Examining Electronic Nicotine Delivery System (ENDS) Cue Reactivity Using a Remote Platform: Cigarette and ENDS Desire in Dual Users

Abstract

Introduction

Prior research shows that in-person exposure to electronic nicotine delivery systems (ENDS) use increases the desire for cigarettes and ENDS. However, less is known about the impact of cues delivered during remote interactions. This study extends previous in-person cue work by leveraging a remote confederate-delivered cue-delivery paradigm to evaluate the impact of dual nicotine vaping (vs. sole smoking) on reactivity to an ENDS cue in individuals who smoke cigarettes.

Methods

N = 52 dual users (DU; current users of both combustible cigarettes and ENDS) and N = 54 sole smokers (SS; users of combustible cigarettes only) observed a study confederate drinking bottled water (control cue) and then vaping an ENDS (active cue). Changes in desire for cigarettes and ENDS were compared between groups post-cue exposure.

Results

Multilevel models, controlling for sex and cigarettes per day, revealed that the remote ENDS cue, but not water, significantly increased the desire for both cigarettes and ENDS. Relative to SS, DU reported greater post-ENDS cue increases in ENDS desire but not cigarette desire.

Conclusions

A remote, confederate-delivered ENDS cue generalizes as a smoking and vaping cue, with DU showing greater reactivity than SS. This study provides the first evidence for the validity and feasibility of a remote, confederate-delivered ENDS cue reactivity paradigm.

Implications

This study provides support for the use of a remote platform, an increasingly popular method of conducting research since the onset of the COVID-19 pandemic, to employ confederate-delivered ENDS cues. Frequent observations of vaping via remote platforms and social media may contribute to the maintenance of single- and dual-product use.

Introduction

Combustible tobacco use remains the leading cause of preventable death in the United States, and 28.3 million Americans smoke cigarettes.¹ Although combustible cigarette smoking has declined, vaping using electronic nicotine delivery systems (ENDS, or e-cigarettes) persists at high levels, and dual use of both ENDS and cigarettes is increasing.^1–4 The Population Assessment of Tobacco and Health study revealed that almost half of adults who use nicotine use multiple forms of tobacco, including dual use of cigarettes and ENDS.⁵ Individuals who smoke commonly turn to ENDS products to attempt smoking reduction or cessation, circumvent smoke-free policies, and enjoy shared features of these products.⁶ While ENDS use may be associated with fewer adverse health outcomes than cigarette smoking, especially if a chronic user of cigarettes switches completely, many dual users (ie, users of electronic and combustible tobacco) continue to smoke the same amount of cigarettes as before initiating ENDS use.⁷ This is concerning as dual use may exacerbate negative health outcomes associated with combustible smoking, including an increased risk of lung conditions like chronic obstructive pulmonary disease and asthma, cardiovascular disease, and more severe nicotine dependence.^7–9

ENDS use closely resembles combustible smoking behaviors including inhalation, exhalation, and hand-to-mouth movements. Given these similarities in form and function, ENDS may act as a Pavlovian cue triggering vaping and smoking urges in observers. Indeed, for young adults who smoke, direct exposure to ENDS and heated tobacco product use serves as a smoking cue, inducing both cigarette desire and smoking behavior.^10–15 Additionally, a recent combined analysis examining cue reactivity among a range of tobacco users showed that dual users exhibited greater ENDS cue reactivity in terms of confederate-delivered ENDS cues increasing ENDS desire and smoking urge and decreasing latency to smoke compared with sole smokers.¹⁶ However, these effects have largely been studied as part of in-person, laboratory paradigms. Tobacco cues presented as still images or videos can be delivered remotely and increase smoking urge among people who smoke.^17–20 However, these cues typically lack social interaction, and ecological momentary assessment studies show that individuals are more likely to use tobacco products when socializing²¹ or exposed to others smoking^22,23 in their natural environments. Together, these results suggest that social interaction is a key part of smoking behavior, and cue paradigms that balance cue standardization with ecological validity can help elucidate key factors maintaining tobacco/nicotine use behaviors.

