Device Features and User Behaviors as Predictors of Dependence Among Never-Smoking Electronic Cigarette Users: PATH Wave 4

Abstract

Introduction:

Electronic cigarettes (ECIGs) vary greatly in their ability to deliver nicotine, which suggests they may also vary in their ability to produce dependence. This study examined individual and combined ECIG device features, and also user behaviors, as predictors of dependence in never-smoking ECIG users.

Methods:

Data were collected from 711 current ECIG users who had smoked <100 cigarettes in their lifetime at Wave 4 of the Population Assessment of Tobacco and Health (PATH) Study. Multivariable linear regressions examined individual (e.g., contains nicotine, uses a tank, flavor preference) and combined (e.g., refillable tanks, refillable mods) device features, and user behaviors (e.g., uses/day) as predictors of dependence, withdrawal, and craving after accounting for demographic variables.

Results:

Results for ECIG dependence and craving showed a similar pattern; higher levels were observed for older age, more frequent past 30-day use, using an ECIG containing nicotine (vs no nicotine), and using a non-refillable cartridge or refillable tank style (vs disposables). Higher withdrawal levels were observed for higher education levels and individual device features of tank (vs no tank), cartridge (vs no cartridge), refillable (vs non-refillable), and “other” flavor preference (vs tobacco flavor). Lower withdrawal levels were associated with a preference for sweet/fruit flavor(s) (vs tobacco flavor).

Conclusions:

Few use behaviors and device features, whether examined alone or in combination, predicted dependence-related outcomes. Findings underscore the challenge with regulating ECIG products in the current marketplace, which is inundated with a myriad of device types.

1. Introduction

Electronic cigarettes (ECIGs) are a class of products that may deliver nicotine to the user via inhalation of aerosolized liquid. A potential consequence of nicotine self-administration via products like ECIGs is the development of dependence, characterized by symptoms such as tolerance, withdrawal, and unsuccessful quit attempts (American Psychiatric Association, 2013). ECIG dependence likely promotes continued ECIG use, which may increase users’ risk for related health consequences (Kennedy et al., 2019) and/or progression to using other tobacco products (Primack et al., 2018). Some ECIG devices are capable of delivering nicotine in doses and at speeds comparable to a cigarette (Ramôa et al., 2016; Wagener et al., 2017), a product known to have a high potential for dependence. However, there also exist many ECIG devices that deliver nicotine in doses much lower than a cigarette (Eversole et al., 2020; Vansickel et al., 2010). This wide variation in the nicotine delivery profiles of ECIGs likely results in wide variation in their potential to produce dependence. Consequently, the U.S. Food & Drug Administration (FDA) has called for research that evaluates the influence of product features, including those for ECIGs, on dependence (FDA, 2019).

The current marketplace is inundated with hundreds of different ECIG products, many of which have features that can be customized: nicotine concentration (~0-70 mg/mL) and formulation (protonated or unprotonated), solvent ratio (0-100% propylene glycol), power output (up to ~200 Watts), and number of coils (Hsu et al., 2018; Massey et al., 2020). These features can have differential effects on nicotine yield and/or delivery, depending on whether they are examined individually or in combination. For instance, when other relevant ECIG features are held constant, relatively higher battery power results in increased nicotine yield and delivery (Eversole et al., 2020; Talih et al., 2015). On the other hand, low-wattage devices used with high nicotine concentrations (e.g., ‘second generation vape pens’) have the potential to deliver levels of nicotine that are comparable to high-wattage devices used with low nicotine concentrations (e.g., ‘third generation mods’) (Wagener et al., 2017). Such results might suggest that nicotine delivery, and thus dependence potential, be evaluated in the context of product feature combinations (i.e., ECIG device types).

