Nicotine Dependence in Dual Users of Cigarettes and E-Cigarettes: Common and Distinct Elements

Abstract

Introduction

In a sample of dual users of cigarettes and electronic nicotine delivery systems (ENDS; e-cigarettes), we evaluated psychometric properties of ENDS versions of the Nicotine Dependence Syndrome Scale (NDSS), the brief Wisconsin Inventory of Smoking Dependence Motives (WISDM), and the Fagerström Test for Nicotine Dependence (FTND). Using the NDSS, we tested the hypothesis that there would be one common underlying factor of dependence across the cigarette and ENDS scales and other product-specific factors.

Aims and Methods

Adult dual users (N = 404) completed baseline cigarette and ENDS versions of the NDSS, WISDM, and FTND, and biweekly surveys of their smoking and vaping. Analyses included bifactor modeling, which helps to identify both a general and product-specific factor for each item, and exploratory factor analyses of the combined cigarette and ENDS NDSS items and examinations of concurrent and predictive validity.

Results

The bifactor model was not a good fit, suggesting the lack of one common underlying dependence factor. Factor analyses revealed separate, similar factors for both products, with only one factor (priority) showing overlap of cigarette and ENDS items. ENDS scales significantly predicted ENDS use over time, but not cigarette use. Cigarette scales did not predict ENDS use over time.

Conclusions

Although the cigarette and ENDS NDSS versions showed similar factor structure, there was not a primary common underlying factor reflecting drive or tolerance, but rather product-specific factors. The cigarette scales were not valid for predicting ENDS use. These results highlight the importance of separately assessing dependence for cigarettes and ENDS in dual users.

Implications

Although underlying dimensions of nicotine dependence may be similar for ENDS and cigarettes, separate, product-specific measures may be needed to understand differences in product-specific dependency and predict changes in use of each product over time.

Introduction

Dual use of combustible cigarettes and electronic nicotine delivery systems (ENDS) is increasingly common among adult cigarette smokers. Among the 19.3% of the US adult population who used tobacco products in 2017, 14.0% used conventional cigarettes, and 19.0% used at least two tobacco products.¹ The most common dual-use combination was cigarettes and ENDS, which constituted almost a third of poly-users (30.1%).¹ This proportion of dual users may have increased as newer ENDS devices entered the market and gained popularity over the past 2 years.

Much controversy exists over whether dual use may be a path away from combustible tobacco use or whether it may undermine cessation efforts.^2–4 Dual use may also exacerbate health symptoms and problems.⁵ Increasingly, concerns have been raised about the risk of ENDS for increasing or sustaining nicotine addiction, and there is growing interest in addressing the needs of individuals who may feel addicted to e-cigarettes and want to quit.^3–7 Knowing more about nicotine dependence related specifically to ENDS products may be important in assisting those who want to quit use of all nicotine products.

A key question to help address this issue is assessing nicotine dependence in dual users and differentiating the potential contribution of each product type (combustible or ENDS) to individual differences in levels of nicotine dependence. Previous findings suggest that dual users may have comparable or higher nicotine dependence levels than cigarette-only smokers.⁸ In addition, knowing whether the same dimensions of nicotine dependence are applicable to both product types or whether there are unique dimensions with each may help to inform intervention efforts. The current study evaluated whether there is one common nicotine dependence dimension underlying use of both products and whether there are product-specific dimensions.

To date, researchers most often have adapted well-established measures of nicotine dependence to use with ENDS by changing wording in the measures’ items to reflect vaping or e-cigarettes, rather than referring to cigarettes.⁹ One exception is the Penn State Electronic Cigarette Dependence Index (PS-ECDI), which was developed using constructs common to other measures of nicotine dependence.¹⁰ Piper et al. examined the internal structure and validity of the PS-ECDI along with two other e-cigarette dependence scales that were adapted from their cigarette versions (the e-cigarette Fagerström Test for Cigarette Dependence—e-FTND and the e-cigarette Wisconsin Inventory of Smoking Dependence Motives—e-WISDM) in a sample of adult dual users of both e-cigarettes and cigarettes.⁹ All three e-cigarette dependence measures showed good correlations with key outcome variables among dual users, such as continued e-cigarette use over time and perceptions of addiction. However, they did not specifically compare the e-cigarette versions of the scales with the traditional cigarette dependence scale versions.