Indeed, prior literature suggests that in-person cues elicit comparable cue effects to other modalities, though findings across studies are mixed.^18,24–26 Still, there is a strong theoretical rationale for prioritizing human interaction in cue-delivery paradigms and previous cue reactivity work has shown that in-person, confederate-delivered cues elicit the desire to smoke cigarettes and use ENDS.^26,27 However, conducting such studies is labor-intensive, expensive, and, as the COVID-19 pandemic showed, infeasible when there are concerns about transmitting illness associated with exhaled aerosols. As remote interactions are now commonplace, developing a method to implement confederate-delivered cue paradigms may circumvent some of the limitations inherent in onsite cue delivery while maintaining ecological validity. This method may also facilitate more efficient data collection and allow for more generalizable national samples.

To this end, the current study extended prior studies on ENDS cue reactivity by developing and testing a fully remote, confederate-delivered cue exposure paradigm. We examined ENDS cue reactivity in an adult smoker sample and compared cue-elicited cigarette and ENDS desire in dual users (DU) and sole smokers (SS). We hypothesized that remote ENDS cues would increase desire for both ENDS and cigarettes and that these effects would be particularly pronounced in DU. We also aimed to evaluate and identify the core components of this remote methodology including strengths, challenges, and considerations.

Methods

Recruitment

Candidates were recruited nationally via social media advertisements (eg, Instagram, Facebook, Reddit) and completed an online screening survey assessing basic eligibility criteria. Eligible candidates were scheduled for a remote screening session via video conferencing to confirm study eligibility. To minimize expectancies or recruitment bias, as in our past studies,^10,12–14 participants were told that the purpose of the study was to examine moods, behaviors, and social interactions during two randomly assigned 5-minute tasks (ie, eating a snack, drinking water, leading a conversation, viewing videos, or smoking or vaping). To maintain the guise of random assignment to a variety of tasks, they were also told that they would not be asked to engage in a behavior that is not typical for them (eg, someone who does not currently vape e-cigarettes would not be randomized to a vaping condition, vegetarians would not be asked to eat meat products, and so forth). Lastly, the screening survey minimized the number of assessment items pertaining to nicotine/tobacco products and included measures related to other health behaviors to avoid a clear emphasis on tobacco use behaviors.

Participants

Inclusion criteria were age 18–60 years, good general health, ability to understand and read English, no current substance use disorder (excluding tobacco use disorder), and no active or untreated suicidal ideation, hallucinations, delusions, or other severe psychiatric symptoms in the past 6 months. Overall smoking criteria included current cigarette smoking (≥7 cigarettes/week for past 1 year) and no current use of smoking cessation medications or plans to quit. DU endorsed current weekly nicotine vaping (≥1 day/week) for at least the past 3 months and denied a desire or plan to cut down or quit vaping. In contrast, SS reported no nicotine vaping in the past year and a maximum lifetime history of nicotine vaping once monthly for up to 4 months. The latter criteria were employed to enhance the generalizability of the sample as infrequent, experimental use of vape products is common among combustible cigarette smokers,²⁸ while also avoiding any past regular ENDS use that may have compromised the differentiation of the DU and SS subgroups.

Screening

The screening session was conducted remotely on a video conferencing platform during daytime hours, with a typical start time of 1:30 pm ± 1.9 hours local time for participants. Prior to the screening visit, candidates were instructed to abstain from recreational drugs and alcohol for at least 24 hours and from smoking and vaping for at least 1 hour. The screening session lasted approximately 30 minutes and consisted of informed consent, explanation of study procedures, and completion of screening surveys to confirm study eligibility. Surveys included general demographic, health, and substance use questionnaires including lifetime smoking, vaping, and cannabis use/vaping behaviors. Trained staff administered a modified 7-day Timeline Follow-Back interview to obtain estimates of daily cigarette smoking and nicotine vaping, and the research version of the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, Fifth Edition (SCID-5)²⁹ to screen for current psychiatric illness.

Approximately 53% of screened candidates (131/246) were deemed eligible to participate. Ineligible candidates did not meet smoking criteria (n = 65), endorsed current untreated psychiatric symptoms (n = 21), reported same-day cannabis use (n = 7), or were currently seeking treatment for smoking or vaping (n = 9). The sample was geographically diverse with 68% residing in the Midwest, 17% in the Northeast/East, and 15% in the Southwest/West regions of the United States.

