ANTECEDENTS AND POSTCEDENTS OF TEMPTATIONS DURING SMOKING CESSATION: AN ECOLOGICAL MOMENTARY ASSESSMENT STUDY
Cigarette smoking is the leading cause of preventable death in the U.S. Most cigarette smokers want to quit but are unable to do so, even when using evidenced-based treatments. A better understanding of the psychological processes underlying relapse may lead to better interventions. Theory and data suggest that cognitive biases, including attentional bias, may elicit craving and temptations to smoke, and undermine cessation attempts. However, few studies have examined these relationships in the field and none have examined the time course of these relationships. The current study used ecological momentary assessment (EMA) to examine the relationship between attentional bias and temptations during the first week of a quit attempt. Participants (n=119) carried around a Personal Digital Assistant (PDA) for one week and completed a smoking Stroop task at random times (random assessments; RAs), and at temptation episodes (TAs). They also completed a self-report measure of attentional bias. The overall goal of the dissertation is to examine the association between attentional bias and temptations. There are three specific aims. The first specific aim will examine the relationship between attentional bias & Assessment Type (RAs vs. TAs) during attempted abstinence from smoking. The second specific aim will examine whether attentional bias provokes temptations during attempted abstinence from smoking. The third specific aim will examine whether temptations provoke attentional bias during attempted abstinence from smoking. Additional exploratory analyses will investigate the reciprocal association between attentional bias and temptations/craving.
Cigarette smoking is the leading cause of preventable death and disease in the United States (Office of the Surgeon General, 2014). Since the surgeon general’s report in 1964, scientific research has associated cigarette smoking with a variety of serious health conditions and death (Centers for Disease Control and Prevention, 2016; Office of the Surgeon General, 2014; U.S. Department of Health and Human Services, 2004; Xu, Bishop, Kennedy, Simpson, & Pechacek, 2015). Considerable effort has been made to prevent and ultimately reduce cigarette smoking. Despite these efforts, approximately 4,000 people smoke their first cigarette each day in the United States (Office of the Surgeon General, 2014), and most attempts to quit smoking end in failure (U.S. Department of Health and Human Services, 2004).
A better understanding of the psychological processes underlying relapse may lead to better interventions. Theory and data suggest that cognitive biases, including attentional bias, may elicit craving and temptations to smoke, and undermine cessation attempts. However, few studies have examined these relationships in the natural environment and none have examined the time course of these relationships in this context. The current study used ecological momentary assessment (EMA) to examine the relationship between attentional bias and temptations during the first week of a quit attempt.
The background material for this dissertation is organized as follows. Chapter 1 reviews the literature on the health effects of smoking and smoking cessation, and the available treatment options for nicotine addiction. Chapter 2 reviews literature on psychological theories of nicotine addiction, with particular focus on Robinson and Berridge’s incentive sensitization theory (1993), which is most germane to the current dissertation study. Chapter 3 reviews the use of Ecological Momentary Assessment (EMA) to examine psychological processes in relapse. Chapter 4 reviews the literature from laboratory and EMA studies that have examined relationships between attentional bias and craving. Finally, Chapter 5 reviews three preliminary studies that are pertinent to current dissertation study dissertation, describes the rationale for the dissertation study, and presents the Specific Aims.
Health Consequences of Smoking
The 1964 report by the United States Surgeon General is often considered to be one of the first public statements supported by scientific research that condemned smoking due to its negative health consequences (Office of the Surgeon General, 2014). Cigarette smoking is known to cause or be associated with numerous negative health conditions and diseases including multiple types of cancers, coronary heart disease, emphysema, stroke, diabetes and chronic obstructive pulmonary disease (Novello, 1990; U.S. Department of Health and Human Services, 2004, 2006, 2012).
Smoking cigarettes is not only harmful to the individual smoker, there are also associated risks to others in the immediate environment, and there are risks to developing fetuses. Exposure to second-hand smoke in one’s environment is associated with increased risk of diseases including heart disease, cancer, and respiratory disease (U.S. Department of Health and Human Services, 2006). Smoking while pregnant is associated with increased risk of miscarriage, low birth weight, and other birth complications (U.S. Department of Health and Human Services, 2004).