Morean et al. developed a specific nicotine dependence scale for e-cigarettes based on the PROMIS (Patient Report Outcomes Measurement System) Nicotine Dependence Item Bank with options for a 22-, 8-, and 4-item version.¹¹ Each of the PROMIS-E versions had strong psychometric properties for both exclusive e-cigarette users and dual users of e-cigarettes and cigarettes.

Strong et al. used data from the PATH (Population Assessment of Tobacco and Health) study to develop and validate a measure of tobacco dependence across a range of tobacco products.⁸ Using item response models, they evaluated items from multiple measures of nicotine dependence and found that, across products, there was a primary single dimension of tobacco dependence that could be used as a common measure to assess tobacco dependence across different kinds of tobacco users. Their findings suggest that although there may be differences in levels of dependence across different tobacco products, a core set of items can validly assess tobacco dependence across the various products. One limitation of this study, however, was that respondents did not answer items separately about their cigarette and ENDS use.

The current study extends these prior investigations assessing nicotine dependence among dual users by investigating the internal consistency, structure, and predictive validity of adapted e-cigarette measures of nicotine dependence and comparing the adapted versions to their respective original cigarette versions. We used an adapted version of the Nicotine Dependence Syndrome Scale (NDSS),¹² as well as the e-FTND¹³ and a brief version of the e-WISDM¹⁴ in a diverse sample of adult dual users, with more extensive focus on the NDSS. The NDSS is based on a multidimensional conceptualization of dependence and assesses five factors of nicotine dependence: drive, tolerance, continuity, stereotypy, and priority.¹² Drive reflects a core feature of nicotine dependence—compulsion to smoke, craving, and withdrawal symptoms. Tolerance reflects increased resistance to the negative effects of smoking; continuity reflects smoking regularity; and stereotypy reflects homogeneity of smoking. Priority reflects behavioral preferences that prioritize smoking over competing considerations. The priority factor is predictive of smokers’ urge intensity when not smoking, reflecting the preeminence of smoking during periods of abstinence, and predictive of smoking cessation outcomes at multiple timepoints.^12,15 Unlike the PROMIS-E, which is a single factor scale, the NDSS allows for a multidimensional examination of the underlying constructs of nicotine dependence, which may play out differently with dual users of ENDS and cigarettes. Although the underlying dimensional structure of the Wisconsin Inventory of Smoking Dependence Motives (WISDM) has been previously investigated in dual users,⁷ less attention has been paid to the NDSS with dual users.

We specifically examined the question of whether there was one common dimension underlying parallel measures of nicotine dependence (for cigarettes and e-cigarettes) among the dual users considering both the e-cigarette versions of the scale and the traditional cigarette version for nicotine dependence. We hypothesized that a common, underlying factor representing a core feature of nicotine dependence would emerge, as well as product-specific factors. Identifying the common and unique factors of dependence for dual users may help in developing more tailored interventions to reduce use of both products.

Methods

Participants

Participants (N = 404) were recruited for a longitudinal observational study through a combination of social media (eg, Facebook), Craigslist ads, local electronic listservs, flyers, and word of mouth. Interested individuals completed an online screener and then a second level of phone screening. Eligibility criteria included: residing in the Chicago metropolitan area; aged 18 years or older; having smoked combustible cigarettes at least once a week in the past 30 days; having used an electronic cigarette (e-cigarette or ENDS) within the last 14 days but not on a daily basis; and indicated susceptibility to using an ENDS in the near future as defined by their response of “moderately” or “very likely” to the following questions: “How likely are you to use an e-cigarette in the next 2 weeks” and “How likely are you to purchase an e-cigarette in the next 2 weeks?” Individuals who did not speak or read English were excluded as well as those who responded inconsistently to screening questions, or those who were unwilling to participate in 7 days of ecological momentary assessments.

Procedures

All procedures were approved by the UIC Institutional Review Board. Eligible participants provided written informed consent at an in-person meeting and completed baseline questionnaire assessments of demographics, tobacco use history and patterns, motives, beliefs, and expectancies for e-cigarettes and cigarette use, and other measures. Participants also completed two 7-day waves of ecological momentary assessment using a study-provided device over the course of the study. After baseline, participants completed biweekly web-based assessments of all tobacco use. Data for the current study come from the baseline questionnaires and the biweekly web-based assessments of tobacco use through the 6-month follow-up. Baseline data were collected between October 2016 and September 2018.