Remote Session Protocol

After screening, eligible candidates were scheduled for the 90-minute remote experimental session. All sessions were conducted in the afternoon, with an average local time session initiation of 2:20 pm ± 2.03 hours. Participants were asked to participate in a private space where they were unlikely to be interrupted. They were then asked to show that they had each item available that could be needed for the 5-minute tasks, that is, a snack item, a water bottle, their cigarettes, and their ENDS device (if a dual user), to maintain the guise that they may be randomized to engage in a variety of behaviors during the study. Once the research assistant verified access to those items, participants were instructed to place all items in a bag and remove them from their immediate surroundings to avoid potential cross-cue contamination during the exposures. Participants were asked to show their participation space on camera to confirm there were no substance-relevant items present.

The session procedures were adapted from methods outlined in our previous studies^{11–14,16,30,31} and modified for a remote platform. Each session began with the participant completing baseline surveys (for details, see Dependent Measures) followed by study “randomization.” The research assistant used the share screen feature to show a list of possible tasks they may be randomly selected completely. The research assistant then used an online generator to “randomly” assign the participant to a task. The generator was pre-fixed such that the study participant unknowingly was always assigned to the conversation task. To facilitate the conversation with the other participant (who was the study confederate), a conversation topic list (favorite TV shows/movies, pets, travel, food, seasonal activities) was given to the participant to select a topic, and interaction prompts to initiate and facilitate the interaction were also provided.

The research assistant then admitted the study confederate into the video session from the remote waiting room and asked them to state their random task assignment. This assignment was also fixed such that the confederate was always “randomized” to drink water for this first task (control cue). To further enhance validity, the research assistant pretended to forget the confederate’s name and their task assignment on at least one occasion to maintain the guise of not knowing the confederate or their seemingly randomized task.

For both interaction periods, the confederate was trained to deliver the cue with approximately 8–10 hand-to-mouth movements in a natural way during the conversation. To mimic the frequent passive visual exposure to the products that may occur in real-life settings, the confederate was trained to maintain a view of the cue in their camera so that it could be seen by the participant during the interactions. After the first interaction, the confederate was sent back into the web-conference waiting room and the participant completed post-cue surveys including unrelated health and personality questionnaires to comprise a 20-minute wash-out period.

Similar procedures were followed for the second task with the participant seemingly receiving conversation as their task, and the confederate receiving the vaping as their task. Surveys were repeated immediately after the second interaction and both conversations were recorded and the number of hand-to-mouth movements were coded by two independent raters to ensure fidelity of the cue presentations. At the end of the study, participants answered a free response item asking them to state what they believed the purpose of the study was, and similar to our prior work,¹⁰ only a small portion (4.5%) of the sample correctly ascertained the study purpose.

Cues

A nonbranded, 12 oz. water bottle served as the control cue because drinking water is a common oral consummatory behavior with frequent hand-to-mouth movements; we used this cue in our prior studies and it is neutral in terms of associations with smoking or vaping.^13,30,32 The ENDS cue was a fourth-generation ENDS device¹⁴ of the confederate’s choice and was approximately 90.6 mm long × 25.47 mm wide, resembling a sleek flash drive. The device was consistent with the ENDS preference of our sample, with 100% (50% modifiable, 50% disposable) reporting fourth-generation ENDS as their preferred device. As stated earlier, these cues were used by the confederate during each conversation with the participant.

Standardization of cue delivery was confirmed by two independent raters rating the number of hand-to-mouth movements, that is, when the cue touched the confederate’s mouth, and positive valence ratings of the social interaction (Two-Dimensional Social Interaction Scale; 2DSIS).³³ The hand-to-mouth movements for all sessions were within the target range of 8–10 movements (M = 9.42 ± 2.1 SD).

Dependent Measures

The primary dependent measures for cue reactivity were the desire for “a cigarette of your preferred brand” and desire for “an electronic cigarette” each rated on a visual analog scale from “not at all” (0) to “most ever” (100).³² Secondary measures of study acceptability included the 2DSIS subscale scores of passive and active engagement. These outcomes were computed as the mean item scores (rated on a 1–4 point scale, with higher scores indicating more positive engagement) for positive passive engagement (agreeable, considerate, attentive, cooperative) and positive active engagement (friendly, spontaneous, talkative, energetic). Participant–confederate interactions were rated for active and passive participation by two independent raters.

