Abstract
Cue reactivity to Electronic Nicotine Delivery Systems (ENDS) has been studied by several researchers, yet the variability in user types (smokers, former smokers, dual users, exclusive ENDS users) and ENDS designs used between the studies may have undermined consistent results. This systematic review aims to give an overview of ENDS cue reactivity and how smoking status and device design may moderate this. A systematic search of Medline, Embase, Web of Science, PubMed and Cochrane was completed. All studies which reported findings on reactivity to ENDS cues in the form of craving or desire for ENDS or cigarettes, attention to cue, delay of gratification or economic decisions were included. Exclusion criteria were non-human subjects, non-adult participants or participants with comorbidities. Literature selection was carried out by 2 independent reviewers. The risk of bias and study quality were assessed using tools developed by Cochrane, BMJ and NHLBI. A total of 711 papers were screened and 22 studies were included in the current review. Study design, research question(s), population of interest, number of participants, dependent variable(s), ENDS generation and nicotine content used and study results were extracted. ENDS cues reliably induced ENDS craving, with no clear moderation by smoking status and no apparent moderation by device generation. In about half of the studies, ENDS cues induced craving for conventional cigarettes. Most studies used a smoker sample, thus limiting the conclusions that can be drawn on the moderation of cue reactivity by smoking status. The quality varied among studies but comparing the findings against the outcomes of only high-quality studies did not yield any different results. The results of this review support the notion of cue reactivity to ENDS, identifies gaps in current research on different user types and implies that ENDS design iterations have little impact on cue reactivity.
Keywords
Introduction
The electronic cigarette is a handheld electronic device that was introduced in the early 2000’s as a replacement or substitute for conventional cigarettes1,2 and has become increasingly popular over the last two decades.3,4 More formally, it goes by the name of Electronic Nicotine Delivery System (ENDS) or Electronic Non-Nicotine Delivery System (ENNDS); although the distinction may be confusing as the delivery of nicotine depends on the nicotine content of the e-liquid rather than the device used. As newer devices allow the user to refill with different e-liquids, an ENNDS may become an ENDS and vice versa depending on which e-liquid is used as a refill. To avoid confusion, this paper will use the term ENDS to refer to the device, regardless of the nicotine content of the e-liquid.
Since their introduction, ENDS have gone through several design iterations and the modern devices look quite different from the original cigarette-like designs.5,6 Parallel to this transformation, the ENDS user profile has diversified into different types such as active conventional tobacco smokers attempting to quit or trying to reduce tobacco consumption (dual users), former conventional cigarette smokers who only vape and ENDS users only, who never smoke tobacco cigarettes.5,7 Due to the nicotine in the solution that is inhaled, ENDS users are susceptible to dependency, 8 yet little is known about how this dependency takes shape in the face of the variety of devices and user profiles. The current systematic literature review aims to give an overview of existing studies on reactivity to ENDS-related cues and the potential moderating role of device design and user status.
ENDs
In principle, the mechanism of all ENDS devices consists of a battery-powered heating element (the ‘atomiser’) which heats the e-liquid, a solution usually consisting of propylene glycol or glycerol (glycerine), nicotine and flavouring agents. The resulting nicotine-containing aerosol (also called vapour) is returned to the user to be inhaled.1,5,6 Initially, ENDS were promoted and marketed as a healthier alternative to tobacco cigarettes as well as to assist with quitting tobacco smoking.1,2 Although the effectiveness of vaping for smoking cessation remains unclear,9,10 the World Health Organisation has recommended that ENDS should not been marketed for this purpose 11 as scientific evidence is inconclusive 12 and their potential health effects are still to be fully understood.13,14 For many, ENDS are still less harmful than smoking and are recommended as a viable alternative for those who struggle to quit, helping them to reduce their consumption of tobacco cigarettes. 15 Nevertheless over time, ENDS use has increased rapidly alongside the marketing of vaping products through advertisements on the Internet, television and in retail outlets,16,17 especially in relation to sweet flavoured and high nicotine content e-liquids.18,19 Simultaneous to the evolving marketing of vaping, the design of ENDS has also undergone major transformations, with four generations of ENDS to date.
First generation ENDS (‘cig-a-likes’ 6 or ‘vape-sticks’ 5 ), started to be commercialised in the mid 2000’s. These devices have a fixed low-voltage battery and are designed to mimic the cigarette experience. 6 Initially, the devices consisted of three separate pieces (atomising unit, e-liquid reservoir, battery), evolving into a two-piece (combining the atomising unit and reservoir) and disposable single-piece.
