Abstract
Objectives:
The cue-reactivity paradigm has been widely used to assess craving among cigarette smokers. Seeking to replicate and expand on previous virtual reality (VR) nicotine cue-reactivity research on nontreatment-seeking smokers, the current study compared subjective reports of craving for cigarettes when exposed to smoking (proximal and contextual) and neutral cues using VR in treatment-seeking and nontreatment-seeking cigarette smokers.
Methods:
Data from two previously published studies in nontreatment seekers from our group (Bordnick et al., 2004; Bordnick, Graap, Copp, Brooks, & Ferrer, 2005) were compared to results with 82 newly enrolled treatment-seeking smokers.
Results:
Overall, VR cues produced similar levels of craving for both treatment seekers and nontreatment seekers across the different cue environments. Specifically, craving was greater for both groups in smoking environments (paraphernalia and party) than those in the neutral environments.
Conclusions:
These findings provide strong evidence that VR is a useful tool that may be used by social workers and other clinical professionals to present smoking (proximal and contextual) cues for assessment and treatment and illustrate the utility of standardizing VR procedures to compare craving between different groups.
Introduction
The most common form of chemical dependence in the United States is nicotine (American Society of Addiction Medicine, 2010), with nearly 1 in 5 (45.3 million) American adults smoking (Centers for Disease Control and Prevention, 2011). Smoking kills nearly half a million Americans annually and is responsible for one out of every five deaths (Centers for Disease Control and Prevention, 2008). This is greater than all the deaths caused by HIV, illegal drug use, alcohol use, motor vehicle injuries, suicides, and murders combined (Centers for Disease Control and Prevention, 2008; Mokdad, Marks, Stroup, & Gerberding, 2004). Cigarette smoking is also the leading preventable cause of morbidity and is estimated to cost the United States nearly $100 billion in lost productivity and an additional $100 billion in health care expenditures annually (Centers for Disease Control and Prevention, 2008). Of the millions of smokers who try to quit each year, most relapse after a week, and only 3–5% are abstinent at the end of a year (Hughes, Keely, & Naud, 2004). Pharmacotherapies increase quit rates, but the majority still relapse (Fiore, Jaén , Baker, et al., 2008).
One factor that may be important in precipitating relapse is craving for cigarettes. Craving plays a motivational role in every major theory of drug dependence (Drummond, 2001), with some suggesting a direct relationship with relapse (Marlatt & Gordon, 1985; Robinson & Berridge, 1993; Siegel, 1989). Two common types of craving are general background craving and cue reactivity. General craving refers to relatively steady and tonic states of craving that usually last hours or days. In contrast, cue reactivity, the focus of the current study, refers to episodic and intense spikes in craving when exposed to smoking cues (Ferguson & Shiffman, 2009; Perkins, 2009).Cigarette smoking occurs in a rich stimulus environment that includes both proximal cues and contextual cues. These stimulus cues can include the sight and smell of cigarettes as well as environments and emotional states. Through repeated pairings with smoking, proximal, contextual individually or both combined (complex cues) become potent conditioned reinforcers and gain evocative properties. Cigarette smokers react to smoking-related stimuli with increased craving and physiological arousal (Carter & Tiffany, 1999).
There is evidence that cue reactivity is an important target in treatments for smoking. In naturalistic studies, smokers often point to irresistible cravings following exposure to smoking cues as a cause of relapse (Bliss, Garvey, Heinold, & Hitchcock, 1989; Shiffman, 1982; Shiffman, Paty, Gnys, Kassel, & Hickcox, 1996). Similarly, in vivo studies using portable electronic diaries report a reliable temporally proximal relationship between recent exposure to smoking cues and relapse (Shiffman et al., 1996, 1997). However, results have not been consistent, leading some researchers to state that cue reactivity predicts relapse (Ferguson & Shiffman, 2009), whereas others have stated that little evidence supports such a relationship (Perkins, 2009). Additionally, some findings in the field are counterintuitive from a Pavlovian perspective. For example, one study observed that treatment-seeking smokers exhibiting greater cue reactivity were less likely to relapse (Powell, Dawkins, West, & Pickering, 2011), and another report observed increased cue reactivity with longer durations of smoking abstinence among nontreatment-seeking smokers who were paid to remain abstinent (Bedi et al., 2011). Such results underscore the need for additional research, particularly among treatment-seeking smokers.