Statistical Analyses

Preliminary Analyses

Participant demographics and smoking/vaping background characteristics were compared between DU and SS using independent-samples or chi-square tests, as appropriate. Sex and cigarettes smoked per day were included as covariates in all analyses to be consistent with prior studies.¹³ Three study participants who were significant outliers on the primary outcomes of desire for a cigarette and ENDS (ie, scores on either outcome >3 SD from the mean) were excluded from further analysis, resulting in a final sample of n = 106 (SS = 54; DU = 52). Of note, five individuals in the DU group reported current cannabis vaping with regular nicotine vaping but not in the past year; they were included in this subgroup as their vaping and smoking behaviors were more aligned with the DU than the SS group. Analyses were repeated by removing these participants and the main results did not change. Finally, to ensure standardization across cue sessions, independent-samples t-tests were performed, comparing the mean scores from the 2DSIS and interaction ratings at both time points.

Primary Analyses

We conducted hierarchical linear multilevel modeling with all models including baseline as the reference to test if cue type (hypothesis 1) and smoking group (DU vs. SS; hypothesis 2) were associated with outcomes (ie, post-ENDS cue desire for cigarettes and ENDS). A two-stage, multilevel approach was used to appropriately specify random effects in the model and accommodate the structure of a nested design. Individual beta weights from stage 1 models were examined to test hypothesis 1. Separate stage 1 models were fit for cigarette and ENDS desire, and both models included cue type (water vs. ENDS) and participant-level variables such as cigarettes/day and sex. Stage 2 models for cigarette and ENDS desire were fit by adding a smoking group by cue type interaction to test hypothesis 2. Beta weights were examined for interactive and main effects and these stage 2 models were compared to the stage 1 models via likelihood ratio test to assess model fit. All significant interactions were probed using pairwise comparisons assessing the effect of cue across smoking groups. Results were considered significant at p < .05. Effect sizes for predictors were assessed using estimated Cohen’s d. In addition to estimated Cohen’s d, conditional R² values were calculated to assess the total variance explained by each model.

All analyses were conducted using the lmerTest³⁴ and lme4³⁵ packages in RStudio.³⁶ Models were originally run using the Restricted Maximum Likelihood (REML) estimation method, but were re-fit using the Maximum Likelihood (ML) per recommendation by Bates et al.³⁵ for model comparison.

Results

Participant Characteristics

The sample was 54% female with a mean age of 38.5 ± 0.6 (SEM) years. The overall sample, n = 54 SS and n = 52 DU, reported a mean frequency of 6.8 ± 0.1 smoking days per week and an average of 10.6 ± 0.7 cigarettes per smoking day. Demographic and tobacco use behaviors between the groups are included in Table 1. Zero-order comparisons revealed that the two groups differed in cigarettes per day and baseline craving for ENDS. Participants in the SS group smoked more cigarettes per day, but the DU group exhibited similar levels of baseline cigarette craving and significantly greater ENDS craving, suggesting that the DU group experienced high levels of tonic craving for both cues before exposure.

Table 1.

Group Characteristics

	Dual users (DU; n = 52)	Sole smokers (SS; n = 54)	p
Demographics and background
Age (y)	38.4 (0.5)	38.7 (0.4)	.809
Sex (% female)a	27 (52%)	30 (56%)	.717
Race	26 (50%)	31 (57%)	.231
White	15 (29%)	18 (33%)	—
African American/Black	11 (21%)	5 (10%)	—
Other			—
Sexual orientation			.166
% Heterosexual/Straight	40 (77%)	43 (80%)	—
% Gay or Lesbian	9 (17%)	4 (7%)	—
% Other	3 (6%)	7 (13%)	—
Alcohol drinks/week	4.3 (0.8)	7.7 (1.2)	.055
Smoking patterns and use
Cigarettes smoked per smoking day	9.3 (0.4)	12.0 (1.0)	.015
FTND (nicotine dependency, 0–10)	4.8 (0.2)	4.9 (0.2)	.725
Hours since last reported cigarette	6.2 (1.6)	4.8 (1.4)	.543
ENDS patterns and use
Penn State ECDI (ENDS Dependence)	8.0 (0.62)	—
Puffs per day	18.3 (2.6)	—
Hours since last vape	7.1 [0.02–336]b	—
Previous ENDS use:
Naïve user (never used ENDS)	—	14 (26%)	—
Lifetime user (no past year regular use)	5* (10%)	40 (74%)	—
Current user (past year regular use)	47 (90%)	—	—
Baseline desire ratings
Regular Cigarette VAS	61.5 (4.7)	63.9 (3.8)	.876
ENDS VAS	51.1 (4.0)	6.2 (1.3)	<.001

Abbreviations: ENDS = Electronic nicotine delivery systems; FTND = Fagerström Test for Nicotine Dependence; VAS = visual analog scale.