Second-generation ENDS (clearomisers or tank systems5,6) have a larger battery. They are often pen-shaped, comparatively larger and can contain more e-liquid than the first generation ENDS. 5 Clearomisers are transparent and the atomiser unit is removable. 6
Third-generation ENDS are known as ‘mods’. They allow the user to modify the battery qualities (voltage, wattage, power and even charging other devices) making them easier to adjust 6 and have been marketed as ‘vapes’. 5
The fourth generation of ENDS is known as ‘pods’, and is sometimes merged with the third generation of mods. It was introduced in 2015 (e.g., JUUL 20 ). These devices have fixed batteries (meaning less control over the settings) which are generally more powerful than the first- and second-generation ENDS. They are suitable for use with nicotine salts rather than freebase nicotine, allowing users to inhale higher nicotine doses with a decreased taste aversion.21,22 Pods are very variable in design, smaller and sleeker than previous ENDS generations and are highly popular.16,22
Along with the evolution of design and marketing, the ENDS user profile has also changed, from current smokers using ENDS as an alternative to conventional cigarettes (dual users), former smokers who only use ENDS and exclusive ENDS users who have never smoked. 23 The latter tend to be younger 23 and are more likely to use flavoured e-liquid, especially fruit, mint, menthol and candy/dessert flavours. 24 This trend of young and non-smoking users has raised concerns regarding potential health risks associated to ENDS use,25,26 particularly linked to harmful substances contained in flavoured vape products,27,28 as well as due to nicotine being highly addictive 29 and how it can be detrimental to adolescent brain development (which continues until users are in their mid 20s), 30 and the association of nicotine with disturbances of cognitive development in teens and young adults. 31 In addition, concerns have been raised about the potential risk of young people who vape to be susceptible to smoke tobacco cigarettes in the future.29,32 -35
ENDS and cue reactivity
Given the first generation’s cig-a-like design and appeal to cigarette smokers, it may not be too surprising that ENDS use was initially considered a variation of conventional cigarette use. 36 With the progression into more modern devices as the newer generations were developed and their appeal to a much wider array of users, it has become clear that ENDS are their own brand of stimulant. Currently, there is no universally accepted framework of behaviours which constitute typical ENDS usage, or at what level of usage someone should be considered either dependent or a regular user.8,37
Cue reactivity is often seen as a fundamental part of addiction, 38 being considered a relevant risk factor in substance-related behaviours, 39 and it may, therefore, be the best starting point for establishing the addictive qualities of a product. The Pavlovian conditioning, or classical conditioning, states that when an individual is exposed to two paired stimuli, these can elicit responses. 40 Cues are neutral stimuli that, through repeated pairings, may become conditioned to a substance or drug leading to cue reactivity. This cue reactivity can happen when an individual interacts with environmental contexts or a stimulus, such as food advertisements, the smell of cigarettes or the sight of alcohol. Conditioned stimuli can cause different natural responses (conditioned or learned responses) such as physiological (eg, salivation, sweating), behavioural (eg, drug seeking and consumption), psychological (eg, anxiety, irritability) and a symbolic expression or a subjective motivational state (craving)41 -43 in response to external cues. 43
The term ‘craving’ has been the focus of many interpretations and often is defined as a synonym of urge or desire. For example, for the American Psychological Association (APA), craving refers to an insistent desire, urge or yearning, which is commonly used as a criteria for drugs related diagnoses. 44 For some, there are distinctions between craving and urge, whereby the former can be defined as a motivational state similar to the subjective desire for the effects of a drug and the latter indicates a behavioural intention to engage in a particular behaviour. 43 Regardless of the lack of consensus for a prevalent definition, craving and urge have been largely associated with cue exposure and cue-reactivity.38,45 -47
Furthermore, substance cues can have an effect on the level of self-reported craving and self-administration behaviours. The individual’s risk of relapse may be predicted by the degree of cue reactivity.48 -50
The classical conditional concept has been used to support several theories related to alcohol and other drugs (AOD) craving, such as that developed by Winkler, 51 which indicates that addicts going through craving could be motivated to look for AODs in order to be relieved of conditioned withdrawal effects. This model was first used to explain heroin addiction and has been applied to alcohol among other drugs. Cue reactivity has long been established in conventional smoking52 -54 as well as for the study and treatment of other disorders such as eating related disorders, 46 gaming, 55 alcoholism 56 and drug use. 57
However, whether cue reactivity prevails over the range of user profiles and ENDS generations is still under investigation. On one hand, one could argue that ENDS cues are similar enough to smoking cues for smokers to respond with craving when exposed to them: ENDS and cigarettes use roughly the same hand-to-mouth gestures; assuming that e-liquid with nicotine content is used, both deliver their nicotine through inhalation; and most (but not all) ENDS create clouds of vapour similar in appearance to cigarette smoke. On the other hand, ENDS designs can differ wildly from both cigarettes and one another. This could mean that, for example, a user of the cig-a-like will not respond with craving at all when exposed to a pod, because the latter looks so different from the device that they have been conditioned on that it does not elicit a response, particularly if it does not produce vapour clouds. Similarly, a smoker may respond only to first-generation ENDS cues with craving, as the later generations are too dissimilar.
