Electronic Cigarette Cue Reactivity in Exclusive Electronic Cigarette Users

Abstract

Background:

Exposure to electronic cigarette (ECIG) cues increases craving for ECIGs among people who smoke cigarettes, including those without prior ECIG use experience. This study aimed to examine the effects of ECIG cue exposure in individuals who use ECIGs regularly but have minimal smoking history (<100 cigarettes lifetime).

Methods:

Participants (N=42) attended two within-subjects conditions that differed by pictorial cue type: ECIG or neutral (water) cues. Each condition included two presentations of the condition-assigned cue type. Before and after each cue presentation, participants completed subjective measures of ECIG craving, anticipated relief from withdrawal, and mood. After the second cue presentation, participants completed ECIG delay discounting and behavioral economic tasks, with ECIG use quantified in puffs or in minutes of ECIG access. Linear mixed-effects regressions and paired-samples t-tests assessed the effects of cue type on these outcomes.

Results:

Ratings of ECIG craving and anticipation of relief from ECIG withdrawal increased significantly from pre- to post-cue exposure for ECIG cues relative to neutral cues (p’s<.05). A similar pattern was observed for negative mood (p<.05). ECIG cues (versus neutral cues) also increased discounting of delayed minutes of ECIG access (p<.05) but not ECIG puffs.

Conclusions:

This is the first study to demonstrate that ECIG cues increase ECIG craving and negative mood in individuals who vape but do not smoke cigarettes. Results have implications for restrictions on where vaping should be permitted.

1. INTRODUCTION

Over 30% of adults who use ECIGs have no prior smoking history, a percentage that doubles to 61.4% for young adults aged 18-24 (Kramarow and Elgadda, 2023). ECIG use can lead to nicotine dependence (Liu et al., 2017), characterized by symptoms such as loss of control, craving, and difficulty concentrating (American Psychiatric Association, 2013). It is therefore unsurprising that up to 76% of people who use ECIGs report an interest in quitting (Alalwan et al., 2022) and up to 75% have made a quit attempt (Simpson et al., 2021). Despite efforts to quit, many have difficulty remaining abstinent (Unger et al., 2020).

Failed quit attempts may be due in part to exposure to cues associated with ECIGs, such as device features or exhaled aerosol (i.e., vape cloud). Cue exposure may elicit craving for ECIGs, changes in mood, and nicotine-seeking behavior (Keijsers et al., 2022). To date, most work investigating the effects of ECIG cue exposure has done so in people who smoke cigarettes, some of whom have no ECIG experience. These studies demonstrate increases in self-reported urge and desire for both cigarettes and ECIGs, as well as increases in cigarette smoking behavior, after exposure to ECIG (vs. neutral) cues across device types (Blackwell et al., 2020; King et al., 2021; Vena et al., 2020a, 2020b). In a systematic review, Keijsers et al. (2022) found no differences in cue reactivity between people who use both cigarettes and ECIGs, only cigarettes, or only ECIGs. However, participants in this latter group either had a history of cigarette smoking or little experience with ECIGs. Importantly, even though over half of young adults who use ECIGs have never smoked cigarettes (Kramarow and Elgadda, 2023), no known studies have investigated the effects of ECIG cue exposure in this population.

Extant work examining ECIG cue exposure has relied primarily on subjective (e.g., self-reported craving) and behavioral (e.g., latency to vape) outcomes (Keijsers et al., 2022). Behavioral economic outcomes also may be sensitive to ECIG cue exposure. ECIG purchase tasks, for example, assess the reinforcing efficacy of ECIGs by asking participants the quantity of ECIG product they would purchase at increasing prices (Cassidy et al., 2020). Such tasks have demonstrated the effects of smoking-related cue exposure on demand for cigarettes among people who smoke; after exposure to smoking (vs. neutral) cues, people who smoke spend more money on cigarettes and are less responsive to increasing prices (Acker and MacKillop, 2013). Another related construct is impulsivity, which can be defined as steeper discounting of delayed rewards (“delay discounting”) (Madden & Johnson, 2010). In people who smoke, steeper discounting is associated with higher levels of nicotine dependence (Amlung and MacKillop, 2014) and greater nicotine consumption (Ohmura et al., 2005). Delay discounting is sensitive to some experimental manipulations, including episodic future thinking (i.e., imagining future events) (Stein et al., 2018) and drug administration (de Wit and Mitchell, 2010). No studies have assessed the influence of ECIG cue exposure on these or other behavioral economic measures.