Values are mean (SEM) or N (%), as indicated. Variables were compared between groups with Student’s t-test, chi-square, or Fisher’s exact test of independence, as appropriate. Variables were missing for participants that were included in analyses. Total ns may not reflect the full sample size.

^*The five individuals who were not current ENDS users remained in the dual use category due to their history of regular vaping in the past AND current cannabis vaping. Excluding these individuals did not change the results of the analyses.

^aParticipant’s biological sex.

^bHours since last vape are not normally distributed and are presented as median [range].

Social Interaction and Desirability Ratings

Ratings showed high active engagement for both the participant (3.6 ± 0.01 SEM) and confederate (3.7 ± 0.01), suggesting that both acted in a friendly, energetic, and talkative manner. Similarly, the ratings showed high passive engagement by the participant (4.0 ± 0.003) and confederate (3.9 ± .01). Independent-samples t-tests showed no significant differences in DU and SS participants on these interaction ratings (active; t(104) = 0.02, p = .98, passive; t(104) = 1.40, p = .17).

Hypothesis 1: Desire for Cigarettes and ENDS Will Increase Following Remote ENDS Cue Exposure

Cigarette Desire

The stage 1 model for cigarette desire revealed a significant main effect of ENDS cue (B = 4.91, SE = 1.35, t(210) = 3.64, p < .001, d = 0.09) such that the remote ENDS cue, but not the water cue, increased cigarette desire relative to baseline cigarette desire levels. The conditional R² (0.079) indicated that the model predicted a high level of variance in cigarette desire.

ENDS Desire

Similarly, the stage 1 model for ENDS desire revealed a significant main effect of ENDS cue (B = 3.98, SE = 1.43, t[188.01] = 2.79, p < .01) such that the remote ENDS cue, but not the water cue, increased ENDS desire relative to baseline. Figure 1 shows change scores from baseline to post-water and post-ENDS cue in response to the remote cue. Raw means for cigarette and ENDS desire by group and timepoint can be found in Table 2.

Figure 1.

Cigarette and electronic nicotine delivery systems (ENDS) desire change scores in response to a remote ENDS cue. Change scores from baseline to post-water and post-ENDS cue for the whole sample are presented. The remote ENDS cue significantly increased desire for a cigarette and ENDS.

Table 2.

Raw Means for Cigarette and ENDS Desire by SS and DU

	SS	DU
Cigarette desire
Baseline (t0)	63.2 (3.8)	61.5 (4.7)
Post-Water Cue (t1)	65.8 (4.2)	61.9 (4.7)
Post-ENDS Cue (t2)	69.6 (4.4)	63.2 (4.5)
ENDS Desire
Baseline (t0)	6.2 (1.3)	51.1 (3.9)
Post-Water Cue (t1)	6.1 (7.4)	53.4 (4.1)
Post-ENDS Cue (t2)	7.4 (2.1)	54.7 (4.1)

Abbreviations: DU = dual users; ENDS = Electronic nicotine delivery systems; SS = sole smokers.

Values are mean (SEM).

Hypothesis 2: The ENDS Cue Will Increase Desire for Cigarettes and ENDS in DU Compared With SS

Cigarette Desire

The stage 2 model for cigarette desire, which included a smoking group by cue type interaction term, did not explain additional variance when compared to the stage 1 cigarette desire model, and did not reveal any additional significant main effects or interactions.

ENDS Desire

Likelihood ratio test indicated a significant improvement in model fit for the stage 2 ENDS desire model versus the stage 1 model (χ²[3] = 112.71, p < .001), suggesting that including the group × cue type interaction term improved the ability of the stage 1 model to explain the observed variance in ENDS desire.

The stage 2 model for ENDS desire did not reveal a significant interaction between smoking group and cue type, but showed a significant main effect of smoking group (B = -45.60, SE = 4.0, t[148.22] = −11.42, p < .001, d = 0.20), such that DU displayed higher overall desire for ENDS than SS. Additionally, there was a significant main effect of ENDS cue (B = 6.74, SE = 2.15, t[192.09] = 3.13, p < .01, d = 0.11) such that the ENDS cue increased desire for ENDS generally. The conditional R² (0.900) indicated that the model predicted a high level of variance in ENDS desire. The stage 1 cigarette desire and stage 2 ENDS desire models are presented in Table 3.