Cigarette and E-Cigarette Dependence Measures

Participants completed three cigarette dependence measures: (1) the Fagerström Test for Nicotine Dependence (FTND)¹³; (2) a 14-item Brief WISDM¹⁶; and the NDSS.¹² Participants also completed parallel e-cigarette dependence measures of each of these. The e-cigarette versions of the scales replaced cigarette terms with terms applicable to ENDS products. For example, if an NDSS item was “If I’m low on money, I will spend it on buying cigarettes instead of buying lunch,” an adapted item for ENDS would be “If I’m low on money, I will spend it on buying e-cigarettes or e-juice instead of buying lunch.” The same modifications were made to the 14-item WISDM and FTND to reflect ENDS use.

Cigarette and E-Cigarette Use Behavior

At baseline, participants self-reported the number of days in the past 7 days and past 30 days that they used each type of tobacco product, along with the amount (number of cigarettes; number of “sessions” of ENDS) of each product used on days used. Following the baseline assessment, participants completed web-based surveys every 2 weeks for 1 year, including self-reported rate and frequency of cigarettes and ENDS use over the past 14 days, type of ENDS device and e-liquid used, and intention to continue using or purchasing ENDS over the following 14 days.

Data Analytic Plan

Descriptive Statistics and Differences Among Cigarette and E-Cigarette Versions of Nicotine Dependence

We calculated Cronbach’s alphas for each version of the three nicotine dependence measures, and we performed t tests to examine whether nicotine dependence scores for each of the three scales differed between the cigarette and e-cigarette versions.

Bifactor Model

A bifactor analysis¹⁷ was used to assess whether there was a common, underlying construct of nicotine dependence across products and to identify product-specific factors. All items from the cigarette NDSS and ENDS NDSS were entered together for the bifactor analysis using MPlus 8.0 for bifactor modeling. Bifactor modeling is useful for evaluating the feasibility of applying a common general latent model structure across heterogenous indicators.¹³ The NDSS was selected for bifactor analysis because participants completed all items for both products, and its multidimensional underlying construction provided a richer opportunity for examining factor structures compared with single factor scales such as the PROMIS-E or FTND. Full-information maximum-likelihood estimation, including robust standard errors, was employed to handle missing data. We specified that all items would be explained by a first common factor and then the product-specific items would load on separate factors. A Comparative Fit Index >.90, Root Mean Square Error of Approximation <.08, and Standardized Root Mean Square Residual <.08 would indicate acceptable model fit.^18,19

Exploratory Factor Analysis

All items from the cigarette NDSS and e-cigarette NDSS were also entered together into an exploratory factor analysis using maximum-likelihood estimation and a Varimax rotation. Factors with an eigenvalue ≥1 were extracted. NDSS items that had a factor loading ≥0.4 were retained in each factor. Factor scales were created by averaging the retained items for each factor, and Cronbach’s alpha values were calculated for the factor scales. Pearson’s correlation coefficients were calculated for the two extracted factors that each included only two NDSS items.

Concurrent, Convergent, and Predictive Validity

To examine construct validity across and within products, correlations were computed between all extracted factor scales from the exploratory factor analysis, other nicotine dependence scale scores, and other measures of nicotine dependence including usage rate and frequency for both cigarettes and ENDS. In order to examine predictive validity after the baseline wave, we computed correlations between nicotine dependence measures and participants’ daily rates of cigarette and ENDS use at baseline and at 6 months.