To date, work addressing ECIG dependence has focused largely on individual features of ECIG products. That work shows higher levels of dependence for use of liquid that contains nicotine (vs no nicotine) (Morean et al., 2019a and 2019b) and that has higher (vs lower) nicotine concentrations (Foulds et al., 2015; Harvanko et al., 2018; Morean et al., 2019a), as well as for use of devices that have a higher (vs lower) power level (Harvanko et al., 2018) or that have a button and/or more than one battery (Foulds et al., 2015). These latter device features (i.e., power, having a button) might serve as better proxies for certain ECIG device types relative to liquid features, though they are not without their limitations. Indeed, buttons that require manual activation can be found on many different device types, making it clear that at least some single features cannot be used to distinguish between devices. Few studies have examined dependence as a function of device type, and thus considered combinations of product features. Two studies suggest that dependence is higher with use of what have been called ‘pod’ devices (e.g., JUUL) than other device types (e.g., vape pens, mods), in samples of youth and young adults (Boykan et al., 2019; Tackett et al., 2021). Though, other work reveals no differences in dependence levels among youth between device types (Vogel et al., 2018).

Of course, other factors play a role in the relation between ECIG product features and dependence. In one of the abovementioned studies that compared ECIG dependence among ‘pod’ and ‘non-pod’ users, those who used ‘pods’ reported more frequent use (i.e., daily use) than those who used ‘non-pod’ devices (Boykan et al., 2020). Consequently, frequency of use was confounded with device type, and previous work demonstrates that dependence levels are higher among those who vape more frequently (e.g., more vapes per day/month) (Morean et al., 2019a and 2019b; Piper et al., 2019; Yingst et al., 2021) and/or who have used ECIGs for a longer period of time (Foulds et al., 2015; Johnson et al., 2018). Also relevant is ECIG users’ history of other tobacco use. The majority of ECIG users are current or former cigarette smokers (Villarroel et al., 2020), though increases in ECIG use over the past few years have been most pronounced for young adults who have never smoked cigarettes (Bandi et al., 2020; Dai & Leventhal, 2019).

The purpose of the current secondary data analysis was to examine ECIG device features, as well as ECIG use behaviors, as predictors of ECIG dependence, craving, and withdrawal. Relative to previous work, our study will provide a more comprehensive assessment of the association between individual and combination device features and these dependence-related outcomes. It also will be the first to assess these associations in an adult sample of never-smoking ECIG users, thereby eliminating the potential confound of existing nicotine dependence.

2. Methods

2.1. Sample

Data derive from Wave 4 of the Population Assessment of Tobacco and Health (PATH) Study, a national-level longitudinal survey of youth and adults in the U.S. (Hyland et al., 2017). Data, collected in households using computer-assisted self-interviews, are available across four waves. The current study drew from Wave 4 (2016-2018) respondents (N=33,822), which included those re-contacted from Wave 1 (n=27,757) along with a “replenishment sample” of participants added in at Wave 4 to bolster power for cross-sectional analyses (n=6,065). The present analysis was limited to N=711 adults aged ≥18 years who were current ECIG users (i.e., n=562 ‘some days’ and n=149 ‘everyday’) and never cigarette smokers (n=49 reported 0 cigarettes and n=662 reported at least one but <100 cigarettes in lifetime) at Wave 4.

2.2. Measures

2.2.1. ECIG Device/Liquid Characteristics.

For individual characteristics, respondents reported on whether the ECIG device/liquid they use most often a) is rechargeable; b) has a tank; c) has a cartridge; d) is refillable; and/or e) contains nicotine (each coded 1=yes or 2=no). Individual characteristics were then combined using the PATH Study-supplied syntax (https://www.icpsr.umich.edu/web/NAHDAP/series/606) to create five ECIG device types: a) disposable (does not use cartridge or tank; not refillable; not rechargeable; 6.0%); b) non-refillable cartridge (uses cartridge; not refillable; rechargeable; 6.6%); c) refillable cartridge (uses cartridge; refillable; rechargeable; 3.2%); d) refillable tank system (uses tank; refillable; rechargeable; 77.2%); and e) refillable mod system (does not use cartridge or tank; refillable; rechargeable; 7.0%). These device groups were dummy coded for regressions with disposable as the comparison group.

Respondents also reported on the concentration of their most-used liquid by selecting one of eight response options: a) I don’t know (treated as missing for analyses; n=142; 20.0%); b) 0 mg or 0.0% (coded 0; n=275; 38.5%); c) 1-6 mg or 0.1-0.6% (coded 1; n=232; 32.6%); d) 7-12mg or 0.7-1.2% (coded 2; n=25; 3.5%); e) 13-17mg or 1.3-1.7% (coded 3; n=5; 0.7%); f) 18-24mg or 1.8-2.4% (coded 4; n=9; 1.3%); g) 25-39mg or 2.5-3.9% (coded 5; n=11; 1.5%); or h) 40+mg or 4.0+% (coded 6; n=12; 1.7%). Participants who reported use of an ECIG that does not contain nicotine (i.e. responded “no” for “contains nicotine”) were coded as 0mg.