A promising and powerful tool for cue-reactivity research is the use of virtual reality (VR) cues. Presenting cues in the laboratory that more closely represents the smoker’s natural environment is an important goal that will allow researchers to more accurately understand cue reactivity and increase the likelihood that treatment effects observed in the laboratory will generalize to the natural environment (Bordnick et al., 2008; Monti & MacKillop, 2007). VR enables nicotine researchers to present photo-realistic proximal, contextual, and complex stimulus cues that closely mimic the real-world scenarios. Smokers are immersed in interactive virtual worlds that target visual, auditory, olfactory, and tactile senses associated with a real experience (Bordnick et al., 2009). In support of VR cues, Bordnick and colleagues observed 2- to 3-fold increased levels of craving in VR environments with smoking paraphernalia as well as a social party setting relative to a neutral VR environment (Bordnick, Graap, et al., 2005), which they replicated in young adult smokers (Traylor, Bordnick, & Carter, 2008). A second potential benefit of VR is the promise of treatments that may be presented in an individual’s home. As the cost of technology decreases, such scenarios are becoming more and more realistic. For example, an individual may provide himself or herself with a booster session prior to attending a party where he or she knows smoking is likely to occur.
To our knowledge, no studies have compared cue-induced craving in treatment-seeking versus nontreatment-seeking smokers using VR cues. For example, across drug and alcohol studies, samples of nontreatment-seeking users have shown greater cue reactivity than samples of users who are actively seeking or undergoing treatment (Wertz & Sayette, 2001; Wilson, Sayette, & Fiez, 2004). Therefore, more research is needed, which builds empirically on previous work in order to gain clearer understanding as to how VR smoking cues actually affect smokers as a critical step in the process of developing effective VR-based cue-exposure treatments to prevent smoking relapse.
The purpose of the current study was to compare subjective reports of craving for cigarettes when exposed to smoking and neutral cues using VR in treatment-seeking and nontreatment-seeking cigarette smokers. We have combined data from two previously published studies on nontreatment seekers from our group (Bordnick et al., 2004; Bordnick, Graap, et al., 2005) and compared those results with newly enrolled treatment-seeking smokers from the current study.
Method
Participants
Eighty-two treatment-seeking cigarette smokers were recruited through local newspaper advertisements. Inclusion criteria included having a Diagnostic and Statistical Manual of Mental Disorders, Fourth Edition, Text Revision (DSM-IV-TR; American Psychiatric Association, 2000) diagnosis of nicotine dependence; desire for smoking treatment; smoked at least 10 cigarettes per day for the past 2 years; and good physical health. Exclusion criteria included having a current DSM-IV-TR (American Psychiatric Association, 2000) psychiatric diagnosis of chronic, severe mental illness (e.g., schizophrenia, bipolar disorder, depression with psychosis, and schizoaffective disorder), or substance abuse other than nicotine dependence; treated with any smoking cessation medications or medications that have a potential effect on nicotine craving, consumption, and related behaviors or mood; history of serious medical conditions; fear of closed spaces or unable to wear the VR helmet; and visual problems that may have an impact on the ability to view VR materials. Regarding nontreatment-seeking smokers, data were obtained from two previously reported studies from our group (Bordnick et al., 2004; Bordnick, Graap, et al., 2005). The procedures used in those studies that have been previously reported closely match those used for treatment seekers in the current study. The current study treatment-seeking smokers and the nontreament-seeking smokers were convenience samples recruited in the same city using similar methods (e.g., flyers and print advertisement).
Design and Procedures
All participants provided informed consent during which they were provided with study information, rationale, potential risks, and benefits. Following informed consent, a member of the research team administered paper-and-pencil questionnaires and assessment instruments to assess the participant’s study eligibility. After completion of all intake assessments, eligible participants underwent a 15-min VR acclimation session to help provide familiarity with the VR experience and assessment scales and study procedures. The VR acclimation environment was unrelated to the study environments and consisted of an office interview setting. A 10-min break was instituted after VR acclimation in the lab.
Participants were seated in a comfortable, nonreclining chair placed upon a vibration platform (simulating bass and tactile sensations of a party). Participants were asked to put on a VR head-mounted display and tracker (eMagin, Hopewell Junction, NY) and make adjustments for comfort. Participants held a gamepad controller (P3000, Saitek, Torrance, CA) in their dominant hand to allow for responses on the visual analog scales at the end of each cue room. Instructions for the VR cue trial were provided from a standard script to each participant. The ambient lighting was then turned off and the VR experimental path, including participant’s preferred cigarette type that appeared in the VR smoking environments, was preselected through a graphical start-up interface. Participants then relaxed for 5 min, with instrumental (e.g., classical) music playing while the screens were dark, until the VR session began.