The current study provides a comprehensive assessment of cue reactivity in individuals who use ECIGs but have never regularly smoked cigarettes, thus eliminating the confounds of existing nicotine dependence and previous experience with similar nicotine-related cues. We hypothesized that ECIG cue exposure would increase ratings of ECIG craving, anticipated relief from withdrawal, and negative mood, as well as demand for and discounting of ECIGs. We also aimed to explore the effects of ECIG cue exposure on heart rate; previous research with people who smoke cigarettes shows mixed findings for cigarette cue effects on heart rate (Drobes and Tiffany, 1997; Tong et al., 2007). Physiological responses may provide insight into autonomic activation related to motivational states such as craving or arousal.

2. METHODS

2.1. Participants

Eligible participants, recruited via flyers and online advertisements, were ≥18 years old and reported using a nicotine-containing ECIG >4 days per week for ≥3 months; these latter criteria increased the likelihood of including those most likely to be dependent on ECIGs. They also reported smoking ≤100 cigarettes/lifetime, and their screening carbon monoxide (CO) levels were ≤5 ppm (M=2.07, SD=0.85 ppm). Exclusion criteria were uncontrolled medical or psychiatric condition(s), past month use of alcohol or marijuana >15 days, past month use of any illicit substances (excluding marijuana), current pregnancy (verified via urinalysis) or breastfeeding, or a desire to quit vaping in the next month.

2.2. Procedures

This within-subjects study, conducted between January 2022 to December 2023, consisted of two, ~2-hour counterbalanced sessions (ECIG or neutral cues) separated by >48 hours. Sessions were scheduled at approximately the same time of day, and held in a controlled laboratory environment (e.g., small tables with computers and vitals monitors; comfortable chairs; no windows). Any items that could serve as possible cues were removed from the environment, including participants’ cellphones. During rest periods, participants were permitted to read or use activity books (e.g., crosswords) provided to them.

In the first session only (see Figure 1), participants completed baseline questionnaires assessing demographics, and ECIG dependence and device characteristics. They also completed a subjective measure of impulsiveness (Barratt Impulsiveness Scale) and a monetary delay discounting task (Monetary Choice Questionnaire). For both sessions, participants were required to abstain from nicotine/tobacco use for ≥8 hours. A bogus pipeline procedure was used (Hiler et al., 2020) such that participants were told that a saliva sample would be tested (but not actually tested) to verify presession abstinence, and that failure to abstain would result in rescheduling their session. Overnight abstinence from combustible products was verified via expired air CO ≤5 ppm. The session commenced with the continuous recording of heart rate followed by a 30- minute rest period. Next, participants experienced two 5-minute cue exposure bouts separated by 20 minutes. The cues were presented twice per session to maximize effects, as larger effect sizes have been shown for multiple (vs. single) cue presentations in studies with people who smoke cigarettes (Betts et al., 2020). Subjective questionnaires (e.g., craving, mood) were administered immediately before and after each of the two cue bouts. After both bouts, participants completed two batteries of behavioral economic tasks, counterbalanced by task version (puffs vs. minutes) and separated by 30 minutes. Participants were compensated $50/session ($100 total). Study procedures were approved by the university’s Institutional Review Board.

Figure 1.

Timeline (in minutes) for each of two sessions, which differ by the type of cue presented (ECIG or neutral).

S = subjective measures (i.e., e-QSU, Schuh-Stitzer ECIG Craving, PANAS). The break in the timeline represents a 10-minute rest period between cue-exposure bouts.

2.3. Cues

Cues were presented via 5-minute picture slideshows. Each set of cues contained 50 unique pictures, displayed for 6 seconds each, in a pre-randomized order that was the same for each participant. Of the 50 ECIG cues, half (n=25) featured one or more people using an ECIG and half (n=25) featured only the device and/or liquid product (see Supplementary Figure 1 for more details). Neutral cue images depicted people drinking water (n=25) or water products (n=25; e.g., glasses/bottles of water) (as in Dowd & Tiffany, 2019). The ECIG and neutral cues were matched on characteristics such as the presence of a person and the environment (i.e., indoors vs. outdoors). The people depicted were different across cue types, and represented different races (56-60% White; 8-12% Black; 8% Asian; 20-28% undetermined) and genders (40% women; 32-36% men; 24-28% undetermined or multiple genders). The same cue set (ECIG or neutral) was presented twice during each counterbalanced session, and research staff were blinded to cue type.