Respondents reported on the flavor of ECIG liquid that they used regularly by selecting all that apply from the following list: a) tobacco; b) menthol or mint; c) clove or spice; d) fruit; e) chocolate; f) alcoholic drink; g) non-alcoholic drink; h) candy, desserts, or other sweets; i) some other flavor. Individual flavors were combined into five flavor categories: a) tobacco only (3.6%); b) menthol or mint only (9.2%); c) sweet/fruit only (i.e., use of one or more flavors such as watermelon, chocolate, candy; 69.5%); d) other only (clove, spice, (non)alcoholic drink, etc; 4.0%); and e) 2+ flavor categories (13.7%; flavor combinations other than multiple sweet/fruit flavors). These flavor categories were dummy coded with tobacco only flavor as the comparison group.

2.2.2. ECIG Use Behaviors.

Use questions included, “On how many of the past 30 days did you use an electronic nicotine product?”; “On average, on the days that you use, how many times each day do you pick up your electronic nicotine product to use it, whether you take one puff or several?”; and “Each time you pick up your electronic nicotine product to use it, about how many puffs do you take?”. Answers to each of these questions were open-ended, and were treated as continuous in analyses.

2.2.3. ECIG Dependence.

The PATH Study provides 16 dependence items validated previously (Strong et al., 2017). One of the items derives from the Diagnostic and Statistical Manual IV (American Psychiatric Association, 1994), “In the past 12 months, did you find it difficult to keep from using electronic nicotine products in places where it was not permitted?” (yes=5; no=1). This item was recoded to be consistent with other items in the measure (as in Shiffman & Sembower, 2020). Eleven items derived from the Wisconsin Inventory of Smoking Dependence Motives (WISDM-68; Piper et al., 2004), and were rated using a 5-point Likert scale (1=not at all true of me to 5=extremely true of me). Example WISDM-68 statements included “I frequently crave electronic nicotine products”; “My urges keep getting stronger if I don’t use electronic nicotine products”; and “Electronic nicotine products control me”. The remaining four items derived from the Nicotine Dependence Syndrome Scale (Shiffman et al., 2004), and also relied on a 5-point Likert scale (1=not at all true of me to 5=extremely true of me). Statements included “I find it really hard not to use electronic nicotine products”, and “I would find it hard to stop using electronic nicotine products for one week.” Scores for all items were averaged to create a mean score for each participant (as in Shiffman et al., 2020). Thus, possible scores ranged from 1 to 5, with higher scores indicating higher levels of ECIG dependence.

The PATH Study also assessed ECIG craving using a separate, single item, “I frequently crave electronic nicotine products.” Participants rated their level of agreement on a 5-point Likert scale ranging from 1 (not at all true of me) to 5 (extremely true of me), with larger scores indicating higher levels of ECIG craving.

2.2.4. ECIG Withdrawal.

Respondents who made an attempt to quit or cut down on their ECIG use in the past 12 months (n=115) also answered 7 questions about their related withdrawal symptoms: “Within days after stopping or cutting down on your electronic nicotine product use in the past 12 months, did you …” a) feel depressed (n=26; 3.7%); b) have difficulty falling asleep or staying asleep (n=22; 3.1%); c) have difficulty concentrating (n=27; 3.8%); d) eat more than usual or gain weight (n=25; 3.5%); e) become easily irritated, angry, or frustrated (n=38; 5.3%); f) feel anxious or nervous (n=43; 6.0%); and/or g) feel more restless than usual (n=38; 5.3%) (all coded 1=yes, 0=no). Items were summed to create an ECIG withdrawal score for each participant (range=0-7), with larger scores indicating higher levels of withdrawal.

