Does laboratory cue reactivity correlate with real-world craving and smoking responses to cues?
Introduction
Smoking is motivated and maintained by nicotine (Benowitz, 2008), but smoking and craving are also influenced by situational stimuli (Kozlowski and Herman, 1984). Surveys reliably link smoking and craving to cues such as presence of other smokers and drinking alcohol (McKennell, 1970, Russell et al., 1974). Similarly, studies using electronic diaries show that people were more likely to smoke when seeing others smoking and when drinking, and less likely to smoke when smoking was prohibited (Shiffman et al., 2002, Shiffman et al., 2014b). The role of situational cues is particularly prominent in relapse (O’Connell and Martin, 1987, Shiffman et al., 1996). These situational linkages are often attributed to conditioned learning due to repeated pairing of smoking with environmental stimuli (Carter and Tiffany, 1999, Niaura et al., 1988). This leads to an emphasis on individual differences, since smokers are likely to differ in smoking patterns and thus in the stimuli they associate with smoking, and might also differ in conditionability (Mineka and Oehlberg, 2008). It also suggests that if individual cue associations could be identified, this might help to individually tailor treatment (Conklin et al., 2010, Drummond, 2000).
Ecological Momentary Assessment (EMA; Shiffman et al., 2008) methods, using electronic diaries to collect detailed real-time data, are increasingly being used to study the effects of cues in smokers’ real-world settings. However, these methods, while they are considered ecologically valid, rely on smokers’ naturalistic exposures to cues in their environment. In contrast, cue reactivity (CR) studies (Carter and Tiffany, 1999, Niaura et al., 1988) attempt to bring cue responses into the laboratory, where they can be manipulated and studied experimentally. As a consequence, CR studies are commonly used to assess the effects of cues on craving and smoking (as well as other drug use; Carter and Tiffany, 1999). However, the degree to which laboratory CR responses predict real-world responses to cues has not been studied. In this paper, we assess the associations between smokers’ laboratory CR cue responses and their real-world cue responses, as assessed by EMA.
CR studies (Carter and Tiffany, 1999) expose smokers to stimuli thought to be associated with craving and/or smoking, such as cigarettes themselves, and compare these responses to those elicited by a neutral stimulus. Increases in craving or in the likelihood of smoking are seen as indicators of specific cue reactivity. Cues studied in this way have included cigarette-specific stimuli (termed ‘proximal cues’ by Conklin et al., 2008), and diverse cues such as positive or negative affect, alcohol cues, etc. (‘distal cues’; Conklin et al., 2008). CR effects have been observed both in abstinent smokers (McClernon et al., 2009, Tiffany et al., 2000) and minimally deprived smokers (Bailey et al., 2010), and reactivity seems similar across these conditions (Carter et al., 2009, Franklin et al., 2007). However, individual differences in CR are not consistently linked to smoking cessation outcomes (Perkins, 2012, Powell et al., 2010, Waters et al., 2004), causing some to question its relevance to cessation and treatment (Perkins, 2009).
A key question in the CR field is whether reactivity observed in laboratory settings actually reflects real-world responses to these cues. Such a link would be expected if CR responses reflect participants’ condition history (Hogarth et al., 2010, Niaura et al., 1998), and necessary if CR responses are to be used to predict outcome (Perkins, 2012) or to tailor treatment (Conklin, 2006, Drummond, 2000). However, to our knowledge no study has examined the degree to which CR measures predict how smokers react to stimuli in the real world. In this paper, we compare data from previously-published studies examining CR responses observed in the laboratory (Shiffman et al., 2013a) and real-world craving and smoking observed in response to analogous cues in EMA data on the same sample (Shiffman et al., 2014b).
EMA data comparing the presence of cues when subjects are and are not smoking provide estimates of the associations between stimuli and smoking (Paty et al., 1992), and can also be used to assess situational correlates of craving (Dunbar et al., 2010). These EMA data are essentially the real-world analog of laboratory cue-reactivity measures, and thus an apt reference to assess the real-world generalizability of CR assessments.