Objectives of the current review
The current systematic literature review aims to create an overview of ENDS cue reactivity, as well as investigate the moderating role that the device’s generation and smoking status of the user (i.e., dual users – who use both ENDS and tobacco cigarettes, former smokers who only vape and users who have never smoked and only use ENDS) may play in ENDS cue reactivity. The research questions are as follows:
Do ENDS cues elicit cravings for conventional (tobacco) cigarettes, ENDS or both?
Is this response moderated by user smoking status?
Is this response moderated by ENDS generation or design?
The review is limited to studies using adult participants with no relevant comorbidities (eg, other addictions, cancer, psychological disorders). The intervention of interest is cue reactivity (be it visual, olfactory, social or other cues), with the primary outcome of interest being cue-induced craving, and the following possible secondary outcomes: delay of gratification, attention to cue, time to next ENDS or cigarette use, and money users are willing to spend to use an ENDS or cigarette. Comparisons of interest would be ENDS cue reactivity to alternative cue reactivity (e.g., tobacco smoking cue, neutral cue or comparing different generations of ENDS devices) and comparing ENDS cue reactivity depending on smoking status: dual user, former smoker who only vapes, or ENDS user who only vapes. To give a complete overview of the state of the field, experimental study designs, cross-sectional, cohort studies, case studies, randomised controlled trials, pre-post tests without a control group, longitudinal studies, as well as literature reviews were included.
Method
Protocol
A PROSPERO protocol was adapted from an earlier logged protocol (CRD42020153927), which was deemed too comprehensive. However, at the time of writing the updated protocol was still under review. The protocol can also be found at the Open Science Framework under the digital object identifier 10.17605/OSF.IO/XK9NA.
Eligibility criteria, information sources and search
The PRISMA guidelines for review58,59 were followed for the current review. Five databases (Ovid Search-MEDLINE, Ovid search-EMBASE, Web Of Science, Pub Med and Cochrane) were searched by a single researcher using the following search terms: [‘vaping’ OR ‘e-cigarette*’ OR ‘electronic cigarette*’ OR ‘electronic nicotine delivery’] AND [‘electronic cigarette dependence index’ OR ‘cue reactivity’ OR ‘cue exposure’ OR ‘cue induc*’ OR ‘vaping cue’ OR ‘reactivity to cue’ OR ‘e-cigarette cue’ OR ‘salien*’ OR ‘stimulus’ OR ‘withdrawal symptoms’] AND NOT [‘animal’ OR ‘mice’ OR ‘cannabis’ OR ‘marijuana’], in the title, keywords or abstract. No language restrictions were considered for the database searches, although all search terms were written in English only. There was no limit of the year of publication for literature searches, since relevant research related to the use of electronic cigarettes is relatively new. The reference lists of selected papers were also scanned for relevant papers.
Inclusion criteria
The inclusion criteria were as follows: any study with an adult participant population targetting cue reactivity to ENDS was included. Cue reactivity was taken as a physiological or psychological response to the cue. Measures of primary interest were reports of craving and desire, but brain responses, cognitive measures and behavioural measures were also included.
Exclusion criteria
The following excluding criteria were observed: the use of non-human subjects, the inclusion of subjects with comorbidity (psychopathological or physical), the inclusion of non-adult subjects, data that had been published in another paper included in the review (these two or more publications were to be treated as one), papers written in any language other than English, studies on vaping any drug other than nicotine or a placebo aerosol liquid, and studies that compared smoking cigarettes to vaping.