2.2.5. Covariates.

All models were adjusted for potential confounds of age, sex (male=0; female=1), and education (less than high school=1; GED/high school graduate=2; some college or associate’s degree=3, bachelor’s degree or advanced degree=4). Also included were ethnicity (not Hispanic/Latino=0; Hispanic/Latino=1) and race (white=0; racial minority=1). The category of racial minority was based on the PATH Study-derived options for race of Black or African American or Other (i.e., American Indian or Alaska Native, Asian Indian, Chinese, Filipino, Japanese, Korean Vietnamese, Other Asian, Native Hawaiian, Guatemalan or Chamorra, Samoan, or Other Pacific Islander). The race and ethnicity variables were treated as separate in all analyses despite potential overlap: 50.6% non-Hispanic white, 22.2% non-Hispanic racial minority, 19.3% Hispanic white, 7.9% Hispanic racial minority. For age, respondents selected the appropriate range from the following options: 18-24 years (code=1, 77.5% of sample); 25-34 years (code=2, 15.3%); 35-44 years (code=3, 3.7%); 45-54 years (code=4, 2.1%); 55-64 years (code=5, 1.1%); and 65-74 years (code=6, 0.3%).

2.3. Data Analysis

Analyses were conducted using Mplus 8.4 on weighted data (i.e., PATH Study-provided Wave 4 cross-sectional non-replicate weights), which adjust for sampling, missing responses, and attrition across waves. All missing data were estimated using full information maximum likelihood. To ensure that primary findings were not due to suppression, Pearson correlations were used first to examine bivariate associations among study variables. A one-way analysis of variance (ANOVA) examined mean differences in dependence, craving, and withdrawal as a function of device group. Tukey’s Honestly Significant Difference post-hoc tests were used to follow-up on significant F tests. Independent samples t-tests examined mean differences in dependence, craving, and withdrawal as a function of individual device characteristics. Then, separate multivariable linear regressions were examined to determine whether ECIG device/liquid characteristics and/or ECIG use behavior are associated with ECIG dependence, withdrawal, and/or craving after accounting for demographic variables (age, sex, race, ethnicity, education). Models 1-3 included individual ECIG device/liquid characteristics and demographics, with ECIG dependence (Model 1), craving (Model 2), and withdrawal (Model 3) as the outcome variables. Models 4-6 included device groups, contains nicotine, and demographics, with ECIG dependence (Model 4), craving (Model 5), and withdrawal (Model 6) as the outcome variables. Finally, models 7-9 included ECIG use behavior and demographics, with ECIG dependence (Model 7), craving (Model 8), and withdrawal (Model 9) as the outcome variables. Variables with significant bivariate associations were allowed to covary. Associations were considered statistically significant when p<.05. Additionally, all analyses described above were performed on a reduced sample (n=602) that excluded respondents who reported using ECIGs on 0 days in the past month. The pattern was overwhelmingly similar, and thus results for the full sample are reported below.

3. Results

3.1. Bivariate Results

Table 1 shows descriptive statistics and bivariate Pearson correlations for all measures. Table 2 presents descriptive statistics and univariate comparisons for outcomes as a function of individual and combined device characteristics. With regard to the former, the one-way ANOVA demonstrated mean differences on dependence (F[4,384]=5.28, p<.001) based on device group. Shown in Table 2, post-hoc tests revealed that higher dependence scores were observed with use of a non-refillable cartridge than with use of a disposable, refillable tank, or refillable mod (all ps < .05; Tukey’s HSD). Mean craving and withdrawal scores did not vary by device group. With regard to flavor, one-way ANOVAs demonstrated mean differences for dependence (F[4,394]=5.43, p<.001) and craving (F[4,402]=3.42, p=.009). Post-hoc tests revealed significant differences for dependence but not craving; dependence scores were significantly higher for tobacco only relative to sweet/fruit only and other only flavors.

Table 1.