2.4. Baseline Measures

2.4.1. ECIG Dependence.

As there currently exists no gold standard for measuring ECIG dependence, two different measures were administered. The Penn State Electronic Cigarette Dependence Index (PSECDI) consists of 10 items summed: not dependent (0-3) or low (4-8), medium (9-12) or high (13-20) dependence. (Foulds et al., 2015). The E-Cigarette Dependence Scale (EDS-4) consists of 4 items summed to create a score that ranges from 0-4 with higher scores indicating higher dependence (Morean et al., 2019).

2.4.2. Impulsiveness.

The Barratt Impulsiveness Scale (BIS-11) includes 30 items (e.g. “I do things without thinking”; “I change hobbies”) rated on a 4-point scale (1=rarely/never to 4=almost always/always) and collapsed into attentional, motor, and non-planning factors (Patton et al., 1995). BIS-11 total scores range from 30-120 with higher scores indicating greater levels of impulsiveness. The standard Monetary Choice Questionnaire (MCQ) (Kirby et al., 1999) contains 27 items for which participants choose between a smaller amount of money immediately ($55 now) or a larger amount of money after a delay ($75 after 61 days). The dependent variable is the rate of delay discounting (k), with higher k values indicating steeper discounting of delayed rewards (i.e., higher impulsivity).

2.5. Outcome Measures

2.5.1. ECIG Craving and Anticipated Relief from Withdrawal.

The Tiffany-Drobes Questionnaire of Smoking Urges Brief (Cox et al., 2001) includes 10 items that were adapted for ECIGs (and thus referred to here as e-QSU-Brief). Items were summed (e.g., Felicione et al., 2022) and collapsed into two factors: Factor 1 (desire to vape; score range=5-35) and Factor 2 (anticipated relief from withdrawal; score range=5-35). The Schuh-Stitzer Craving Scale (Schuh and Stitzer, 1995) contains 4 items, also adapted here for ECIGs, each presented on a 0-100 visual analogue scale.

2.5.2. Mood.

The Positive and Negative Affect Schedule (PANAS; Watson et al., 1988) consists of 20 items addressing positive (e.g., excited, proud) and negative (e.g., upset, nervous) mood states. Participants rate how they feel “right now” (1=very slightly/not at all to 5=extremely), and higher scores on each scale (range=10-50) indicate higher positive or negative mood.

2.5.3. Behavioral Economic Tasks.

All behavioral economic tasks were administered using two versions, quantifying ECIG use either in puffs or in minutes of access (Supplementary materials). Both versions were included given that the best unit of measurement for quantifying ECIG consumption remains unknown, and may differ depending on the type of device used. For both, participants were instructed to assume that the ECIG use would be from their own device and for their own personal use (i.e., puffs/access could not be given away).

The MCQ described above (see 2.4.2) also was adapted for this study to measure delay discounting related to ECIG use. Henceforth referred to as the e-MCQ (to differentiate from the standard MCQ), participants choose between a smaller amount of ECIG use immediately (13 puffs now) or a larger amount of ECIG use after a delay (20 puffs after 61 minutes). Higher k values indicate steeper discounting of delayed rewards.

For the ECIG purchase task, participants were instructed to enter the number (open-field format) of puffs or minutes of access they would purchase at increasing prices ($0.01 to $100.00) (Johnson et al., 2017). Several demand indices were obtained: intensity (amount purchased at $0.01), breakpoint (highest price at which any commodity was purchased), O_max (maximum amount spent on commodity), P_max (price at which maximum amount was spent), and elasticity (purchasing sensitivity as a function of increasing price).

The Multiple-Choice Procedure (MCP) required participants to choose between a hypothetical fixed amount of ECIG use (10 puffs; 10 minutes of access) and an increasing amount of money ($0.01 to $5.12) (Vansickel et al., 2012). The dependent variable was the crossover point, or the last price at which ECIG use was chosen over money. The crossover point was recorded as $5.12 when ECIG use was selected over money on all trials, and as $0.00 when money was chosen over ECIG use on all trials.

2.5.4. Heart Rate.

Heart rate (beats per minute [bpm]), included as secondary outcome, was recorded every 20 seconds throughout sessions via a pulse oximeter placed on participants’ preferred hand (506 NP3, Criticare Systems, Inc).

2.6. Data Analysis

Data were analyzed using R Statistical Software 4.2.3 (R Core Team, 2023) with the libraries lme4 (Bates et al., 2015), emmeans, and beezdemand (Kaplan, 2018). Data from the subjective measures were analyzed using linear mixed-effects models (LMMs) fitted by restricted maximum likelihood to assess effects of cue type (ECIG vs. neutral) × bout (one vs. two) × time (pre- vs. post-cue exposure). Bout and time were included as fixed factors, while random slopes were modeled for cue type to allow for variability of the cue type effects across participants. Random intercepts were modeled for participants. Sum coding was applied to categorical variables to assess the effect of each factor on the dependent variable while collapsing across other factors in the model (Brehm & Alday, 2022). Post-hoc comparisons were conducted using emmeans with Tukey’s p-value adjustment (p’s<.05).