The primary effects from both the CR and EMA studies have been reported in previous publications. The CR study (Shiffman et al., 2013a) assessed daily smokers’ craving and smoking in response to a range of smoking-relevant cues. Craving was increased by exposure to smoking and alcohol cues, and decreased by positive affect; however, negative affect and smoking prohibition cues did not, on average, affect craving. As previously reported, both the probability of smoking and the amount smoked increased in proportion to craving, and in proportion to the observed increases in craving, though not in a cue-specific way (Shiffman et al., 2013a).
Similarly, the data from the EMA study also showed significant cue effects: smoking was associated with situational cues including others smoking and alcohol consumption, and inversely with smoking restrictions (Shiffman et al., 2014b). Negative affect was not associated with smoking, but positive affect showed curvilinear effects. Importantly, there was significant individual variation in response to cues, and when this was taken into account, cues, including affect, were very robust predictors of smoking. Knowing a smoker's affective state at any moment allowed one to predict whether s/he was smoking with 70% accuracy (Shiffman et al., 2015), again demonstrating the importance of individual differences even when average effects in the sample as a whole were not significant.
Our primary papers on the CR (Shiffman et al., 2013a) and EMA (Shiffman et al., 2014b) studies focused on the effects of cues in each paradigm. Here, we aim to directly compare the laboratory CR and real-world EMA responses observed to assess whether CR is an accurate proxy for real-world responses to cues. If so, one would expect the magnitude of reactivity observed for each individual during CR exposure to correlate with that individual's responses in real-world observations. For example, if a particular smoker exhibits a high degree of craving reactivity to smoking cues during the CR sessions, one would similarly expect to see that person's craving be higher when such cues are present in real-world settings.
Note that such individual associations do not rest on there being a shared, common reaction across all smokers. Thus, for example, some smokers may be “negative affect smokers” and others not, with the result that the overall effect of negative mood on smoking is not significant across the whole sample of smokers. Nevertheless, one would expect that the particular individuals who demonstrate negative affect smoking in the CR assessment would show negative affect smoking in the EMA assessment. The validity of such individual assessments even in the absence of group-wide significant effects was illustrated in a previous EMA study. In that study, there was no overall significant relationship between negative mood and smoking (Shiffman et al., 2002). However, within this overall ‘null’ effect, there were substantial individual differences in the relationship between mood and smoking, and analyses showed that the individual associations predicted which individuals were most vulnerable to relapse (Shiffman et al., 2007). Accordingly, in addition to examining the cues that had group-wise significant effects in each paradigm, we also examine the correlation of individual differences in other cues as well.
For the data to be informative on individual differences, it needs to be demonstrated that each of the measures shows substantial between-subject variation; i.e., that some respond more strongly than others, or even that some respond in one direction (e.g., smoking more when upset) while others respond in the opposite direction (smoking less when upset). The potential for correlation across methods depends on such individual differences; if nearly everyone responds in the same way to a particular cue, individual differences – and correlations – would be minimized. Thus, we precede the analysis of correlations between CR and EMA data with an examination of between-person variation in cue associations within each method.
Section snippets
Subjects
Subjects were 190 daily smokers who completed both the EMA assessment and the CR assessment (i.e., providing data on the neutral cue and one or more active cues). Volunteers had to be at least 21 years old, smoking 5–30 cigarettes per day, smoking for ≥3 years, at their current rate for ≥3 months, and not planning to quit within a month. Subjects in these analyses averaged 40.19 (±11.58) years old, on timeline follow-back reported smoking an average of 14.98 (±5.89) cigarettes per day, and had
Individual variability in response
Fig. 1 shows the distribution of associations between cues and craving and smoking for both CR and EMA. It can be seen that craving changes after each CR cue (vs. neutral) demonstrated a wide range of individual differences, including both increases and decreases in craving, compared to neutral cue response. Similar dynamics held for total puff time following each cue (compared to neutral cue response). EMA data are graphed in Fig. 1 as within-subject correlations between each cue and smoking
Discussion
For the first time, we tested whether craving and smoking observations from laboratory-based CR testing were actually related to corresponding measures of real-world craving and smoking, assessed via EMA, in the same sample. The results were highly consistent: There was no reliable relationship between CR measures and real-world behavior. The average correlation between the two was 0.03, and there was no robust evidence that the correlations were reliably greater than zero. (Two correlations
Role of funding source
This work was supported by grant R01-DA020742 (PI: Shiffman) from the National Institutes of Health, National Institute on Drug Abuse. Additional support was provided by National Science Foundation Graduate Research Fellowship (Dunbar), National Center for Research Resources (KL2-RR024154-03; Tindle), and National Cancer Institute R25-CA057703-15 (Dunbar) and R01-CA141596-02 (Tindle).