Bivariate correlations among study variables

Variables	M (SD) or n (%)	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
Sociodemographics
1. aAge, M (SD)	NA	--
2. Female, n (%)	270 (38.0)	0.07	--
3. Racial Minority, n (%)	214 (30.1)	−0.03	0.07	--
4. Hispanic, n (%)	193 (27.1)	0.02	0.04	−0.01	--
5. bEducation, M(SD)	NA	0.02	0.02	0.08	−0.05	--
Individual ECIG Characteristics
6. Rechargeable, n(%)	644 (90.6)	−0.04	−0.07	−0.09	−0.04	−0.02	--
7. Tank, n(%)	538 (75.7)	−0.1	−0.1	−0.04	−0.01	−0.03	0.38	--
8. Cartridge, n(%)	266 (37.4)	0.03	0.02	0.05	0.06	0.01	0.16	−0.02	--
9. Refill, n(%)	588 (82.7)	−0.11	−0.05	−0.07	−0.02	0.01	0.5	0.6	−0.12	--
10. Nicotine, n(%)	446 (62.7)	0.05	−0.03	−0.06	−0.07	0.05	0.08	−0.03	0.03	−0.01	--
11. cConcentration, M(SD)	0.81 (1.23)	−0.03	−0.1	−0.12	−0.09	0.02	0.11	−0.17	0.13	−0.09	0.61	--
User Behaviors
12. Past 30-day ECIG use, M(SD)	10.78 (11.61)	−0.01	−0.09	−0.09	−0.11	0.01	0.16	0.06	−0.1	0.06	0.32	0.33	--
13. Time per day ECIG use, M(SD)	11.62 (23.23)	−0.02	−0.05	−0.04	−0.05	0.01	0.07	0.01	−0.01	−0.03	0.21	0.22	0.46	--
14. ECIG puffs, M(SD)	6.94 (16.70)	−0.01	−0.06	0.02	0.05	0.02	−0.03	−0.07	−0.02	−0.07	0.05	0.04	0.13	0.15	--
Outcomes
15. Dependence, M(SD)	1.60 (0.80)	0.22	−0.07	−0.02	−0.01	−0.04	0.06	−0.07	0.12	−0.09	0.27	0.23	0.41	0.22	0.15	--
16. Craving, M(SD)	1.79 (1.12)	0.12	−0.1	−0.08	−0.03	0.02	0.06	−0.03	0.05	−0.02	0.27	0.26	0.46	0.24	0.09	0.87	--
17. Withdrawal, M(SD)	1.90 (2.41)	0.02	0.18	−0.01	0.07	0.01	0.03	0.16	0.12	0.11	0.14	0.07	0.08	−0.02	−0.04	0.56	0.46	--

Note. Bold values denote statistical significance, p<.05.

^aAge was coded categorically as 1=18-24 years old (77.5%), 2=25-34 years old (15.3%), 3=35-44 years old (3.7%), 4=45-54 years old (2.1%), 5=55-64 years old (1.1%), 6=65-74 years old (0.3%).

^bEducation was assessed as 1=less than high school, 2=GED/High school graduate, 3=some college or associate's degree, 4=bachelor's degree or advanced degree.

^cConcentration was coded as 0=0 mg or 0.0%, 1=1-6 mg or 0.1-0.6%, 2=7-12mg or 0.7-1.2%, 3=13-17mg or 1.3-1.7%, 4=18-24mg or 1.8-2.4%, 5=25-39mg or 2.5-3.9%, 6=40+mg or 4.0+%

Table 2.

Dependence, craving, and withdrawal scores as a function of individual and combined features

	Dependence, M (SD)	Craving, M (SD)	Withdrawal, M (SD)
Combination Device Characteristics
Disposable	1.12 (0.16)	1.27 (0.59)	2.33 (4.04)
Non-refillable Cartridge	2.14 (1.16)	2.23 (1.26)	1.09 (2.17)
Refillable Cartridge	1.77 (0.83)	2.07 (1.21)	2.00 (2.10)
Refillable Tank System	1.56 (0.76)	1.76 (1.12)	2.16 (2.52)
Refillable Mod System	1.56 (0.72)	1.69 (1.01)	0.75 (1.04)
p-value	<.001	.058	.413
Flavors
Tobacco only	2.43 (1.19)	2.42 (1.24)	3.00 (2.74)
Menthol only	1.75 (.74)	2.06 (1.19)	2.17 (2.37)
Sweets/fruit only	1.53 (.74)	1.70 (1.08)	1.67 (2.36)
Other only	1.22 (.47)	1.25 (.62)	3.75 (2.87)
2+ flavors	1.79 (.98)	2.03 (1.25)	2.00 (2.47)
p-value	<.001	0.009	0.382
Individual Device Characteristics
Rechargeable
Yes	1.62 (0.80)	1.81 (1.13)	1.93 (2.40)
No	1.38 (0.81)	1.52 (1.04)	1.60 (3.05)
p-value	.199	.238	.764
Tank
Yes	1.57 (0.77)	1.77 (1.12)	2.14 (2.50)
No	1.71 (0.92)	1.86 (1.14)	1.28 (2.07)
p-value	.154	.533	.096
Cartridge
Yes	1.73 (0.88)	1.86 (1.15)	2.23 (2.51)
No	1.53 (0.75)	1.75 (1.11)	1.66 (2.33)
p-value	.016	.326	.218
Refill
Yes	1.57 (0.76)	1.77 (1.12)	2.02 (2.41)
No	1.78 (1.00)	1.85 (1.13)	1.28 (2.37)
p-value	.073	.630	.232
Nicotine
Yes	1.73 (0.86)	1.97 (1.19)	2.13 (2.38)
No	1.24 (0.48)	1.28 (0.71)	1.38 (2.44)
p-value	<.011	<.001	.133