VR sessions began and ended with presentation of two neutral cue environments (Neutral 1 and Neutral 2). In between neutral cue presentations, participants were presented with a party room and paraphernalia room. Each environment lasted for 3 min, totaling 12 min of exposure. After each cue room, participants reported their craving ratings, via hand controller, which were projected into the VR environment. After completion of the final VR-session ratings, participants were debriefed about their VR experience.
Study Measures
Smoking History
A self-report of smoking behavior was used to gather information regarding years of smoking and the use levels.
Structured Clinical Interview for the DSM-IV-TR
The DSM-IV-TR (American Psychiatric Association, 2000) interview was used to assess participants for nicotine-dependence diagnosis and to exclude those with major psychiatric diagnoses.
Cigarette Craving Visual Analog Scale (CCVAS)
The CCVAS served as our primary dependent measure and was used to measure cigarette craving 4 times during VR presentation of cues (described below). Participants were asked to rate “their greatest craving for smoking” “at this time” by selecting a position along a line anchored on the left by “none” and on the right by “more than ever.” The CCVAS was projected into the VR environment and participants responded using a computer gamepad. The CCVAS has been used to measure nicotine craving in our previous VR cue-reactivity studies (Bordnick, Graap, et al., 2005; Bordnick, Traylor, Graap, Copp, & Brooks, 2005; Traylor et al., 2008). The CCVAS was administered following exposure to each of the cue conditions (Neutral 1, paraphernalia, party, and Neutral 2).
VR Session
All the participants were exposed to 3 VR environments (cue rooms) during the VR session for 3 min each. During each VR cue room, participants were guided through the environments on a 3-min timed path. After each cue room, participants completed a craving measure using a gamepad that was projected into the VR environment on a white background. Note that due to improvements in computer technology, VR sessions for treatment seekers had improved graphics and effects relative to nontreatment seekers. However, the relative sophistication of the graphics in the VR environments for the nontreatment and treatment seeking studies were both cutting edge for their respective time periods. Furthermore, the VR cues in both studies contained both proximal smoking cues and contextual smoking cues that were depicted with photo-realistic textures/graphics, minimizing the difference in graphics and effects. The types of environments (Neutral 1, paraphernalia, party, and Neutral 2) were similar across studies. The neutral cue environment was presented first and last, with the paraphernalia and party environments presented in between the neutral cue environments. The order of the paraphernalia and party environments was randomized across subjects.
The neutral cue environment consisted of a room, without the presence of smoking cues, in which participants could look around. The room had two screens on each side, which displayed nature documentary videos of buffalo and flamingos. The videos were approximately of 1.5 min durations each. The same neutral cue environment was used for both neutral cue presentations.
The smoking paraphernalia environment consisted of a room in a house, including a home bar area and a living room scenario (contextual cues) containing furniture, with various smoking (proximal) cues on tables including packs of cigarettes of user’s preferred brand, alcoholic beverages, coffee, ashtrays, and burning cigarettes. This room contained no people or social interactions.
The party environment (complex cues) consisted of an indoor and outdoor party setting in a 1-story house. In the indoor party, participants were exposed to people drinking, eating, and talking, with alcoholic drinks and smoking proximal cues (cigarettes and cigarette packs) on tables. The outdoor party area consisted of a patio with people smoking cigarettes and drinking alcohol and the participant was offered a cigarette.
For treatment seekers, olfactory cues were presented using the Scent PaletteTM (Envirodine Studios, Inc. Canton, GA) system. The Scent Palette system emits scents at triggered time points in the VR environments. In the VR neutral cue environments, olfactory stimuli included lavender scent that served as a control stimulus. In the smoking paraphernalia room, cigarette smoke, raw tobacco, beer, and coffee scents were presented, and in the party room, cigarette smoke, beer, pizza, and popcorn scents were presented. Olfactory cues were not used in the earlier studies in nontreatment seekers.
Statistical Analyses
Data were analyzed using StatView 5.0 (SAS Institute Inc., Cary, NC). Descriptive statistics were compiled for demographic and smoking variables and parametric and nonparametric tests were used as appropriate. A one-way analysis of variance (ANOVA) was conducted with craving score means from the four VR environments as the within-subject factor. A one-way ANOVA was also conducted on mean craving scores between treatment-seeking and nontreatment-seeking smokers for each of the VR environments. All tests were two tailed, and statistical significance was set at p < .05. All data are presented as mean ± 1 standard deviation.