Study selection
Paper selection occurred in three stages. First, all duplicates were removed. Then, paper screening occurred based on the title and abstract. Finally, the sample of papers that was left was reviewed based on the full paper contents. During this final stage, paper accretion through identifying relevant publications in the literature which were not uncovered by the database search occurred as well. Newfound publications were also assessed for inclusion/exclusion criteria.
Reviewing took place through the Covidence interface, with two reviewers independently screening and selecting all papers to verify inclusion criteria and to confirm relevance. Group discussions were conducted when discrepancies regarding the inclusion of studies occurred.
Data extraction
The following variables were extracted from the selected studies: study design, the population of interest (age group, smoking and vaping background), number of participants, independent variables, dependent variable(s), generation and type of ENDS device used in each study (when reported), nicotine content e-liquid used (if applicable), procedure, research question(s), results, funding and reported conflicts of interest.
As part of the paper evaluation, quality control tools were used. Specifically, the Cochrane Risk of Bias tool (ROB 2.0) 60 was used for assessing the quality of randomised trials, and the Risk of Bias In Non-randomised Studies of Interventions (ROBINS-I) 61 for cohort-type studies. The PRISMA 2009 Checklist 62 was used for assessing systematic reviews and meta-analyses. Finally, the Study Quality Assessment Tools developed by the National Heart, Blood and Lung Institute (NHLBI) 63 were used to assess overall quality.
Table 1 shows a summary of the selected papers, their design, their main findings and the risk of bias.
Summary of the selected papers, their design, main findings and risk of bias.
Abbreviations: NA, not applicable; NR, not reported; RoB, risk of bias; SQ, study quality. RoB indicators – L, low; SC, some concern; H, high. Quality indicators – G, good; F, fair; P, poor.
Results
Study selection
The search was completed in October 2020 and resulted in a total of 1009 papers: Cochrane (n = 59), Ebscohost (n = 182), Embase (n = 131), MEDLINE (n = 67), Pubmed (n = 111), Scopus (n = 577). After duplicate removal, 711 papers remained. Of those, 679 were excluded based on their title or abstract, leaving 32 papers from the original search for the full-text review. During this stage, one more paper was identified through the bibliographic references resulting in a total of 33 papers for full-text review. Of those 33, two were removed for including non-adults among their participants, eight were removed because they did not include cue exposure in their design and one was removed because it was a protocol with no outcomes reported. Thus, the current systematic literature review included a total of 22 papers. See Figure 1 for the flowchart of the selection process; see Table 1 for the data extracted and assessed risk of bias per study.
Flowchart of the paper selection process.
Preliminary findings
Over half the papers (n = 12) used a between-subjects randomised design. Three papers had a mixed pre-post test design, where participants completed a baseline measure, were randomly allocated to a condition and completed a post-exposure measure; five had within-participant experimental designs; one study had a pre-post test with no control group; and one carried out a regression analysis.
Overall, the quality of the studies was high, with 14 receiving a ‘good’ grade, five receiving a ‘fair’ grade and three a ‘poor’ on the NHBLI tools. The risk of bias was low overall, with four studies receiving ‘some concerns’ and one being identified as ‘high risk’ (Table 1). In terms of funding and potential conflicts of interest, several authors reported being previously affiliated with the Food and Drug Administration (FDA) or various pharmaceutical companies; none reported affiliation with the tobacco or vaping industry. Similarly, funding was received from a number of sources outside of academic institutions such as the National Cancer Institute (NCI) or National Institute for Drug Abuse (NIDA); none from the tobacco or vaping industry, although one study 64 used vape products (devices and cartridges) provided by the manufacturer, White Super E-cigarette.
Almost all studies (n = 19) used smoking participants (either exclusively smoking or dual users) in their experiments; of those, one compared smokers to former smokers and those who had never smoked, 65 one compared smokers to dual users, those who are exclusively ENDS users and those who are neither smokers nor ENDS -users 66 ; one compared daily to intermittent ENDS users to former smokers. 67 Of the remaining three papers that did not include smokers, one included participants who had never smoked and did not currently smoke, 68 one assessed only exclusive ENDS users, 69 and one included both current and former smokers but analysed the groups separately. 70 All papers used visual cues.