Results

Participants (N = 404; 58.4% male) averaged 34.6 years of age (SD = 12.5); 40.1% were White, 31.7% were Black, 12.1% Hispanic, and 11.6% self-identified as Asian. Most participants were employed part- or full-time (62.38%); 25.9% had no more than a high school education; 50.2% completed some college; and 23.9% had a college degree or more. Participants smoked an average of 8.4 (SD = 8.26) cigarettes per day on 25.2 (SD = 7.92) days in the past 30 days. Most participants (60.3%) primarily smoked menthol-flavored cigarettes. Less than half of participants (43.1%) were currently trying to reduce the number of cigarettes smoked per day. Participants used ENDS an average of 4.4 (SD = 7.05) sessions per day and 15.0 (SD = 10.96) days in the last 30 days. Most participants used a rechargeable device, with 51.7% using a device with a refillable tank or cartridge. Sweet flavored e-liquid was most popular, with 50.5% of participants primarily using sweet flavors including fruit, candy, dessert, and other similar flavors. Although 28% did not know the nicotine concentration of their e-liquid, only 2% of participants reported using 0 mg/mL nicotine e-liquid; 18.5% reported using ≤3 mg/mL; 23.2% reported using 4–6 mg/mL; 15.1% reported 7–12 mg/mL; and 13.6% reported using ≥13 mg/mL. At 6 months, data were available for 373 participants (92.3% of baseline sample).

Differences in Nicotine Dependence Scores Across Products

For each scale, participants had higher average scores for the cigarette dependence versions than the corresponding ENDS versions (all ps < .001). For the NDSS, the mean cigarette dependence score was 2.8 (SD = 0.70) compared with 2.4 (SD = 0.76) for the ENDS version (t = 8.50, p < .0001). For the WISDM, the mean cigarette dependence score was 28.5 (SD = 9.56) compared with 23.0 (SD = 10.24) for ENDS; and for the FTND, the mean cigarette scale score was 3.7 (SD = 1.88) compared with 2.9 (SD = 1.72) for ENDS. Both the WISDM and NDSS had excellent internal consistency for both product versions (coefficient alphas of ≥.85), whereas the coefficient alphas for the FTND were acceptable, but less robust (.65 and .70).

Concurrent and Predictive Validity: Correlations Between Nicotine Dependence Measures Across Products, Over Time, and With Usage Data

Table 1 presents the correlations between the nicotine dependence measures both within and across products, as well as the correlations between the nicotine dependence measures and usage of both cigarettes and ENDS. As can be seen from the table, the correlations were uniformly higher within products than across products. For example, the cigarette NDSS correlated .78 with the cigarette WISDM and .56 with the cigarette FTND, but correlations with the ENDS versions of the scales ranged from .33 to .14. Similarly, each cigarette dependence scale significantly correlated with cigarette rate and frequency (rs ranged from .38 to .53 with concurrent measures at baseline and from .37 to .46 prospectively over 6 months), but not so with ENDS rate and frequency. However, each of the ENDS dependence measures was significantly correlated with ENDS rate and frequency both concurrently at baseline (rs range from .33 to .47) and prospectively over 6 months (rs range from .25 to .31). Each of the dependence measures also significantly prospectively predicted usage rates, even after controlling for baseline use rates. For the cigarette dependence measures, partial correlations between the respective measure and 6 months cigarette smoking rates were .24 (p < .0001) for the FTND, .15 (p < .004) for the NDSS, and .19 (p < .002) for the WISDM. Correspondingly, the partial correlations between the ENDS dependence measures and 6 months ENDS usage rates were .20 (p < .0002) for the FTND-E, .12 (p < .02) for the NDSS-E, and .14 (p < .009) for the WISDM-E.

Table 1.

Means and Correlations of Nicotine Dependence Measures Across Products, at Baseline (N = 404) and at 6 Months (N = 373)

	Mean (SD)	Cigarette rate (cigs/ day)	Cigarette frequency (days/mo)	ENDS rate (sessions/ day)	ENDS frequency (days/mo)	NDSS (cigarette)	NDSS (ENDS)	WISDM (cigarette)	WISDM (ENDS)	FTND (cigarette)	FTND (ENDS)	Cigarette rate (cigs/day): 6 mo	ENDS rate (sessions/day): 6 mo
Cigarette rate (cigs/day)	8.4 (8.26)	1.00	0.47***	−0.03	−0.10*	0.38***	0.04	0.42***	−0.00	0.53***	0.03	0.55***	−0.03
Cigarette frequency (days/ mo)	25.2 (7.92)		1.00	−0.14**	−0.16***	0.42***	−0.10*	0.45***	−0.16***	0.45***	−0.13*	0.41***	−0.15**
ENDS rate (sessions/day)	4.4 (7.05)			1.00	0.58***	−0.03	0.33***	−0.00	0.42***	0.01	0.47***	−0.10*	0.31***
ENDS frequency (days/ mo)	15.0 (10.96)				1.00	−0.05	0.47***	−0.05	0.51***	−0.07	0.42***	−0.06	0.26***
NDSS (cigarette)	2.8 (0.70)					1.00	0.33***	0.78***	0.27***	0.56***	0.14**	0.32***	−0.04
NDSS (ENDS)	2.4 (0.76)						1.00	0.26***	0.80***	0.11*	0.55***	−0.02	0.22***
WISDM (cigarette)	28.5 (9.56)							1.00	0.31***	0.61***	0.12*	0.37***	−0.04
WISDM (ENDS)	23.0 (10.24)								1.00	0.08	0.57***	−0.13*	0.25***
FTND (cigarette)	3.7 (1.88)									1.00	0.25***	0.46***	−0.04
FTND (ENDS)	2.9 (1.72)										1.00	−0.09	0.31***
Cigarette rate (cigs/day): 6 mo	6.2 (6.59)											1.00	−0.04
ENDS rate (sessions/day): 6 mo	3.6 (6.95)												1.00