Cigarette smoking in the military is of particular concern. Current rates of smoking are higher in the military compared to the civilian population. Approximately 24% of active duty military members in the U.S. are current smokers compared to 21% of the civilian population ages 18-65 (The Department of Defense, 2013). Smoking directly affects military members’ ability to perform their duties which can jeopardize not only the individual smoker, but also the unit as a whole.
Smoking Prevalence and Monetary Impact
In the 40 years following the release of the surgeon general’s report outlining the damaging effects of smoking in 1964, an estimated 12 million people died from smoking related illnesses (U.S. Department of Health and Human Services, 2004). Cigarette smoking is the largest single contributor to preventable deaths in the United States with more than 443,000 people dying each year from smoking related diseases (Office of the Surgeon General, 2014).
As noted above, the negative effects of smoking also extend to bystanders, where it is estimated 42,000 deaths each year in the U.S. that are attributable to second-hand smoke (Centers for Disease Control and Prevention, 2016). Annually, smoking related diseases cost nearly $170 billion in direct medical expenses and $156 billion in lost productivity (Office of the Surgeon General, 2014; Xu et al., 2015).
In spite of the overwhelming evidence outlining the risks associated with smoking, approximately 43 million people smoke cigarettes in the United States with an estimated 4,000 people trying cigarettes for the first time each day (Office of the Surgeon General, 2014; U.S. Department of Health and Human Services, 2004).
Health Consequences of Smoking Cessation
Smoking cessation has been shown to have immediate and long-term health benefits including reducing the risk of suffering a stroke, reducing the risk of developing numerous types of cancers, and reducing the risk of developing coronary heart disease (U.S. Department of Health and Human Services, 2004). Overall, people who quit smoking live longer than those who continue to smoke (Centers for Disease Control and Prevention, 2016). Unfortunately, even with the development of pharmaceutical, telephone, group interventions, and online cessation resources (which will be described in more detail later), smoking cessation rates remain low. Even though three out of four smokers report that they would like to quit, less than 5% of people who attempt to quit smoking remain abstinent for greater than 3 months (U.S. Department of Health and Human Services, 2004). It is also estimated that one third of smokers who are able to quit for one year begin smoking again (U.S. Department of Health and Human Services, 2004).
Smoking cessation is difficult in part due to the addictive nature of nicotine. Out of the thousands of chemicals contained in modern cigarettes, a large amount of research has implicated nicotine as a major factor in cigarette addiction. The addictive qualities of nicotine have been extensively studied and implicated as a major factor in tobacco dependence and with difficulty in cessation or maintaining abstinence. Relevant research on the addictive nature of nicotine and cigarettes is reviewed below.
Studies have investigated the acute hedonic and cognitive effects of smoking of smoking and nicotine. Varying methods of nicotine administration have been studied including intravenous, transdermal, and inhalation. Nicotine absorbed from cigarette smoke has been linked to improvements in both attention and memory (Heishman, Kleykamp, & Singleton, 2010). A meta-analysis of 41 double-blind, placebo controlled studies found that smoking improved attention and memory (Heishman et al., 2010). A literature review on the acute subjective effects of nicotine and smoking and has demonstrated that nicotine and smoking have acute pleasurable effects (Kalman, 2002). Smokers often report relief of negative symptoms, particular negative affect, as a positive outcome associated with smoking (Baker, Brandon, & Chassin, 2004). In sum, smokers may become dependent on the acute hedonic or cognitive effects of nicotine.
Tolerance to and withdrawal from nicotine are hallmark criteria of nicotine dependence (American Psychiatric Association, 2013). Repeated and chronic exposure to nicotine leads to dependence (Baker et al., 2004). Simply put, individuals who have developed a nicotine addiction are dependent on this substance to function normally. When individuals do not take in the required level of nicotine, they will experience nicotine withdrawal. The symptoms of nicotine withdrawal vary depending on the individual but are characterized by negative changes in mood, cognition, and physical symptoms (American Psychiatric Association, 2013). During withdrawal, individuals can expect to experience dysphoria, anxiety, difficulty concentrating, increased appetite, weight gain, sleep disruptions, and increased cravings which may last for weeks or months following cessation (Baker et al., 2004). These negative symptoms have been theorized to be a major factor in maintaining smoking. Avoidance of withdrawal symptoms has been indicated as a reason individuals fail in a cessation attempt (Robinson & Berridge, 1993).