Note. p-values reflect significance values from one-way ANOVAs (device group) or independent samples t-tests (individual device characteristics).

For individual device characteristics, independent samples t-tests demonstrated mean differences on dependence based on whether participants’ ECIG contains nicotine and has a cartridge. Specifically, higher dependence scores were associated positively with use of an ECIG than contains nicotine (vs does not) and that contains a cartridge (vs does not) (p<.05). Additionally, participants who used an ECIG that contains nicotine reported higher levels of craving, relative to those who did not (p<.001).

3.2. Multivariable Regression Results

Multivariable regressions indicated that higher dependence was associated with older age, using an ECIG with nicotine, using an ECIG with a non-refillable cartridge or refillable tank system (compared to a disposable ECIG), and more ECIG use in the past 30 days (see Table 3). Predictors of higher craving included older age, white race, using an ECIG with nicotine, using an ECIG with a non-refillable cartridge or a refillable tank system (compared to a disposable ECIG), and more ECIG use in the past 30 days. Finally, higher withdrawal was predicted by higher education level, using an ECIG with a tank or a cartridge, using a refillable ECIG, and using other only flavor (relative to tobacco). Lower withdrawal was predicted by using sweet/fruit only flavor (relative to tobacco).

Table 3.

Multivariable regressions predicting ECIG dependence, craving, and withdrawal.

	Dependence			Craving			Withdrawal
Variables	B	SE	p	B	SE	p	B	SE	p
Sociodemographics
Female	−0.14	0.11	0.198	−0.19	0.14	0.17	0.01	0.64	0.993
Racial Minoritv	−0.17	0.12	0.148	−0.30	0.15	0.036	0.08	0.70	0.914
Hispanic	0.02	0.15	0.867	−0.06	0.16	0.701	0.78	0.74	0.296
Age	0.30	0.09	0.001	0.25	0.09	0.004	−0.09	0.24	0.702
Education	0.03	0.06	0.658	0.05	0.07	0.487	0.57	0.28	0.04
Adjusted R 2		0.13			0.07			0.04
Individual ECIG Characteristics 1
Rechargeable	−0.25	0.60	0.685	−0.29	0.71	0.684	−1.99	1.19	0.094
Tank	0.21	0.13	0.122	0.24	0.18	0.192	1.37	0.51	0.007
Cartridge	0.16	0.14	0.229	0.11	0.18	0.549	1.63	0.58	0.005
Refill	−0.09	0.25	0.73	−0.08	0.31	0.791	2.50	0.70	<.001
Nicotine	0.51	0.13	<.001	0.69	0.17	<.001	0.52	0.62	0.401
Concentration	0.06	0.04	0.191	0.08	0.07	0.218	0.26	0.15	0.089
Menthol onlv	−0.10	0.37	0.778	0.13	0.47	0.778	−0.73	0.82	0.368
Sweets/fruit only	−0.07	0.37	0.86	−0.05	0.45	0.918	−1.87	0.77	0.016
Other only	−0.51	0.38	0.171	−0.61	0.47	0.194	2.26	0.71	0.001
2+ flavors	−0.07	0.39	0.867	−0.07	0.48	0.892	−1.56	0.82	0.057
Adjusted R 2		0.26			0.18			0.46
Combined ECIG Characteristics 1
Nicotine	0.51	0.10	<.001	0.67	0.12	<.001	0.81	0.70	0.247
Non-refillable cartridge	0.90	0.25	<.001	0.91	0.31	0.003	−1.39	1.87	0.459
Refillable cartridge	0.31	0.24	0.189	0.44	0.35	0.219	−0.94	1.83	0.605
Refillable tank system	0.47	0.15	0.001	0.51	0.22	0.019	−0.01	1.69	0.997
Refillable mod system	0.27	0.19	0.156	0.26	0.28	0.357	−1.99	1.79	0.267
Adjusted R 2		0.24			0.16			0.12
User Behaviors
Nicotine	0.28	0.11	0.012	0.30	0.14	0.032	1.00	0.68	0.14
Past 30-day ECIG use	0.02	0.01	<.001	0.04	0.01	<.001	0.03	0.03	0.449
Times per day ECIG use	0.00	0.00	0.24	0.00	0.00	0.142	−0.01	0.01	0.439
ECIG puffs	0.01	0.00	0.119	0.00	0.00	0.858	−0.01	0.01	0.135
Adjusted R 2		0.27			0.24			0.10