ENDS = electronic nicotine delivery systems; FTND = Fagerström Test for Nicotine Dependence; NDSS = Nicotine Dependence Syndrome Scale; WISDM = Wisconsin Inventory of Smoking Dependence Motives.

*p < .05, **p < .01, ***p < .001.

Bifactor Analysis

Although we had hypothesized that there would be one common, general strong nicotine dependence factor across both the cigarette and ENDS items, along with several product-specific factors, the bifactor model did not fit the data well: the Root Mean Square Error of Approximation was 0.092; Comparative Fit Index was 0.659; Tucker-Lewis Index was 0.617, all of which indicated that this was not a close fit and that the bifactor model should be rejected. In addition, the factor loadings on the first factor were low, apart from six items across the cigarette NDSS and ENDS NDSS which had item estimates between 0.60 and 0.81. These items were three similar items from the priority factor of the NDSS for cigarettes and ENDS. These items included “Even if traveling a long distance, I’d rather not travel by plane because I’m not allowed to smoke,” “Sometimes I decline offers to visit with my non-smoking family/friends because I know I’ll feel uncomfortable if I smoke,” and “I tend to avoid restaurants that don’t allow smoking, even if I would otherwise enjoy the food” and the corresponding ENDS items.¹² Beyond this first factor, all subfactors were product-specific and did not include items from both the cigarette and ENDS NDSS.

Exploratory Factor Analysis

Eight factors with eigenvalues greater than 1.0 explained 53.6% of the variance in the combined cigarette and ENDS items. The factors with the item factor loadings are shown in Supplementary materials table. The exploratory factor analysis replicated the results of the bifactor analysis in that only one factor emerged (factor 2, accounting for 9.4% of variance) which included items from both the cigarette NDSS and ENDS NDSS. This factor reflected environmental restrictions, including all items from the NDSS priority factor for both cigarettes and ENDS, the same items from the common factor in the bifactor analysis. All the other factors included single product items, with three of these reflecting ENDS items and four reflecting cigarette items. The first factor reflected ENDS items of drive and tolerance, accounting for 10.8% of the variance. Similarly, the strongest factor including only cigarette items also included drive items from the cigarette NDSS.

Correlation Between NDSS Factors Extracted

Scale scores for the factors were created by averaging items that loaded above .4 for each factor (factor scale score means shown in Supplementary materials table; correlations between factor scale scores shown in Table 2). In general, correlations between factors were stronger within products than across products. For example, the factor scale containing ENDS Drive and Tolerance items correlated .51 and .58 with the other ENDS factor scales, but only .13–.22 with the cigarette only factor scales. This pattern of results suggests that there is only modest overlap in indicators of dependence across products.

Table 2.