Treatments for Nicotine Addiction
As noted above, most smokers want to quit. Smokers undergoing a quit attempt may do so utilizing pharmacological and/or behavioral interventions. Nicotine replacement therapy (NRT) is one pharmacological treatment used to aid smokers undergoing a quit attempt. NRT is not a single treatment and consists of various products inducing transdermal patches, lozenges, gum, nasal spray, and oral inhaler which all contain specific levels of nicotine (U.S. Food and Drug Administration, 2010). These treatments are intended to work by replacing the nicotine a smoker typically obtains from cigarettes and then reducing the amount of nicotine ingested overtime in order to reduce negative withdrawal effects (Ehrman et al., 2002). A Cochrane review of all types of NRT currently available indicate use of NRT improves cessation rates by 50-70% compared to smokers who attempt to quit unaided (Stead et al., 2012).
Bupropion and varenicline are two other FDA approved cessation medications developed to improve rates of quitting (U.S. Food and Drug Administration, 2010). Unlike NRT, these medications do not contain nicotine. The effectiveness for bupropion was found to be similar to NRT and medication adherence was considered good due to the low incidence of significant negative side effects (Hughes, Stead, Hartmann-Boyce, Cahill, & Lancaster, 2014). Varenicline has been shown to improve cessation rates compared to control groups receiving no medication and is more effective in extended treatments compared to bupropion (Hajek et al., 2013).
Research studies of psychological and behavioral intervention programs indicate they improve the odds a smoker will quit compared to non-treatment control groups. A review of counseling programs targeting tobacco cessation have shown individual counseling is effective in assisting smokers to quit (Lancaster & Stead, 2008). These findings were consistent with the clinical practice guidelines published by the U.S. Public Health Service which recommends utilizing intensive counseling for smoking cessation (U.S. Department of Health and Human Services, 2008).
Despite the development of effective pharmacological and behavioral interventions, over half of all smokers undergoing a quit attempt do so without unaided (Chapman & MacKenzie, 2010). Of those, only 3-5% who attempt to quit unaided are able to stay quit by the 6-12 month follow-up (Hughes, Keely, & Naud, 2004).
In summary, most smokers report the desire to quit, and there are effective pharmacological and behavior intervention programs that assist individuals in quitting. Yet most people choose to quit unaided with poor overall cessation rates.
The current study enrolled a large proportion of African American smokers. Therefore, a brief review of smoking in African Americans will follow. Research investigating racial disparities of smoking and cessation has found significant differences among racial groups. Compared to white smokers, minority smokers have lower rates of cessation despite being more likely to be light or intermittent smokers (Trinidad, Perez-Stable, White, Emery, & Messer, 2011). African-American smokers are approximately 50% less likely to be successful in quitting compared to non-Hispanic White smokers (Trinidad et al., 2011). Research investigating what underlies these disparities has found several factors that may contribute to lower cessation rates in minority populations including being less likely to have complete smoking bans at home, being less likely to be advised to quit smoking by a health care professional, and being less likely to utilize smoking medications and NRT (Trinidad et al., 2011). Further research investigating the causes of lower cessation rates is necessary since minority smokers experience a greater level of adverse health effects from smoking compared to non-Hispanic White smokers (Trinidad et al., 2011).
(Robinson et al., 2015)
Summary
In summary, cigarette smoking remains a significant public health problem. Most smokers want to quit. However, despite the presence of efficacious smoking cessation therapies, most smokers attempt to quit on their own, and most quit attempts end in failure. Relapse rates are particularly high for minority smokers. A better understanding of the psychological processes underlying smoking cessation and relapse may lead to better interventions. The next chapter reviews some pertinent literature on the psychological processes underlying relapse.
CHAPTER 2: PSYCHOLOGICAL PROCESSES UNDERLYING RELAPSE
As noted in the previous chapter, a better understanding of the psychological processes underlying relapse may lead to better interventions. A review of theories pertaining to relapse will be reviewed next. Major theories of addiction and relapse include positive reinforcement theory, negative reinforcement theory, and, most pertinent to the current study, incentive-sensitization theory.
Positive reinforcement theory of drug use is based on learning and conditioning principles where individuals initiate and maintain drug use due to the positive rewards resulting from drug use (Wise & Koob, 2014). Smokers report experiencing various rewards of smoking including improved mood, attention, and concentration (Heishman et al., 2010). Through classical conditioning, smoking cigarettes becomes associated with these rewards. However, it has been noted that while positive reinforcement may play a role in smoking, it is unlikely to be sufficient in maintaining addiction as the acute effects of smoking are relatively modest (compared to other drugs of abuse) and the pleasure experienced may decrease over time due to tolerance.