For the e-MCQ, an Excel-based tool (Kaplan et al., 2016) generated overall log-transformed k values for each participant and cue condition. Individual-level responding was examined for consistency, with those below the recommended cutoff of 75% (n=1 for each task version) excluded from analyses (Kaplan et al., 2016). Differences in log-transformed discounting rates as a function of cue condition were analyzed using LMMs with cue type set as a fixed factor and random intercepts modeled for participants. The MCQ (monetary) was included as a covariate to account for baseline level of impulsiveness.

ECIG purchase task data were checked for nonsystematic responding using pre-defined criteria (Stein et al., 2015; trend, bounce, absence of reversals from zero). Nine participants failed ≥1 criterion and were excluded from analyses. An additional n=1 (puffs) and n=2 (minutes) were excluded for implausible purchases (e.g., 20000 puffs; >1440 minutes). All demand indices were observed from the data, except for elasticity which was calculated using the exponentiated demand equation (Koffarnus et al., 2015). Demand curves were fitted with k set to 4 to provide the best fit. Paired-samples t-tests (p<.05) compared log-transformed purchase task demand indices and MCP crossover points between cue conditions. For the behavioral outcomes, a sensitivity analysis was conducted such that order of task version (i.e., puffs vs. minutes) across conditions was included as a covariate. Importantly, the pattern of results remained the same.

Heart rate data were averaged into 5-min bins for pre-, during, and post-cue exposure timepoints, and analyzed using a LMM with cue type (ECIG vs. neutral) × bout (one vs. two) × time (pre vs. during vs. post) as fixed factors, and with random intercepts modeled for participants. Significance is reported at p<.05.

3. RESULTS

Participant characteristics (N=42) are shown in Table 1, and LMM results are shown in Table 2.

Table 1.

Demographic Characteristics (N=42)

	M (SD) or N (%)
Age (years)	21.17 (3.13)
Gender
Women	27 (64.30%)
Men	15 (35.70%)
Race
White	33 (78.60%)
Black	1 (2.40%)
Multiracial	4 (9.50%)
Other/Unknown	4 (9.50%)
Ethnicity
Hispanic	9 (21.40%)
Non-Hispanic	33 (78.60%)
Education (years)	14.81 (2.55)
# Cigarettes in Lifetime	14.95 (24.43)
ECIG use
Duration (years)	2.79 (1.53)
Days/week	6.50 (0.86)
ECIG Device Type
Disposable1	31 (73.81%)
Reusable Pod-Style2	12 (27.27%)
Undetermined	1 (2.27%)
ECIG Brand
Hyde	20 (47.62%)
ELF Bar	5 (11.90%)
JUUL	5 (11.90%)
Vuse	5 (11.90%)
Other	5 (11.90%)
Multiple	2 (4.76%)
Liquid Flavor
Fruit	16 (38.10%)
Menthol/Mint	15 (35.71%)
Blue Razz Ice	7 (16.67%)
Other	4 (9.52%)
Liquid Nicotine Concentration (%)3	4.93 (0.45)
ECIG Dependence
PSECDI4	10.60 (4.36)
EDS-45	2.07 (0.76)
BIS-11 Total6	64.21 (9.79)

¹Fully disposable with prefilled pod/cartridge containers

²Reusable with prefilled or refillable pod/cartridge containers

³n=37

⁴Penn State ECIG Dependence Index (range 0-20)

⁵Electronic Cigarette Dependence Scale (range 0-4)

⁶Barratt Impulsiveness Scale-11 (range 30-120)

Table 2.