Table 3.

Multilevel Models Examining the Effect of Smoking Status on Cigarette and ENDS Desire

	Cigarette desire				ENDS desire
Predictors	Estimates	CI	p	Estimated d	Estimates	CI	p	Estimated d
(Intercept)	59.93	54.55–65.30	<.001	—	48.74	42.01–55.48	<.001	—
Water Cue	1.28	−1.37–3.93	.342	0.10	2.40	−1.89–6.69	.272	0.11
ENDS Cue	4.91	2.25–7.56	<.001	0.09	6.74	2.50–10.97	.002	0.11
Sex	4.54	−2.48–11.57	.204	0.18	4.12	−2.79–11.04	.242	0.18
Cigarettes/day	7.15	3.64–10.66	<.001	0.09	3.01	−0.51–6.54	.093	0.09
Smoking Status	—	—	—	—	−45.60	-53.46 to −37.74	<.001	0.20
Smoking Status × Water Cue	—	—	—	—	−2.51	−8.18–3.16	.384	0.15
Smoking Status × ENDS Cue	—	—	—	—	−5.55	−11.17–0.07	.053	0.14
		Random Effects
σ2	96.23				95.22
τ00	303.47 SubId				286.76 SubId
ICC	0.76				0.75
N	106 SubId				106 SubId
Observations	318				294
Marginal R2 / Conditional R2	0.133 / 0.791				0.599 / 0.900

Abbreviations: CI = confidence interval; ENDS = electronic nicotine delivery systems.

Multilevel models demonstrate the effects of cue and person-level predictors on desire for cigarettes and ENDS. Best fitted models (stage 1 cigarette model and stage 2 ENDS model) are presented. Models are fit with random intercept for participant, denoted by SubId in the table. Estimated Cohen’s d are included as a measure of effect size for each predictor.

Discussion

This study provides the first evidence, to our knowledge, of the validity and feasibility of a fully remote, confederate-delivered ENDS cue reactivity paradigm. Consistent with our first hypothesis, the ENDS cue, but not water cue, significantly increased ratings of desire for both cigarettes and ENDS in the overall sample. As such, the present results demonstrate that exposure to vaping behavior, delivered in remote format, generalizes as a smoking cue in adults who smoke. This finding extends prior research showing a similar effect following exposure to direct, in-person cues.^12,13,15 While the group by cue interaction term was not significant (p = .053) and thus our second hypothesis that ENDS cue would increase desire for cigarettes and ENDS in DU versus SS was not supported, including that term in the model improved overall fit. Prior in-person laboratory studies showed that ENDS cue exposure evokes heightened ENDS desire in DU,¹⁶ therefore, it is possible the virtual cue exposure produces somewhat attenuated reactivity.

Consistent with emerging literature using remote methods to deploy other cue paradigms, including videos, ecological momentary assessment,³⁷ and combined in-person and imaginal cues,^31,38 these results demonstrate both the feasibility (successful enrollment, cue effects consistent with in-person work) and acceptability (positive ratings of social interactions) of measuring cue response in participants’ home environments. While there are challenges inherent in the remote methodology, including connectivity issues, detecting fraudulent participants, and ensuring data integrity and participant eligibility, protocol modifications (eg, ID verification, redundancy of eligibility items across several measures) can be successfully implemented to mitigate these issues. Importantly, such methods allow for enhanced reach of a diverse national sample. This can permit researchers to compare effects in varied and specific subgroups, such as people who vape both nicotine and cannabis or individuals with psychiatric comorbidities, that may be otherwise difficult to recruit from a single catchment area. This reach allowed the current sample to extend prior work and comprise ages beyond the typical young adult ENDS user and for diversity across racial backgrounds and geographic locations. Also, the deployment of confederate-delivered cues in participants’ natural environments enhances ecological validity as cues may be more salient and typical of the type of exposures people are likely to experience in their daily lives. Lastly, the effect sizes for the ENDS cue (cigarette desire, d = 0.09; ENDS desire, d = 0.11) and smoking group (ENDS desire, d = 0.20) were small and indicated limited individual contributions to variance in cigarette and ENDS desire. This suggests other unmeasured factors may play a role and future research could investigate additional predictors to refine our understanding. However, the high conditional R² values for the cigarette desire model (0.791) and ENDS desire model (0.900) indicated that the overall models explained a substantial portion of the variance.