In addition, 18 studies64 -67,69 -82 informed about the type (generation or commercial brand) of ENDS used in their experiments as cues (e.g., in person use or visualisation, in videos, advertisement, or printed ads) (Table 1). Moreover, ENDS were used either as a unique active cue or compared with other ENDS generations or tobacco cigarette cues. In this way, 10 studies used first generation ENDS,64 -67,69,73,75,77,78,80 five studies used second generation,67,73 -76,81 four studies used third generation65,66,70,78 and two studies reported the use of fourth generation ENDS.70,78 Furthermore, two studies reported the use of participants’ own ENDS (unspecified),69,78 and two studies reported the commercial brand but not the ENDS specifications.79,82 In regards to nicotine content, only 10 studies reported the ENDS nicotine levels used in their experiments.64,69 -72,74,77,79 -81 E-liquids nicotine content ranged from 0 to 59 mg/ml (salt). Further information about ENDS types and nicotine levels used in the reviewed studies, including related outcomes, is presented in the sub-sections below and in Table 1.
Do ENDS elicit cravings for conventional (tobacco) cigarettes ENDS or both?
The main outcome investigated in this narrative systematic review was related to cue reactivity to ENDS. To achieve this, information related to three primary associated cravings (ie, conventional tobacco cigarettes, ENDS or both) was obtained from the reviewed studies. The resulting findings are presented as follows.
Cue exposure to ENDS was found to enhance craving for conventional (tobacco) cigarettes in 10 studies,67,70 -77,82 while eight studies reviewed did not demonstrate such a relationship.65,66,68,78,79,83 -85 To ensure that this dichotomy was not the result of study quality or design, the groups were compared for whether studies had been conducted in person or online, their number of participants, year of publication, the generation of ENDS used and the quality or Risk of Bias assessment. No differences were found.
Cue exposure to ENDS induced craving for ENDS was found in 10 studies.66,68,70 -76,78 Although Nichols et al 69 investigated this, they did not find a significant result. However, this study had a relatively small sample size (n = 7) and, therefore, a low statistical power. In addition, while the results found in Dowd and Tiffany 78 provided a clear indication of increased ENDS cravings to ENDS cue exposure, results did not establish whether ENDS increased cravings for conventional (tobacco) cigarettes (although it was found that ENDS did not reduce these cravings).
Cue exposure to ENDS induced cravings (desire and/or urge) for both conventional (tobacco) cigarettes and ENDS.70 -76 Vena et al 70 and Van Heel et al 80 studied cue exposure to ENDS inducing craving reduction for both tobacco cigarettes and ENDS, and found significant results (moderated by nicotine).
Is this response moderated by user smoking status?
Smoking status can play an important role in the way that a patient responds to ENDS cue exposure, in relation to attention and attitudes. 65 In this study, smoking status was viewed in relation to both smoking conventional (tobacco) cigarettes and ENDS (dual users). Our findings in this regard are described next.
Most studies used participants who smoked tobacco, although the level of cigarette use varied between studies. Three studies specifically analysed non-smokers (either as a sample or as a subgroup of interest),65,66,69 three studies specifically targetted dual users,66,78,84 and two had exclusive ENDS users as a (sub) group.66,69
With that caveat in mind, smoking status may play a moderating role. Comparing tobacco smokers to ex-smokers and non-smokers, current smokers spent more time looking at ENDS cues than the other two groups. 65 In addition, this was the only group where ENDS cues elicited a smoking urge 67 or craving for both cigarettes and ENDS. 65 On the other hand, there is the analysis by Blackwell et al 66 who carried out an online study comparing ENDS cue reactivity in current smokers and former smokers who were either daily vapers or non-vapers and found no influence of smoking status on cue reactivity (for either tobacco cigarettes or ENDS craving), which caused the authors to raise questions regarding the validity of online cue reactivity research. Nichols et al 69 even reported no ENDS craving after exposure to ENDS cues for non-smoking ENDS users; yet as mentioned above, the sample in this study was small. Interestingly, Vena et al 70 found that ENDS cue exposure increased smoking urge and desire in current smokers, but only enhanced smoking desire (and not urge) in former smokers. However, the size of the latter group was considerably smaller (20 former smokers as opposed to 62 current), so statistical power may have been an issue.
Tan et al 84 compared the responses of exclusive tobacco smokers and dual users to vaping cues in anti-ENDS public service announcements. They found that for smokers, the ENDS cues lowered the intention to use ENDS or buy and smoke cigarettes, yet at the same time increased the urge to use ENDS or smoke (compared to no cue). Thus, the inclusion of a cue affected smokers’ rational acceptance of the advertisement’s message, but still automatically triggered the craving. No such effect was found for dual users.