Mixed-Effects Models Predicting Subjective and Physiological Outcomes

Outcome Measure	Estimate	SE	t	p	95% CI
					Lower	Upper
e-QSU-Brief Factor 1
Intercept	23.32	1.23	18.95	<.001	20.88	25.76
Cue type (ECIG)	0.82	0.41	2.02	.050	0.02	1.63
Bout (2)	0.49	0.15	3.27	.001	0.20	0.77
Time (Post)	0.34	0.15	2.27	.024	0.05	0.62
Cue type (ECIG) × Bout (2)	−0.01	−0.15	−0.06	.952	−0.30	0.28
Cue type (ECIG) × Time (Post)	0.45	0.15	3.03	.003	0.16	0.74
Bout (2) × Time (Post)	0.11	0.15	0.74	.459	−0.18	0.40
Cue type (ECIG) × Bout (2) × Time (Post)	−0.25	0.15	−1.70	.090	−0.54	0.04
e-QSU-Brief Factor 2
Intercept	13.59	1.29	10.51	<.001	11.03	16.15
Cue type (ECIG)	0.44	0.32	1.36	.181	−0.20	1.07
Bout (2)	0.38	0.12	3.19	.002	0.15	0.62
Time (Post)	0.29	0.12	2.40	.017	0.05	0.52
Cue type (ECIG) × Bout (2)	0.10	0.12	0.82	.415	−0.14	0.33
Cue type (ECIG) × Time (Post)	0.24	0.12	2.00	.046	0.01	0.47
Bout (2) × Time (Post)	0.01	0.12	0.07	.941	−0.22	0.24
Cue type (ECIG) × Bout (2) × Time (Post)	−0.07	0.12	−0.62	.537	−0.31	0.16
Schuh-Stitzer ECIG Craving
Intercept	56.36	3.92	14.37	<.001	48.59	64.14
Cue type (ECIG)	3.08	1.03	2.98	.005	1.03	5.13
Bout (2)	2.15	0.40	5.44	<.001	1.38	2.92
Time (Post)	1.30	0.40	3.30	.001	0.54	2.07
Cue type (ECIG) × Bout (2)	−0.06	0.40	−0.14	.888	−0.82	0.71
Cue type (ECIG) × Time (Post)	0.93	0.40	2.36	.019	0.16	1.70
Bout (2) × Time (Post)	−0.53	0.40	−1.35	.179	−1.30	0.24
Cue type (ECIG) × Bout (2) × Time (Post)	−0.93	0.40	−2.36	.019	−1.70	−0.16
PANAS Positive Affect
Intercept	19.60	0.95	20.59	<.001	17.71	21.48
Cue type (ECIG)	0.13	0.50	0.26	.796	−0.86	1.13
Bout (2)	−1.33	0.13	−10.24	<.001	−1.59	−1.08
Time (Post)	−0.46	0.13	−3.52	.001	−0.71	−0.21
Cue type (ECIG) × Bout (2)	−0.01	0.13	−0.09	.927	−0.27	0.24
Cue type (ECIG) × Time (Post)	0.04	0.13	0.32	.749	−0.21	0.29
Bout (2) × Time (Post)	0.22	0.13	1.69	.092	−0.03	0.47
Cue type (ECIG) × Bout (2) × Time (Post)	−0.14	0.13	−1.05	.294	−0.39	0.12
PANAS Negative Affect
Intercept	14.23	0.69	20.66	<.001	12.87	15.60
Cue type (ECIG)	0.59	0.21	2.75	.009	0.16	1.01
Bout (2)	0.29	0.09	3.28	.001	0.12	0.46
Time (Post)	0.21	0.09	2.33	.021	0.03	0.38
Cue type (ECIG) × Bout (2)	0.21	0.09	2.33	.021	0.03	0.38
Cue type (ECIG) × Time (Post)	0.26	0.09	2.94	.004	0.09	0.44
Bout (2) × Time (Post)	−0.02	0.09	−0.27	.790	−0.20	0.15
Cue type (ECIG) × Bout (2) × Time (Post)	−0.04	0.09	−0.48	.635	−0.22	0.13
Heart Rate1
Intercept	70.45	1.51	46.67	<.001	67.45	73.45
Cue type (ECIG)	0.51	0.19	2.73	.007	0.15	0.86
Bout (2)	−0.76	0.18	−4.30	<.001	−1.11	−0.42
Time (Post)	−0.08	0.25	−0.33	.743	−0.57	0.40
Time (Pre)	0.86	0.25	3.41	.001	0.37	1.34
Cue type (ECIG) × Bout (2)	−0.12	0.18	0.65	.515	−0.46	0.23
Cue type (ECIG) × Time (Post)	0.02	0.25	0.08	.940	−0.47	0.50
Cue type (ECIG) × Time (Pre)	0.13	0.25	0.53	.599	−0.35	0.62
Bout (2) × Time (Post)	−0.03	0.25	−0.11	.909	−0.51	0.46
Bout (2) × Time (Pre)	0.07	0.25	0.28	.780	−0.41	0.55
Cue type (ECIG) × Bout (2) × Time(Post)	0.18	0.25	0.71	.477	−0.31	0.66
Cue type (ECIG) × Bout (2) × Time (Pre)	−0.14	0.25	−0.54	.588	−0.62	0.35

Note. Significant items are bolded.