The current investigation had numerous strengths, including the translation of a well-established confederate-delivered cue paradigm to the remote format and the use of validated craving measures for repeated assessments. This paradigm also enhanced the ecological validity of the cue exposures by delivering cues and measuring responses within participants’ natural environments. In addition, considerable steps were taken to ensure the guise of task randomization, including the use of a broad study description (ie, no emphasis on smoking or vaping) and specific confederate and research assistant scripts, to reduce potential demand characteristics. Still, the findings should be interpreted with some limitations in mind. First, given the remote nature of the study, it was not possible to control cue procedures to the same extent as in the laboratory. We attempted to mitigate this by training confederates and research assistants on managing connectivity issues and environmental interruptions. Also, all interactions were recorded and rated for fidelity to the cue-delivery protocol, with high fidelity across sessions. Still, it is possible that the remote delivery contributed to attenuated cue effects when compared with prior in-laboratory work.¹⁶ Second, cues were presented in a fixed order, with the water cue always preceding the ENDS cue to control for carryover effects.^39,40 This procedure is common in cue research and similar effects have been found in studies using between-subjects designs,^10,41 but this does not preclude the possibility that cue responses could have been influenced by temporal effects. Third, participants were asked to display items relevant for all “possible study tasks,” including nicotine/tobacco products, to maintain the guise that they may be randomized to use those products. It is possible that this brief engagement with their cigarettes/e-cigarettes resulted in higher baseline craving than we have observed in prior studies. However, this is the first study in which we have examined individuals in their natural environments and these higher baseline craving scores may represent true craving levels that are seen in participants’ natural environments. Further research is warranted to understand how baseline craving for cigarettes and ENDS may be different in natural versus laboratory environments.

In sum, this study is the first to translate an established in-laboratory, confederate cue paradigm to a fully remote methodology. Findings demonstrated that remote ENDS exposure acts as a smoking and vaping cue in adults who smoke cigarettes and dual-use cigarettes and ENDS. The potential for heightened reactivity among those who smoke/dually smoke and vape nicotine highlights the importance of considering how the frequency of ENDS exposures both in daily living and online/via social media may serve to maintain dual and poly-product use behaviors, rendering it more difficult to quit or reduce use of these products. Specifically, the form and function similarities between ENDS and cigarettes may render complete switching from cigarettes to ENDS difficult for some and may partially explain high rates of dual-product use.^2,3 Regulatory policy on ENDS should consider that their continued proliferation may have unique impacts on groups most susceptible to substance-relevant cues, including those who use multiple substances.^42,43 Future research is warranted to enhance our understanding of the motivational processes underlying desire for ENDS and cigarettes among users of both products and those seeking to quit smoking or all products. Additionally, as remote studies are now commonplace, others may also make use of remote cue paradigms as this is a promising approach for studying cue reactivity beyond local and regional samples.

Funding

This research was supported by the National Institute on Drug Abuse (R01-DA044210). Salary support was provided by the National Institute on Drug Abuse (K99-DA054260 to EIB) and National Institute on Alcohol Abuse and Alcoholism (T32-AA026577 to KM).

Declaration of Interests

ACK reported receiving grants from Pfizer for previous studies and consulting with the Respiratory Health Association outside the submitted work.

Author Contributions

Krista Miloslavich (Conceptualization [equal], Formal analysis [lead], Methodology [equal], Project administration [equal], Writing—original draft [lead], Writing—review & editing [lead]), Emma Brett (Conceptualization [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal], Project administration [equal], Supervision [equal], Writing—original draft [equal], Writing—review & editing [equal]), Daniel Fridberg (Conceptualization [equal], Funding acquisition [equal], Investigation [equal], Methodology [equal]), and Andrea King (Conceptualization [equal], Funding acquisition [lead], Investigation [lead], Methodology [lead], Resources [lead], Supervision [lead], Visualization [equal], Writing—review & editing [equal])

Data Availability

The data sets used and analyzed during the current study are available from the corresponding author on reasonable request.