Contributors
Author Shiffman designed the study and analysis. Authors Scholl, Dunbar, Tindle, and Ferguson participated in the development of the protocol. Author Scholl oversaw conduct of the study. Authors Li and Shiffman conducted the analysis. Author Shiffman drafted the paper and all authors contributed comment and have approved the final manuscript.
Conflict of interest
Dr. Shiffman consults to and has an interest in eRT, which provides electronic diary services for clinical research.
Acknowledgements
The authors are grateful to Thomas Kirchner, and Deborah Scharf for help launching this study and for input on study design; to Anna Tsivina, Joe Stafura, Rachelle Gish, and Aileen Butera for their work conducting research sessions; to Chantele Mitchell-Miland for data management and preparation; and to Alexandra Cardy for assistance with manuscript preparation. We benefited from the comments of Michael Sayette and Kenneth Perkins on an earlier version of this manuscript.
References (58)
- et al.
Reactivity to smoking cues and relapse: two studies of discriminant validity
Behav. Res. Ther.
(1988) - et al.
Bringing the real world into the laboratory: personal smoking and nonsmoking environments
Drug Alcohol Depend.
(2010) - et al.
Neural substrates of smoking cue reactivity: a meta-analysis of fMRI studies
NeuroImage
(2012) - et al.
Dissociable effect of acute varenicline on tonic versus cue-provoked craving in non-treatment-motivated heavy smokers
Drug Alcohol Depend.
(2013) - et al.
Brain reactivity to smoking cues prior to smoking cessation predicts ability to maintain tobacco abstinence
Biol. Psychiatry
(2010) - et al.
Reactivity to nicotine cues over repeated cue reactivity sessions
Addict. Behav.
(2007) - et al.
The relevance of recent developments in classical conditioning to understanding the etiology and maintenance of anxiety disorders
Acta Psychol.
(2008) - et al.
Individual differences in cue reactivity among smokers trying to quit: effects of gender and cue type
Addict. Behav.
(1998) - et al.
Prediction of lapse from associations between smoking and situational antecedents assessed by ecological momentary assessment
Drug Alcohol Depend.
(2007) - et al.
Evidence for greater cue reactivity among low-dependent vs. high-dependent smokers
Addict. Behav.
(2010)
The magnitude and reliability of cue-specific craving in nondependent smokers
Drug Alcohol Depend.
The impact of cigarette deprivation and cigarette availability on cue-reactivity in smokers
Addiction
Clinical pharmacology of nicotine: implications for understanding, preventing, and treating tobacco addiction
Clin. Pharmacol. Ther.
Comparison of methods for measurement of smoking behavior: mouthpiece-based computerized devices versus direct observation
Nicotine Tob. Res.
Varenicline effects on craving, cue reactivity, and smoking reward
Psychopharmacology (Berl.)
Generalized craving, self-report of arousal, and cue reactivity after brief abstinence
Nicotine Tob. Res.
Meta-analysis of cue-reactivity in addiction research
Addiction
International Affective Pictures System: Digitized Photographs
Environments as cues to smoke: implications for human extinction-based research and treatment
Exp. Clin. Psychopharmacol.
Proximal versus distal cues to smoke: the effects of environment on smokers’ cue-reactivity
Exp. Clin. Psychopharmacol.
Evaluation of the brief questionnaire of smoking urges (QSU-brief) in laboratory and clinical settings
Nicotine Tob. Res.