Dowd and Tiffany 78 studied dual users and found that cue exposure had to be matched to the type of dependency that it aimed to elicit: cigarette cues for tobacco craving and ENDS cues for eliciting ENDS craving. Within that design, cigarette cravings were stronger than ENDS cravings. Finally, Dawkins et al 77 reported an interaction between previous ENDS use and reactivity to non-cigarette-like ENDS cues: participants who had previously used ENDS showed increased cravings for cigarettes regardless of whether the ENDS resembled a cigarette, while participants without ENDS experience only reported increased cravings after exposure to a cig-a-like ENDS.
Is this response moderated by ENDS generation or design?
ENDS engineering and design has rapidly evolved in a short time. Many changes in capacity, duration and nicotine content have occurred in the devices since the first cig-a-like ENDS was marketed. The role of ENDS generation and design as a moderator in ENDS cue reactivity was an additional matter of interest in this review, and the results are as follows.
Overall, the reviewed studies were carried out between 2012 and 2020 and most of them used for their experiments, first- and second-generation ENDS. Only two studies used exclusively third generation,71,72 and one study reported using fourth generation ENDS (JUUL) as a cue. Thus, two studies65,66 directly compared first and third generation ENDS and their respective effects on craving. No differences were found; both generations elicited craving to a similar extent. Comparing the results from studies that used exclusively first,69,73,77,80,85 second74,76,81 or third71,72 generation ENDS, the only two studies69,85 that did not find cue reactivity used first generation devices. At the same time, two other studies using first generation devices did find an effect. Moreover, Nichols et al 69 had a small sample (n = 7) and Jo et al 85 compared craving after exposure to various ENDS advertisements; the control condition contained no visual image of ENDS but did state ‘Use Evermist E-Cigs’, all other conditions added either a visual cue, a more elaborate tag line, or both. No pre-exposure measures were taken, so it was not possible to say if all advertisements induced craving to the same amount, or if none produced any craving at all. King et al examined the role of first 73 and second74,76 generation ENDS design (similar/less-similar to conventional cigarettes), finding that both generation designs increased tobacco cigarette craving. Vena et al 70 used a fourth generation ENDS as a cue but asked participants to indicate their desire for a third and a fourth generation ENDS. They found that smokers reported an increased desire for both generations of ENDS, while former smokers only reported higher desire for fourth generation ENDS (JUUL), suggesting that generalisation may coincide with cue salience.
Further observations
Secondary outcomes
Secondary outcomes are a fundamental component of all systematic reviews. In this subsection, we include secondary outcomes pre-established in our review protocol as well as all those reported in the studies reviewed.
More than half of the studies (n = 15) investigated secondary outcomes,65 -67,69-71,74,76-79,83 -86 with some studies exploring more than one. Smoking latency, which relates to the participant’s ability to abstain from smoking tobacco cigarettes (measured in time) versus obtaining money as a reward, was determined in five studies.70,71,74,78,79 In studies developed by Vena et al70,71 the majority of participants chose to smoke over the money compensation, and Dowd and Tiffany 78 reported that participants spent more money to get cigarettes than ENDS. Additional secondary outcomes investigated in selected reviewed studies are presented next, including significant results on analyses: urge to vape,66,69 urge to smoke,70,85 intention to quit smoking or remain abstinent from smoking,66,67 desire to smoke and vape, 66 self-efficacy (for smoking cessation and continued abstinence), 67 intentions to smoke or vape, intentions to buy ENDS and traditional cigarettes. 84 In this latter study, Tan et al reported that anti-vaping advertisements without vapour was associated with lower intention to buy ENDS and tobacco cigarettes in the smokers group; and in dual users, the exposure to the advertisement reduced the intention to purchase ENDS. In terms of ENDS rewards, De La Garza et al, 79 found that although a 16 mg/ml dose of nicotine though ENDS cue exposure (use) relieved withdrawal symptoms, it did not produce a rewarding feeling when compared to participants’ own tobacco cigarettes. Finally, in relation to the effect on mood, King et al 74 found that mood was negatively influenced by tobacco cigarette cue exposure. Additional related information is provided in Table 1.