¹‘During’ cue exposure is reference group for variable ‘time’ in heart rate model.

3.1. ECIG Craving and Anticipated Relief from Withdrawal

For the e-QSU-Brief Factor 1 (Figure 2A), there was a significant cue type × time interaction (b=0.45, SE=0.15, t=3.03, p=.003) such that ECIG craving increased significantly from pre- to post-exposure for ECIG cues (Tukey’s p<.05), but not for neutral cues. Scores also were higher at the post-exposure timepoint for ECIG cues than for neutral cues (Tukey’s p<.05). There also was a main effect of bout (b=0.49, SE=0.15, t=3.27, p=.001), with ratings of ECIG craving higher for bout two than one (collapsed across cue type and time). Likewise, for Factor 2 (Figure 2B) there was a significant cue type × time interaction (b=0.24, SE=0.12, t=2.00, p=.046); scores of anticipated relief from ECIG withdrawal increased significantly from pre- to post-exposure for the ECIG cues (Tukey’s p<.05) but not neutral cues. Finally, there was a main effect of bout (b=0.38, SE=0.12, t=3.19, p=.002), with scores increased from bout one to bout two.

Figure 2.

Mean (+/− SEM) values for (A) e-QSU-Brief Factor 1 (desire to vape; score range=5-35); (B) e-QSU-Brief Factor 2 (anticipated relief from withdrawal; score range=5-35); (C) Schuh-Stitzer ECIG Craving (score range=0-100); and (D) PANAS Negative Affect (score range=10-50). Asterisks (*) denote a significant difference from baseline (Pre-Bout 1) within a cue condition. Pound signs (#) denote a significant difference between ECIG and neutral cue conditions at that timepoint. The break in the x-axis represents the 10-minute rest period between cue-exposure bouts.

The model predicting ECIG craving via the adapted Schuh-Stitzer scale (Figure 2C) revealed a significant three-way interaction of cue type × bout × time (b=−0.93, SE=0.40, t=−2.36, p=.019). For the ECIG cues, ratings of ECIG craving increased significantly from pre exposure at the first bout to all other timepoints: post-exposure at bout one, and pre- and post-exposure at bout two (Tukey’s p<.05). In the neutral cue condition, scores increased significantly from pre-exposure at bout one to post-exposure at bout two (Tukey’s p<.05).

3.2. Mood

For the PANAS negative affect scale (Figure 2D), there were significant interactions between cue type × time (b=0.26, SE=0.09, t=2.94, p=.004) and cue type × bout (b=0.21, SE=0.09, t=2.33, p=.021). For cue type × time, negative affect increased significantly from pre- to post-cue exposure in the ECIG cue condition (Tukey’s p<.05) but not the neutral cue condition. Across cue conditions, negative affect scores were significantly higher post-exposure for ECIG cues compared to neutral cues (Tukey’s p<.05). For the cue type × bout interaction, negative affect increased significantly from bout one to bout two for ECIG cues (Tukey’s p<.05), but not for neutral cues. Moreover, negative affect scores were significantly higher for ECIG cues than neutral cues at bout 2 (Tukey’s p<.05).

For the PANAS positive affect scale, only significant main effects of bout (b=−1.33, SE=0.13, t=−10.24, p<.001) and time (b=−0.46, SE=0.13, t=−3.52, p=.001) were revealed. Positive affect decreased significantly from bout one (M=20.93, SE=0.96) to bout two (M=18.26, SE=0.96), collapsed across cue type and time. It also decreased significantly across timepoints (collapsed across cue type and bout): 20.05 (SE=0.96) for pre-cue exposure versus 19.14 (SE=0.96) for post-cue exposure.

3.3. Behavioral Economic Tasks

Table 3 shows descriptive statistics for the behavioral economic tasks across ECIG and neutral cue conditions. Cue type significantly influenced delay discounting (e-MCQ) for the minutes version of the task (b=0.11, SE=0.05, t=2.19, p=.034), with ECIG cues eliciting steeper discounting than neutral cues. No effects of cue type were observed for the puffs version (b=0.01, SE=0.03, t=0.52, p=.608).

Table 3.

