’s larger social environment. However, our findings may inform intervention design for harm reduction, in conjunction with future work that more precisely quantifies what level of consumption (frequency, nicotine concentration, and device characteristics) are most closely associated with transitions to exclusive e-cigarette use. Further longitudinal research is also needed to examine how changes in device characteristics, particularly in the current rapidly changing e-cigarette device landscape, are associated with variation in dual use transitions over time.

We also found that transitions between dual and exclusive use differed across sociodemographic characteristics historically related to tobacco use disparities (National Cancer Institute, 2017), such that smokers who remained or reverted to cigarette-only use rather than remaining/becoming dual users were more likely to be older, male, African American, less educated, and earn less income, offering insights that population-level tobacco prevention and control efforts could target. These findings warrant further investigation to understand why subgroups of smokers who remain cigarette-only smokers may be less likely to engage with emerging tobacco products such as e-cigarettes. Although prolonged dual use may expose adults to toxicants in both conventional and e-cigarettes, dual use is one signal of engagement with a potentially less harmful product, with the goal of continued movement toward exclusive use of lower-harm products. Thus, future policy alternatives to reduce the burden of combusted product use could be targeted to reach populations of smokers who appear more reluctant to make transitions away from exclusive use of combustible products.

Several limitations should be considered when evaluating the internal and external validity of this study. All data were self-reported; thus, there may be bias in tobacco use behaviors and dependence scores. We were unable to include additional dual user groups that are of interest to policymakers, such as transitions from established dual use to cessation of tobacco use completely, since the sample size for these groups was too small to include in analytical models and draw meaningful inference. We also limited our models to only those who were “current, established” users of either or both products in Wave 1; however, a sensitivity analysis assessing whether the main parameters of interest would change if we opened up the analytic sample to also include all past-30-day users at Wave 1 revealed that doing so yielded no significant changes to our conclusions under this alternative approach (Appendix A). (Supplemental material appears as an online-only addendum to this article on the journal’s website.)

In addition, the PATH Study assessed two kinds of tobacco product dependence—all tobacco products that are not e-cigarettes and e-cigarette–specific dependence. Thus, conventional nicotine dependence would not be specific to cigarettes alone if a participant used a tobacco product other than conventional cigarettes or e-cigarettes. The PATH Study also did not include the full range of questions from the WISDM or the Nicotine Dependence Syndrome Scale (NDSS) but adapted items from these scales in designing the survey. Finally, our findings of associations between nicotine dependence and dual use 1 and 2 years later should not be interpreted as causal, and further research is needed on these relationships over longer periods. To our knowledge, this is the first known study to look at nicotine dependence and smoking behaviors between dual user groups across 3 years and to correlate e-cigarette device characteristics and reasons to use with these transitions.

Conclusion

Since e-cigarettes were introduced to the United States, they have held promise as an alternative that could lead to reduction or cessation of combustible cigarette smoking, but evidence on that potential is, at best, mixed. Furthermore, for many smokers, transitioning to e-cigarette use may not result in quitting smoking altogether but rather the use of both products for an extended period, or even an eventual transition to cigarette-only smoking. This study provides new evidence that dual use among established smokers and transitions to and from dual use are associated with higher nicotine dependence compared with remaining a cigarette-only smoker or e-cigarette-only user and that the regularity (i.e., number of days) with which an adult uses each of these products is associated with transitions to exclusive rather than dual product use.

Our findings suggest that, although e-cigarette-only use may be less harmful than cigarette-only use, many highly dependent smokers continued to use both products, exposing them to potential toxicants from both combusted and e-products, which further complicates crafting regulatory strategies and prevention efforts that aim to improve public health by reducing the burden of tobacco products.