Craving enhancing factors
Elements that intensify, increase or amplify the effect of a cue, can have significant influence in a person’s cue-reactivity. The ENDS industry has capitalised on this in marketing and, as such, understanding the use and effect of cue enhancing factors in patients’ cravings is paramount for future advances in ENDS cue-reactivity research. Relevant information gathered in this review in this regard is presented in this subsection.
Vena et al 72 found that artificially enhanced cues (eg, a more pronounced smoke cloud) enhanced subjective cravings (for both cigarettes and ENDS) accordingly. Kroemer et al 81 found that sweet tasting e-liquids enhanced conditioning effects, that is, resulting in higher subjective craving at subsequent cue exposure. Moreover, sweet flavours enhanced the effect of nicotine on brain responses to ENDS cues (when paired with sweet and nicotine versus paired with either sweet or nicotine). 81 Similarly, using functional Magnetic Resonance Imaging-fMRI, Garrison et al68,69 found higher activation in the nucleus accumbens, a brain area associated with reward and addiction, after exposure to stimuli of fruit-flavoured ENDS rather than tobacco flavoured ENDS.
Craving reducing factors
Here we report the findings on craving reducing factors analysed in our systematic review. In accordance with previous studies developed in tobacco cigarette craving and withdrawal symptoms, 87 in this review, nicotine was the primary craving reducing factor. Next, we describe the studies which reported findings in this area.
After ENDS cue exposure (ENDS use), four studies69,77,79,80 reported that nicotine acted as a moderating factor in self-reported craving (urge or desire) reduction.64,69,79,80 De La Garza et al 79 found that 16 mg/ml-nicotine e-liquid reduced craving for ENDS but was not seen as equally rewarding as smoking tobacco. In particular, Van Heel et al 80 replicated these findings, reporting a reduction in craving for both cigarettes and ENDS after vaping, with a stronger reduction for using a nicotine-containing e-liquid (36 mg/ml) than a nicotine-free e-liquid. Nichols et al 69 found lower self-reported craving after controlled ENDS use, which was followed by an enhanced brain response to an ENDS cue (video). In this study, the effect of ENDS use in reducing the urge for e-cigarettes was larger than the effect of ENDS cue videos. Finally, Dawkins et al 64 found that both using nicotine-containing (18 mg/ml) and nicotine-free e-liquid reduced the desire to smoke compared to no ENDS use; however, only the nicotine e-liquid subsequently reduced anxiety and irritability, and improved concentration. In addition, the same authors reported that for smoking cessation, nicotine content appeared to be more important (acting as a moderator factor) for males than for females.
Discussion
The current systematic literature review set out to aggregate and evaluate the results from existing research on cue reactivity to electronic nicotine delivery systems (ENDS). Twenty-two studies were identified and analysed. While further research is needed to fully understand whether ENDS related cues increase craving for conventional cigarettes, most studies examined in this review agreed that they do enhance craving for ENDS. It was equally difficult to tell with certainty whether any relationship between ENDS cues and craving is moderated by smoking and vaping status.
Unsurprisingly, participants who have never smoked conventional (tobacco) cigarettes nor ENDS did not show reactivity to ENDS cues; yet at the same time there were no systematic differences in reactivity between dual users, exclusive smokers, or exclusive ENDS users found across the studies. It was noticeable that two studies used fMRI to investigate the brain response to ENDS cues in exclusively ENDS users 69 and early vapers or those at risk of vaping. 68 Although the results in the former were not significant, the latter show some interesting results in regards to neural responses to sweet/fruit flavours and users’ poor memory to warning labels. The relevance of these studies lies in the elimination of the effect of the tobacco cigarette smoker over the result, which could represent a clearer way to help understanding the complexity surrounding ENDS cue reactivity in ENDS users.
Furthermore, cue reactivity appeared remarkably consistent across ENDS generations. In this regard it is important to highlight that the majority of studies examined in this systematic review used first to third generation ENDS, and only two studies mentioned the use of fourth ENDS generation in their research, with one of them focussed on the JUUL cue exposure. 70 First to third ENDS generation, generally do not hold the power and capacity that fourth generation ENDS do, including the nicotine concentration levels that these devices are designed to deliver. Further research and reviews focussed on fourth generation ENDS is needed, in order to better understand the effect that these powerful devices are having among ENDS users on nicotine dependence and their potential health implications. Taken together, these results suggest that ENDS cue reactivity generalises reasonably well across both cigarette smokers and ENDS users and device designs for ENDS craving. The use of enhanced stimuli (sweetness and enhanced smoke clouds) elicited an enhanced craving response. In addition, nicotine was the primary mediating factor for reducing craving. Although only four studies examined this, the findings align with previous reports for tobacco cigarettes. 87 The therapeutic use of ENDS for tobacco cessation and replacement is an ongoing investigation and recommending it for these purposes requires sensible considerations.