Descriptive Statistics for Behavioral Economic Tasks

	ECIG Cues M (SE)	Neutral Cues M (SE)
Puffs Version Tasks
Delay Discounting
e-MCQ k	0.06 (0.06)	0.06 (0.07)
ECIG Purchase Task
Intensity	492.34 (66.62)	516.56 (64.99)
Breakpoint	17.61 (5.13)	14.01 (3.54)
Omax	33.38 (9.79)	24.62 (6.09)
Pmax	13.14 (5.25)	11.21 (3.57)
Elasticity (α)	0.009 (0.002)	0.009 (0.002)
Multiple Choice Procedure
Crossover Point	1.99 (1.76)	1.81 (1.56)
Minutes Version Tasks
Delay Discounting
e-MCQ k	0.05 (0.07)	0.04 (0.06)
ECIG Purchase Task
Intensity	447.90 (64.82)	615.26 (68.96)
Breakpoint	16.36 (4.57)	16.43 (4.56)
Omax	31.25 (9.81)	28.55 (6.66)
Pmax	14.44 (4.65)	15.84 (4.62)
Elasticity (α)	0.007 (0.001)	0.007 (0.002)
Multiple Choice Procedure
Crossover Point	1.90 (1.78)	1.75 (1.53)

Note. Significant outcomes are bolded.

For the ECIG purchase task, no effects of cue type were observed for the puffs version (t’s(31)=−0.86-0.72, p’s=.394-.959, d’s=−0.15-0.13), or for the minutes version (t’s(30)=−0.67-1.00, p’s=.327-.539, d’s=−0.12-0.18) with one exception. For this latter task version, there was a significant effect of cue type on demand intensity (the amount purchased at $0.01) (t(30)=−2.15, p=.040, d=−0.39); however, intensity was unexpectedly higher for neutral cues than ECIG cues. There also were no effects of cue type on log-transformed MCP crossover points on the puffs (t(41)=0.77, p=.445, d=0.12) or minutes (t(41)=−0.55, p=.586, d=−0.08) task versions.

3.4. Heart Rate

For heart rate, only main effects were found to be significant: cue type (b=0.51, SE=0.19, t=2.73, p=.007), bout (b=−0.76, SE=0.18, t=−4.30, p<.001), and time (pre- vs. during cue exposure) (b=0.86, SE=0.25, t=3.41, p=.001). Heart rate (bpm) was significantly higher for ECIG cues (M = 70.96, SE = 1.52) relative to neutral cues (M = 69.95, SE = 1.52) (collapsed across bout and time), at bout one (M = 71.21, SE = 1.52) relative to bout two (M = 69.69, SE = 1.52) (collapsed across cue type and time), and pre-cue exposure (M = 71.31, SE = 1.53) relative to during exposure (M = 70.37, SE = 1.53) (collapsed across cue type and bout).

4. DISCUSSION

Among individuals who use ECIGs but not cigarettes, exposure to ECIG cues (vs. neutral cues) significantly increased ECIG craving and anticipation of relief from ECIG withdrawal. Findings are consistent with previous work examining ECIG cue exposure among individuals who have a history of cigarette smoking (King et al., 2021; Vena et al., 2020a). Such effects also are generally consistent across multiple modalities of cue delivery, including pictures (current study), confederates (King et al., 2018, 2021; Vena et al., 2020a), and videos (Blackwell et al., 2020; King et al., 2016; but see Nichols et al., 2016). Results also demonstrated an increase in negative mood in response to ECIG (vs. neutral) cues, an effect shown previously for individuals who smoke cigarettes in response to smoking cues (Drobes and Tiffany, 1997; Wray et al., 2011).

Interestingly, ECIG cue exposure increased hypothetical delay discounting when ECIG use was measured in minutes of access but not in puffs. Minutes of ECIG use offers more flexibility in that users can take as many or as few puffs as they desire within the specified time period, though also provides no information on the amount of ECIG liquid consumed. In the current study, the puff numbers used in the task (i.e., maximum=22) may have exceeded the number desired by our sample, masking any cue effects. We did not ask participants to self-report their average number of puffs taken or minutes vaped per day, thereby limiting our ability to determine if the quantities were representative of their usual patterns. Notably, some individuals who use ECIGs may have trouble estimating their consumption in puffs (Cassidy et al., 2017), while others prefer this unit of measurement (Liu et al., 2021). The most appropriate unit for quantifying ECIG use is unknown, and may differ depending on the type of ECIG device used. Individuals who use pod-style ECIGs, which were the preferred device types among the current sample, have described their use in terms of the frequency of replacing their pod (reusable products) or their entire device (fully disposable products) (Liu et al., 2021). It has been suggested that researchers consider such device-specific units for behavioral economic assessments when sampling individuals whose device types are homogenous (Strickland et al., 2021).