Induction of smoking urge through imaginal and in vivo procedures: physiological and self-report manifestations
J. Abnorm. Psychol.
What does cue-reactivity have to offer clinical research?
Addiction
Do smokers crave cigarettes in some smoking situations more than others? Situational correlates of craving when smoking
Nicotine Tob. Res.
Limbic activation to cigarette smoking cues independent of nicotine withdrawal: a perfusion fMRI study
Neuropsychopharmacology
The Fagerström test for nicotine dependence: a revision of the Fagerström tolerance questionnaire
Br. J. Addict.
Influence of affective manipulations on cigarette craving: a meta-analysis
Addiction
The associative basis of cue-elicited drug taking in humans
Psychopharmacology
The interaction of psychosocial and biological determinants of tobacco use: more on the Boundary Model
J. Appl. Soc. Psychol.
Cited by (32)
Trauma cue-elicited emotional response profiles influence acute experiences of alcohol urge
2022, Journal of Behavior Therapy and Experimental PsychiatryDistress tolerance and reactivity to negative affective cues in naturalistic environments of cannabis-using emerging adults
2022, Drug and Alcohol DependenceCitation Excerpt :Interestingly, no associations between affective reactivity and cannabis craving were observed in this emerging adult, largely CUD, sample. Though our results were partially inconsistent with previous laboratory research, it is not uncommon for laboratory or retrospective findings and EMA findings to be in conflict, and the source of such conflict is important to understand in subsequent research (Shiffman et al., 2015, 2008). Future work should also examine patterns of cannabis use following briefly administered NA cues.
Relative expected value of drugs versus competing rewards underpins vulnerability to and recovery from addiction
2020, Behavioural Brain ResearchCitation Excerpt :A related set of findings have been reported with the cue-induced craving model, where drug cues have been shown to increase self-reported (verbalised) desire for the drug to the extent that those cues signal the probability that drug choice will produce the drug [41]. Again, however, cue induced increases in subjective craving are not reliably associated with dependence severity [42,43]. The conclusion, therefore, is that drug cues increase drug choice by raising the expected probability of receiving the drug (consistent with the unified decision making model), but sensitivity to this effect does not appear to play a role in vulnerability to dependence, an observation that contradicts cue-reactivity theory of dependence.
Naturalistic measurement of dual cue attentional bias in moderate to heavy-drinking smokers: A preliminary investigation
2020, Drug and Alcohol DependenceSex differences in affect-triggered lapses during smoking cessation: A daily diary study
2018, Addictive BehaviorsCitation Excerpt :Indeed, recent work has demonstrated that negative affective states are related to cigarette cravings and subsequent lapses (Ferguson & Shiffman, 2014). Berlin et al. (2003) found that acute positive affect was related to increased smoking and risk of relapse (Shiffman et al., 2015). Findings overall have been mixed, with the majority of studies suggesting that lapses are related to negative affective states (e.g., Ameringer & Leventhal, 2010), while still others have found that lapses are more closely related to positive affective states, including good mood and relaxation (e.g., Vinci et al., 2017).
Reprint of The effect of N-acetylcysteine and working memory training on cocaine use, craving and inhibition in regular cocaine users: correspondence of lab assessments and Ecological Momentary Assessment
2018, Addictive BehaviorsCitation Excerpt :EMA can also be useful in validating laboratory assessments. For example, one can identify the conditions under which data from laboratory assessments are associated with EMA data (Linas et al., 2016; Litt, Cooney, & Morse, 2000; Ramirez & Miranda, 2014) and when they are not (Shiffman et al., 2015). The primary aim of this study was to investigate whether 25-days of treatment with 2400 mg/day NAC combined with WM-training is effective in reducing cocaine use, craving and inhibition in a randomized, double-blind, placebo-controlled trial.
- 1
Li and Tindle have changed institutions since the current work was completed. Li is now at the Department of Biostatistics and Computational Biology at the Dana Farber Cancer Institute, while Dr Tindle is at the Department of Medicine, Vanderbilt University.