In this review, secondary outcomes were a topic of interest of more than half of the studies examined. Latency to smoking cigarettes was the behavioural outcome most reported among many others explored, and according to the latency observations, participants had more cravings for tobacco cigarettes than ENDS, which may be associated to different factors, such as ENDS nicotine content, device used, smoking or vaping status or method of visualisation. More targetted and focussed studies are needed to help elucidating reasons behind this kind of results. According to the remaining findings, the role of ENDS cues in associated secondary outcomes are still quite unclear, but some of the results found may represent areas where future research efforts could be invested, such as the manipulation of ENDS characteristics (such as vapour clouds, flavour, nicotine levels) towards reducing intentions to buy or use ENDS or conventional cigarettes and, in addition, the potential role of ENDS in helping to reduce withdrawal symptoms.
The consistency in findings regarding cue-induced cravings for ENDS is remarkable when one considers the vast differences in design that exist between the generations. The earlier cig-a-likes look very different to the more recent pods; the latter of the two often bearing more resemblance to a USB flash drive, a portable radio or a tape roller than something that would traditionally dispense nicotine. This consistency could be considered relevant, as the scientific community would be hard-pressed to keep their knowledge up-to-date with every new design that is released. Instead, this ease of generalising may imply the strength of the learned response.
Limitations
Two types of limitations can be discussed: limitations that are to do with the literature itself, and limitations of this review. Regarding the limitations from the literature, not all of the studies included were considered to be of fair or good quality; similarly, in several studies, there were concerns regarding bias. To minimise the chance of those studies biassing our current outcomes, the authors checked their results against a sample that only contained papers deemed of high quality and low risk of bias, and no differences emerged in the findings. Thus, we do not think that including the studies of lesser quality introduced any problems.
Furthermore, few studies explicitly addressed the influence of user smoking status on ENDS cue reactivity. This appears to reflect the earlier image of ENDS: an alternative nicotine provider to conventional cigarettes. However, ENDS have long expanded on their role of nicotine replacement therapy and, consequentially, little is currently known about users who vape for reasons other than limiting their cigarette use. A similar statement can be made on the influence of device design, although this can still be indirectly observed as the more recent studies employ more recent generation ENDS. Thus, it appears that ENDS design has little effect on cue reactivity, but a formal experiment to confirm this would be warranted.
Addressing the second type of limitation: a universal issue with literature reviews is publication bias.88,89 Since the current topic is quite new, intuitively we suspect that this particular topic might have a relatively low bias here as both a null result (i.e., ENDS cues do not elicit cravings) as well as a non-null result would be considered intriguing.
Conclusion
As ENDS design and its user base continue to evolve, more research will need to be devoted to how an ENDS dependency may take shape. A starting point for this will likely be cue reactivity. The variability in ENDS designs and user cultures may pose a challenge for researchers trying to find the fundamental principles of ENDS dependency, although it is encouraging to find that so far ENDS design has had limited effect on whether cue reactivity was established. Future studies will need to determine whether such a generalisation can also be made across user types and could further expand on the types of cues.
Footnotes
Acknowledgements
The authors wish to thank Dr Clara Bah who assisted in the proofreading of the manuscript.
Author Contributions
MK: Conceptualization-Supporting, Data curation-Lead, Formal analysis-Lead, Methodology-Supporting, Validation-Lead, Writing-original draft-Lead, Writing-review & editing-Lead. MCVC: Data curation-Supporting, Formal analysis-Supporting, Validation-Supporting, Writing-review & editing-Supporting. SH: Conceptualization-Lead, Funding acquisition-Equal, Methodology-Lead, Project administration-Supporting, Supervision-Equal, Writing-review & editing-Supporting. MT: Conceptualization-Equal, Funding acquisition-Equal, Methodology-Supporting, Project administration-Lead, Supervision-Equal, Writing-review & editing-Supporting.
Funding:
The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This paper was funded through a Health Research Council (HRC) of New Zealand Explorer Grant, number 18/738. The HRC had no role in the study design, collection, analysis or interpretation of the data, writing the manuscript, or the decision to submit the paper for publication
Declaration of conflicting interests:
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.