Correlations Between Factors Extracted From the Cigarette NDSS and ENDS NDSS

	1. ENDS	2. ENDS and cigarette	3. ENDS	4. Cigarette	5. Cigarette	6. Cigarette	7. ENDS	8. Cigarette
1. ENDS NDSS: drive and tolerance	1.00	0.34***	0.51***	0.22***	0.13*	0.20***	0.58***	0.15**
2. ENDS and cigarette NDSS: priority		1.00	0.09	0.34***	0.42***	0.20***	0.25***	0.21***
3. ENDS NDSS: continuity			1.00	0.10	−0.02	0.34***	0.29***	0.13*
4. Cigarette NDSS: drive				1.00	0.52***	0.20***	0.12*	0.46***
5. Cigarette NDSS: stereotypy					1.00	0.07	0.25***	0.34***
6. Cigarette NDSS: continuity						1.00	−0.01	0.27***
7. ENDS NDSS: stereotypy							1.00	0.05
8. Cigarette NDSS: tolerance								1.00

ENDS = electronic nicotine delivery systems; NDSS = Nicotine Dependence Syndrome Scale.

*p < .05, **p < .01, ***p < .001.

Concurrent and Predictive Validity of Factor Scores

As seen in Table 3, the factor scores with ENDS items correlated moderately and significantly with ENDS usage rates at both baseline and 6 months. For example, the ENDS factor of drive and tolerance correlated .32 with ENDS rate at baseline and prospectively .23 with ENDS rate at 6 months. In contrast, this same factor scale was negatively correlated with cigarette rate at baseline (r = −.24, p < .01) and unrelated to cigarette rate at 6 months (r = −.07). The factor score reflecting cigarette drive was correlated significantly with cigarette rate and frequency at both baseline and 6 months (rs between .28 and .46), but was unrelated to the ENDS rate at either baseline (r = −.01) or 6 months (r = −.03). Thus, the factor scale scores show good validity within product. Overall, the cigarette dependence factor scores had no significant correlation with ENDS rate over time. In contrast, the ENDS dependence factor scores were inversely related to cigarette use at baseline, with higher ENDS dependence associated with lower cigarette rate and frequency.

Table 3.

Correlations Between Factors Extracted, Other Measures of Nicotine Dependence at Baseline (N = 404), and Cigarette and ENDS Rate at 6 Months (N = 373)

	Cigarette rate (cigs/day)	Cigarette frequency (days/mo)	ENDS rate (sessions/day)	ENDS frequency (days/mo)	WISDM (cigarette)	WISDM (ENDS)	FTND (cigarette)	FTND (ENDS)	Cigarette rate (cigs/day): 6 mo	ENDS rate (sessions/day): 6 mo
ENDS NDSS: drive and tolerance	−0.01	−0.14**	0.32***	0.45***	0.22***	0.80***	0.09	0.55***	−0.07	0.23***
ENDS and cigarette NDSS: priority	0.22***	0.16**	−0.01	0.04	0.46***	0.31***	0.34***	0.19***	0.21***	−0.01
ENDS NDSS: continuity	−0.02	−0.13*	0.15**	0.26***	0.09	0.48***	−0.05	0.27***	−0.06	0.11*
Cigarette NDSS: drive	0.35***	0.46***	−0.01	−0.05	0.75***	0.24***	0.53***	0.10*	0.28***	−0.03
Cigarette NDSS: stereotypy	0.42***	0.41***	−0.03	−0.03	0.59***	0.13**	0.56***	0.11*	0.35***	−0.07
Cigarette NDSS: continuity	−0.01	0.02	−0.05	−0.08	0.25***	0.18***	0.05	0.04	0.00	−0.00
ENDS NDSS: stereotypy	0.05	−0.04	0.40***	0.49***	0.20***	0.62***	0.11*	0.47***	0.02	0.23***
Cigarette NDSS: tolerance	0.30***	0.31***	0.02	−0.03	0.49***	0.13**	0.32***	0.05	0.20***	−0.01

ENDS = electronic nicotine delivery systems; FTND = Fagerström Test for Nicotine Dependence; NDSS = Nicotine Dependence Syndrome Scale; WISDM = Wisconsin Inventory of Smoking Dependence Motives.

*p < .05, **p < .01, ***p < .001.

Discussion

The current study evaluated the factor structure and concurrent and predictive validity of an e-cigarette adapted version of the NDSS among a sample of adult dual users of both cigarettes and ENDS. We hypothesized that there would be a common strong factor, representing a core feature of nicotine dependence, across both the cigarette and e-cigarette versions of the NDSS. However, the bifactor analysis revealed that this model was not a good fit to the data. Similarly, the factor analysis also did not find a primary, general nicotine dependence factor that incorporated key items of drive or tolerance for both cigarettes and ENDS in a common factor. Both analyses found that the only items across both product scales that factored together were ones that reflected the NDSS construct of priority, or prioritizing the ability to use a product over being in situations where use might be restricted, such as in clean-air policy environments. Environmental restrictions on smoking may drive ENDS use among some smokers,^20,21 and examining comprehensive environment tobacco use restrictions may provide more insights into patterns of total tobacco use.