Additionally, ECIG cue exposure did not significantly increase behavioral economic demand for ECIGs (via the purchase task) or choice for ECIGs (via the MCP). Research examining motivation for nicotine has found similar dissociations between subjective (e.g. satisfaction) and behavioral economic (e.g., demand) data (Cohn et al., 2020). Acute drug motivation is multidimensional (MacKillop et al., 2012), and dimensions may differ in their ability to predict clinically relevant outcomes. Indeed, studies have found that both delay discounting (MacKillop & Kahler, 2009) and behavioral economic demand (MacKillop & Murphy, 2007) predict drug-use behavior. The predictive validity of cue-induced subjective craving on drug-use behavior has been debated (Perkins, 2009), though a recent meta-analysis suggests it is reliably associated with drug use (Vafaie & Kober, 2022). Characteristics of the study may also explain the lack of significant findings for the behavioral economic outcomes. Perhaps the ECIG cues used in this study were not sufficiently salient to affect these outcomes. It is also possible that participants experienced survey fatigue, or the cue effects had dissipated by the time participants completed the second battery of behavioral tasks. For the purchase task specifically, another possibility is that the participants misunderstood or did not attend to the tasks. Indeed, nearly one quarter of participants responded non-systematically on the purchase task and were excluded from the analyses. Future work may benefit from requiring more intensive training on the task, constraining the maximum amount available for purchase, or asking participants to confirm their responses (Stein et al., 2015). For the MCP, findings may have been limited by the maximum monetary amount of $5.12. Some work has used this same amount (Vansickel et al., 2012), and others have used a higher maximum value (i.e., $10.24) (Breland et al., 2020; Maloney et al., 2020). In these latter studies, the average crossover points for own-brand ECIGs were on the lower end of the scale ($1.35-1.50). Similarly, the average crossover points observed here were relatively low ($1.75-1.99), and participants chose ECIG use over the maximum value on only 16% of trials. Still, future research using the MCP may consider using a higher maximum value.

Several study limitations may have influenced findings. Pre-session ECIG abstinence was not biochemically verified. While Betts and colleagues (2020) found little evidence that abstinence moderates cue-induced craving in people who smoke cigarettes, the influence of abstinence on ECIG cue reactivity is unknown. Also, the behavioral economics tasks were hypothetical in nature. Relative to real cigarette-related rewards, hypothetical rewards have produced comparable effects for purchase tasks (Wilson et al., 2016) but less pronounced effects for delay discounting tasks (Lawyer et al., 2022). Another possible limitation involves the ECIG cues. Even for some effects that were significant, the magnitude of difference was arguably small. Many of the images depicted pod-style devices (i.e., JUUL, Hyde Edge) preferred by our sample, though slightly more depicted mod-style devices (i.e., Aspire ESP). Future research should explore the degree to which the personal relevance of cues affects cue reactivity or consider matching cues to participants’ preferred devices. A different modality of cue exposure also may have produced stronger effects. For cigarette smoking, pictorial cues generate larger effect sizes than other modalities (e.g., in-vivo, video) (Betts et al., 2020); however, the strongest modality for people who use ECIGs is unknown. With use of a confederate vaping in-person, participants would be exposed to not only the device but also the sight and smell of the exhaled aerosol and the social cue of someone vaping. We had concerns about navigating the use of a confederate, as the study began when many COVID restrictions were still in place, including the use of masks and distancing. These factors may have further impacted the laboratory environment, consequently affecting participants’ mood or behavior.

In conclusion, this study is the first to demonstrate that ECIG cues increase ECIG craving and negative mood in individuals who use ECIGs but have little cigarette smoking experience. Given that many of these individuals report a desire to quit (Douglas et al., 2023), their ability to do so may be hindered by exposure to ECIG cues. Only 19 states, the District of Columbia, and Puerto Rico have indoor air laws that prohibit the use of ECIGs in places such as private worksites, restaurants, and bars (Centers for Disease Control and Prevention, 2024). Repeated exposure to ECIG cues in these environments may perpetuate craving for ECIGs, making quitting difficult and promoting relapse. Thus, study results have the potential to information regulations related to environmental exposures.

Highlights.

• ECIG cues increased ECIG craving and negative mood in people who use ECIGs exclusively

• ECIG cues increased discounting of delayed minutes of ECIG access but not ECIG puffs

• ECIG cue exposure did not increase behavioral economic demand for ECIGs

• Exposure to ECIG cues may perpetuate ECIG craving, making quitting difficult

Data Availability Statement

Data will be available upon reasonable request to the corresponding author.