Results
Participants
Among treatment seekers, participants were 85% African American (n = 70), 12% Caucasian (n = 10), 1% Hispanic (n = 1), and 1% other (n = 1). Mean age was 45.5 ± 9.3 years and 48% were female. Participants reported smoking an average of 23.8 ± 7.37 cigarettes per day. Participants reported 2.9 ± 3.2 past quit attempts. For commonly available participant characteristics in the data set of nontreatment seekers, mean age was 32.7 ± 12.6 years, 61% were female, and participants reported smoking 25.4 ± 7.0 cigarettes per day. Comparison of participant characteristics between treatment seekers and nontreatment seekers showed no significant differences in gender (z = .89, p = .37) and smoking rates, F(1, 103) = .81, p = .37, but age was significantly greater in the treatment-seeking group, F(1, 103) = 28.6, p < .001.
Craving
Figure 1 shows mean craving levels for treatment seekers and nontreatment seekers across the four cue environments. Comparing parallel cue environments between treatment seekers and nontreatment seekers, one-way ANOVA revealed significantly lower craving in the Neutral 1 environment in nontreatment seekers compared to treatment seekers, F(1, 103) = 4.92, p = .03, Cohen’s d = .60. No differences in craving were observed between the two groups in the paraphernalia environment, F(1, 103) = .42, p = .52, d = .17, and party environment, F(1, 103) = 1.42, p = .24, d = .30, and these differences were no longer present in the Neutral 2 environment, F(1, 103) = .008, p = .93, d = .02.
Self-reported craving for cigarettes (0–100) following exposure to the Neutral 1, paraphernalia, party, and Neutral 2 VR cue environments for treatment-seeking (gray bars) and nontreatment-seeking cigarette smokers (black bars). Error bars represent +1 SEM. *p = .03 (Neutral 1—treatment seekers vs. nontreatment seekers). SEM = standard error of the mean; VR = virtual reality.
Focusing on craving among treatment seekers, one-way ANOVA revealed significantly greater craving in the paraphernalia, F(1, 162) = 13.9, p < .001, d = .58, and party, F(1, 162) = 23.62, p < .001, d = .76, environments compared to the Neutral 1 environment (Figure 1). No significant difference was observed between the paraphernalia and party environments, F(1, 162) = 1.16, p = .28, d = .17, nor between the Neutral 1 and Neutral 2 environments, F(1, 162) = .02, p = .88, d = .02. Among nontreatment seekers, one-way ANOVA revealed craving was similarly greater in the paraphernalia, F(1, 44) = 24.30, p < .001, d = 1.46, and party, F(1, 44) = 23.17, p < .001, d = 1.42, environments compared to the Neutral 1 environment. No significant difference was observed between the paraphernalia and party environments, F(1, 44) = .05, p = .82. However, craving was significantly lower in the Neutral 1 compared to Neutral 2 environment, F(1, 44) = 10.84, p = .002, d = .97.
Conclusion
The primary purpose of the current study was to compare craving in treatment-seeking and nontreatment-seeking cigarette smokers exposed to neutral and smoking-related (proximal and contextual) cues using VR. Overall, VR cues produced remarkably similar levels of craving for both treatment seekers and nontreatment seekers across the different cue environments. Additionally, craving was greater for both groups in both smoking environments (paraphernalia and party) compared to the neutral environments. Increased craving has been previously observed among nontreatment-seeking smokers exposed to both VR cues (Baumann & Sayette, 2006; Bordnick et al., 2004; Bordnick, Graap, et al., 2005; Bordnick, Traylor, et al., 2005; Lee et al., 2003; Traylor et al., 2008) and non-VR cues (Carter et al., 2006; Carter & Tiffany, 2001; Conklin, Robin, Perkins, Salkeld, & McClernon, 2008; Conklin & Tiffany, 2001; Drobes & Tiffany, 1997; LaRowe, Saladin, Carpenter, & Upadhyaya, 2007; Tiffany, Cox, & Elash, 2000). These results extend previous findings in which VR successfully increased craving (Baumann & Sayette, 2006; Bordnick et al., 2004; Bordnick, Graap, et al., 2005; Bordnick, Traylor, et al., 2005; Lee et al., 2003; Traylor et al., 2008) in a new population compared to the treatment-seeking cigarette smokers.
The consistency in the magnitude of craving observed in treatment and nontreatment seekers speaks to one of the strengths of VR, which is the ability to present proximal, contextual, and complex cues in a standardized format. The need for standardized cue presentation methods across studies has been suggested by some researchers (Carter et al., 2006; Stritzke, Breiner, Curtin, & Lang, 2004). Prior studies have employed pictures (Carter et al., 2006; Conklin et al., 2008) imagery (Drobes & Tiffany, 1997; Tiffany et al., 2000), scripts (Conklin & Tiffany, 2001) as well as VR (Baumann & Sayette, 2006; Bordnick et al., 2004; Bordnick, Graap, et al., 2005; Bordnick, Traylor, et al., 2005; Lee et al., 2003; Traylor et al., 2008), and this may lead to a lack of consistency from one study to the next. For example, one lab may use a photo of a young smoker and another lab may use a photo of cigarettes, without a person smoking. Although both labs can claim to present pictures of cigarettes and use similar procedures of cue exposure, there is a lack of consistency or standard of smoking cues and stimuli presented, which may limit reliability. VR allows for researchers to standardize exposure methods and types of cues presented across studies and laboratories as done in the current report. VR provides a consistent framework across studies, thus limiting the variation in methods across laboratories, allowing for cue studies and findings to be compared equally.