Note.

¹Tobacco is the reference group for flavor.

²Disposable ECIG is the reference group.

Significant items are bolded.

4. Discussion

The purpose of this secondary data analysis was to examine the associations among ECIG device features and user behavior with dependence, craving, and withdrawal in a sample of never-smoking ECIG users. For individual features, only the presence of nicotine (vs no nicotine) was associated with greater dependence and craving (Morean et al., 2019a, 2019b). By contrast, multiple individual device features were associated with withdrawal; greater withdrawal was observed for use of a tank or cartridge (vs no tank or cartridge use, respectively), a refillable device (vs not refillable), and other only flavors (vs tobacco), while lower withdrawal was observed for use of sweet/fruit only (vs tobacco). When features were combined to create device types, users of non-refillable cartridge and refillable tank systems (vs disposables) had higher dependence and craving. As for user behavior, more days of ECIG use in the past month was associated with higher dependence and craving scores, as has been reported previously for former and current smokers who use ECIGs (Johnson et al., 2018; Morean et al., 2019a, 2019b; Piper et al., 2019; Yingst et al., 2021). Together, findings not only provide a comprehensive examination of the various factors that may influence ECIG dependence-related domains, but also eliminate the potential confound of previous or current cigarette use.

Overall, the observed pattern of results is not surprising. On the one hand, certain individual and combination ECIG features may be better able to deliver nicotine to the user than other features (Rüther et al., 2018; Wagener et al. 2017). Consequently, it seems reasonable to expect that these same individual or combination features would be associated significantly with ECIG dependence or craving. For instance, participants in the current study may have used specific non-refillable cartridge (e.g. JUUL) or refillable tank (e.g. Kanger pens) systems that deliver higher levels of nicotine than those who used the disposable systems (e.g. blu). On the other hand, this association is complicated by the abundance of device/liquid combinations that interact in different ways to influence nicotine delivery (Wagener et al., 2017). As an example, early ECIG models seemed more easily differentiated such that ‘first generation’ devices (cigalike styles) generally delivered less nicotine than ‘second generation’ devices (vape pen styles), which generally delivered less nicotine than ‘third generation’ devices (mod styles) (Rüther et al., 2018; Wagener et al. 2017). With each successive generation, increased power output presumably led to increased nicotine delivery. Today, however, there exists a ‘fourth generation’ of devices (pod styles), some of which have relatively low power levels but contain liquid nicotine that is salt-based and available in very high concentrations. Such devices may or may not be able to deliver more nicotine than those from earlier generations (Maloney et al., 2020; Yingst et al., 2019). Moreover, recent work demonstrates that even devices with relatively high-power capabilities (i.e., ‘third generation’ mods) can deliver very little nicotine to the user (Eversole et al., 2020). Perhaps devices categorized as refillable mods in the current study included such brands, thus explaining our finding that none of the dependence-related outcomes were predicted by this device type. These observations may explain the overall lack of differences in dependence-related domains across device features.