The more core constructs of dependence that would reflect internal drives or urge, withdrawal or loss of control, did not factor together across products. Our finding of no strong common dependence factor across products contrasts with that of Strong et al.,⁸ who found a final common tobacco dependence latent construct across products, which included items from the NDSS tapping drive and tolerance. However, the dual use group in those analyses answered dependence questions for the cigarette or more generic “tobacco product” reference and did not specifically answer items separately about their cigarette and ENDS use. Thus, asking specifically about each product may help to distinguish between different motives for use or product-specific symptoms.

Both the cigarette and e-cigarette versions of the dependence measures showed good internal reliability and concurrent construct validity. In addition, the underlying factor structure of the scales was similar for both the cigarette and ENDS versions. Each product-specific measure also showed good predictive validity for the corresponding product use variable over time. That is, as expected with these well-established cigarette scales, the three cigarette dependence measures were strong predictors of cigarette rate and frequency over 6 months; and the three e-cigarette dependence measures were good predictors of ENDS usage concurrently and over 6 months. However, the cross-product predictive validity of the measures was poor; the cigarette dependence measures showed no association with ENDS use at 6 months, while the ENDS dependence measures were not associated with cigarette use at 6 months. These findings suggest that a common “nicotine” or “tobacco” dependence factor does not interchangeably predict both cigarette and ENDS use, but instead, there may be different drivers of use patterns for these two separate products. The negative correlations between the ENDS dependence factors and cigarette smoking rates also suggest that there are different determinants of use of these products, and that the greater dependence on ENDS or motives for ENDS use may be one way to offset cigarette use.

Because of our inclusion criteria and overall goal of the parent study (focusing on the early stage use of ENDS and patterns of use over time), we recruited a sample of dual users who were primarily cigarette smokers, but who, at baseline, expressed an intention to continue ENDS use. A limitation of our study, therefore, is that we did not include daily ENDS users, which also may have deflated the levels of ENDS nicotine dependence in this sample. There were also no inclusion criteria related to motivation to quit cigarettes or to switch to ENDS; consequently, our participants had a range of reasons for using ENDS, some including desire to reduce smoking. Given these inclusion criteria, it was not unexpected that the overall cigarette dependence scores would be higher than those for the corresponding ENDS versions, and perhaps because of the different reasons for ENDS use, the predictive validity of the cigarette versions of the scales for ENDS use was poor. Others have also found, however, that among dual users, nicotine dependence may be lower for e-cigarettes than for cigarettes.²²

A limitation of our study was the relatively imprecise assessment of the amount and nicotine concentration of the ENDS products used. Assessing nicotine intake in a nonintrusive manner is a current challenge in the field. There was substantial variability among the participants in the types of ENDS devices used, nicotine concentration of the e-liquid or pod, as well as device features including battery voltage. ENDS product characteristics (including voltage, e-liquid nicotine concentration, battery, and device type) can affect inhalation toxicity, nicotine delivery, and sensory experience, so device specificity may have meaningful impact on ENDS nicotine dependence.^21,23–25

In sum, our results show that the ENDS-modified versions of the cigarette nicotine dependence measures maintain their strong psychometric properties, similar factor structure to the original versions, and have good concurrent and predictive validity with regard to ENDS product use. However, the measures of nicotine dependence for cigarettes did not show good validity for ENDS use or changes in ENDS use over time, and the ENDS versions were also unrelated to cigarette use. These findings suggest the importance of separately assessing nicotine dependence by product among dual users in order to better understand patterns of dual use and changes in product use over time.

Supplementary Material

A Contributorship Form detailing each author’s specific involvement with this content, as well as any supplementary data, are available online at https://academic.oup.com/ntr.

Funding

Research reported in this publication was supported by the National Cancer Institute and Food and Drug Administration Center for Tobacco Products grant R01CA184681. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Cancer Institute or the Food and Drug Administration.

Declaration of Interests

None declared.