Another strength of the current study is the relatively large sample size (n = 82), which to our knowledge is the largest study using VR-based cue reactivity. Previous studies using VR approaches have included sample sizes of 1 (Bordnick, Traylor, et al., 2005), 10 (Bordnick et al., 2004), 13 (Bordnick, Graap, et al., 2005), 20 (Baumann & Sayette, 2006; Traylor et al., 2008), and 22 (Lee et al., 2003). Traditional cue-reactivity studies using picture and in vivo methods have sample sizes ranging from 19 (LaRowe et al., 2007) to 296 (Shiffman et al., 2003). A number of studies (Carter et al., 2006; Conklin et al., 2008; Conklin & Tiffany, 2001) have sample sizes of 66, 62, and 60, respectively, which more closely resemble the current studies’ sample size of 82. Larger sample sizes and increased power allow for greater ability to detect differences between types of smoking cues (contextual, proximal, and complex) and their relative impact on craving. However, similar results and effect sizes are consistently found in VR cue studies using smaller samples, indicating that VR-based cue exposure is capable of large effects with smaller samples.
Results in the current study should be considered with the following limitations. First, craving was measured using subjective single-item response on a scale of 0–100. While this is a common measure in drug craving studies for nicotine, alcohol, cocaine, and marijuana (Bordnick et al., 2008, 2009; Reid, Mickalian, Delucchi, & Berger, 1999; Traylor et al., 2008), craving may be multidimensional and require a more elaborate measure beyond simple craving level (Carter, Bordnick, Traylor, Day, & Paris, 2008; Tiffany, Carter, & Singleton, 2000).
Second, due to improvements in available technology, treatment seekers were exposed to VR cues with olfactory cues compared to nontreatment seekers. Differences between the olfactory cues were not compared between methods, so the effects of the addition of scent cannot be determined. Therefore, the differences in VR cues may have erased such differences in the current study and made the groups’ craving level appear more similar. However, the similar levels of craving observed in the Neutral 2 condition between treatment seekers and nontreatment seekers suggest that the two groups responded similarly to the VR cues. Additionally, both groups reported similar levels of smoking, providing further support to the validity of the observed craving levels.
Third, craving observed in the VR environment may not represent craving levels experienced in the natural environment. The current study did not compare VR craving levels to craving levels experienced outside the lab (natural environment). Future work will focus on the assessment of craving across settings from VR to the natural environment. This comparison will determine the validly of VR as a tool for eliciting “real-world” craving responses.
Implications for Social Work Practice
Over the past 8 years, VR has been successfully used to assess cue-evoked craving in drug and alcohol studies. The lack of widespread use of VR has often been linked to the cost of equipment. As with any technology, costs decrease as a function of adoption and use. Currently, VR systems can be purchased for approximately US$3,000, putting this technology in reach of social work researchers and clinicians.
VR systems are currently used by mental health practitioners for assessment and treatment of anxiety disorders, addictions, obesity, and other behavioral health conditions. Social work clinicians provide the majority of treatment for alcohol and drug dependence. Social workers can capitalize on the use of VR to stay at the forefront of emerging technologies in mental health. Furthermore, using VR as a standardized platform for assessing cue reactivity will allow social workers and other human service professionals to explore the role of craving in relapse during clinical sessions aimed at decreasing smoking.
In conclusion, the results of the current study provide strong evidence that VR is a useful tool for presenting smoking proximal, contextual, and complex cues. The current results illustrate the utility in standardizing VR procedures in order to facilitate comparisons in craving between different groups. Future studies should focus on comparing the types of (proximal vs. contextual) cues and the impact of olfactory stimuli on craving and relapse. This study provides further support for the use of VR cue exposure clinically in the assessment of craving and relapse. Finally, the current study represents an important step in developing VR to assess craving for cigarettes in treatment-seeking cigarette smokers.
Footnotes
Declaration of Conflicting Interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was supported by a National Institute on Drug Abuse (NIDA) Grant#5R42DA016085.