There also was no association between outcomes and two of the ECIG use behaviors (i.e., ECIG uses/day and ECIG puffs/use), likely due to challenges with measuring these more detailed patterns of use (see Blank et al., 2016 for review). While various recommendations have been made for survey items to assess ECIG use behavior (Pearson et al., 2018; Weaver et al., 2018), users themselves remain unsure of how best to describe their patterns (Cooper et al., 2016). ECIG consumption can by quantified in various ways (e.g., number of puffs, number of cartridges, mL of liquid), and the optimal measure may depend on the features of the device that is used. For example, users of cig-alikes may prefer quantifying their use in terms of the number of cartridges, while users of refillable devices (e.g. those with tanks) may prefer using mL of liquid (Cassidy et al., 2017).

Study results also are likely influenced by several limitations, including the measures used to assess device/liquid features. First, the PATH Study survey did not measure ECIG features related to battery capabilities, such as whether the power level was fixed or variable. Interestingly, however, ECIG users may be largely unaware of the power level of their device (Harvanko et al., 2018; Rudy et al., 2017). Second, the terms used to differentiate between certain features may not have been inclusive and/or consistent with those used by consumers. For example, participants were not asked whether their device used a ‘pod,’ another term that can describe the container that holds the liquid. While ‘pod’ style devices like JUUL are currently among the more popular device types (Cwalina et al., 2020), they did not begin inundating the ECIG market until 2017 (Huang et al., 2019) and data collection for the PATH Study Wave 4 was launched in 2016. Similarly, the term ‘disposables’ as used in the PATH Study survey likely reflects older, ‘first generation’ styles, as the more modern disposables (e.g., PuffBar, Hyde) appeared on the market even later than pod-style devices. Also, the terms that were included to describe features like the liquid container (‘cartridge’, ‘tank’) may not be differentiated in the same manner between users, or between users and the research community (Alexander et al., 2016; Coleman et al., 2018; Pearson et al., 2020). Adding to this confusion may be the terms used by the industry, such as that for blu® PLUS+ (https://www.blu.com/en/US/flavors/blu-plus-tanks). This product, which was available on the market at the time of data collection, consists of a liquid container that is marketed as a ‘tank’ but is clearly similar in appearance to what has been traditionally referred to as a ‘cartridge’ both by manufacturers and researchers (CDC, 2020).

To circumvent some of these limitations with ECIG terminology, researchers have included pictures of the devices to which they are referring (Morean et al., 2020) and/or asked respondents to report on their brand of ECIG used (Nardone et al., 2019). The PATH Study survey did include an item that asked respondents to choose the picture of a device type that best represented the one that they use: a) ‘first generation’ cigalikes (6.1%), b) ‘second generation’ vape pens (22.7%), or c) ‘third generation’ mods (71.2%). Still, none of the pictures resembled the more modern device types, such as those that look like a USB flash drive (e.g., JUUL) or a tear drop (e.g., Suorin). The PATH Study item that assessed device brand was asked only of those ECIG users who own a device (n=222). Also, because this question was open-ended, many respondents failed to provide details that would allow for differentiating between different device types of the same brand. For those who stated that they use ‘blu’ ECIGs, for instance, it is unknown whether their device was a blu® Device (pod-style), a blu® Disposable (cigalike that is completely disposable), or a blu® PLUS+ (cigalike that is rechargeable). For these reasons, a better approach might be to ask users to provide pictures of their own ECIG so that device features can be confirmed.

In conclusion, device features, and consequently use behaviors, are limited in their ability to predict ECIG dependence. Available ECIG products are complex in their design, with individual features combining in various ways to interact with behavior and determine nicotine delivery. Also because of this complexity, different ECIG products necessitate different patterns of use to achieve comparable levels of nicotine. Moreover, the terminology used to describe individual and combination features can be inconsistent among users, as well as between users and the research community (Ozga et al, 2021). Current surveillance systems might improve upon these limitations by asking respondents to provide pictures of the devices/liquids that they use. These same surveys might assess other features of ECIG use that are relevant to dependence but that were not measured here (e.g., vape clouds; Bold et al., 2018; Morean et al., 2019a), and that could ultimately be used to ensure the development of reliable and valid measures of ECIG dependence. Findings underscore the challenge with regulating ECIG products in the current marketplace, which is inundated with a myriad of device types.

Highlights.

• Few individual device features and use behaviors were associated with dependence

• Differences in dependence scores were observed as a function of device type

• Measurement of device features is complicated by the current ECIG marketplace