In contrast to positive reinforcement, negative reinforcement theory indicates that individuals use drugs primarily to avoid unpleasant affective, cognitive and physical withdrawal symptoms. As noted earlier, when abstaining from smoking, regular smokers typical experience several negative withdrawal symptoms including irritability, difficulty concentrating, and increased cravings when they abstain from smoking (American Psychiatric Association, 2013). Smoking will relieve these symptoms, a form of “escape behavior”, and smoking consistently throughout the day prevents an individual from experiencing withdrawal in the first place. Smoking may also be negatively reinforcing if it removes negative affect caused by another (non-withdrawal related) source (Baker et al., 2004).
However, some researchers have argued that negative reinforcement is unlikely to provide a complete account of tobacco addiction (Robinson & Berridge, 1993). For example, human and animal studies of addiction have shown individuals will reinitiate drug use even after withdrawal symptoms have subsided (Shaham, Shalev, Lu, De Wit, & Stewart, 2003). Most importantly, at many episodes of relapse, individuals report being in a good (or neutral) mood, meaning that there is no negative state to alleviate through smoking (Shiffman, Paty, Gnys, Kassel, & Hickcox, 1996). It is difficult to understand how negative reinforcement accounts for relapse when individuals are not experiencing any withdrawal or negative affect. Moreover, negative reinforcement does not easily account for drug use initiation.
Incentive-Sensitization Theory
In the Incentive-Sensitization Theory presents a biopsychological theory of addiction (Robinson & Berridge, 1993) (IST 93). IST 93 outlines three major features of addiction including drug craving, a strong tendency to relapse back to drug use both early in the quitting process and after long periods of abstinence, and the relationship between drug Wanting and drug Liking (Robinson & Berridge, 1993).
Craving can be defined as “a strong desire or sense of compulsion to take the drug” by the International Classification of Diseases (World Health Organization, 2016). The Incentive-Sensitization Theory is named due to idea that it is the sensitization of neural pathways and the enhanced incentive value assigned to drugs and drug cues that underlies addiction (Robinson & Berridge, 1993). It is hypothesized that the neural pathways become sensitized through drug taking which influences the process of assigning or attributing incentive salience to the mental representations of drug cues (Robinson & Berridge, 1993).
More specifically, according to IST 93, under normal circumstances a stimulus is processed by a brain system the authors refer to as the “Pleasure Integrator”, and identified as a pleasant stimulus. This process leads to Liking. The Liking system (Pleasure Integrator) communicates with a Wanting system that a pleasurable stimulus has been identified. An “Incentive Salience Attributor” interacts with another mechanism, termed “Associative Learning” by IST 93, and the end result of this processing is that incentive salience is assigned to the mental representations of stimuli paired with pleasure. When this attribution of incentive salience occurs, these “incentive stimuli” provoke Wanting. That is, the Wanting system mediates “subjective craving”, attentional bias, and consumption (Robinson and Berridge, 1993).
Usually, incentive salience is only assigned to stimuli that are associated with pleasurable outcomes. However, Robinson and Berridge (1993) argue that drugs, including nicotine, can cause neuroadaptations that give rise to an attribution of incentive salience (Wanting) without pleasure. After repeated drug use, “sensitization” occurs in some individuals (as noted above, this is why the theory is termed “incentive sensitization” theory), with the result that drugs and drug-related cues are craved more than would be expected based on levels of Liking.
Furthermore, according to IST 93, drug Liking decreases as drug Wanting increases. Wanting increases because due to incentive sensitization, as just described. IST suggests that there is tolerance (not sensitization) to drug Liking over time. After repeated drug administration, individuals typically experience tolerance to that amount of drug and experience less pleasure with each subsequent administration. It becomes necessary to consume greater amounts of the substance to experience a similar effect. This is why some addicted individuals report that they previously found drugs pleasurable but they no longer do so – but they still want them! Therefore, although Wanting and Liking usually go together (in non-addicted individuals), IST 93 views these concepts as separate constructs.
As just noted, repeated pairing of drug cues with drug use over time increases the attractiveness of drug-associated stimuli and increases a person’s Wanting for drugs (Robinson & Berridge, 1993). IST 93 is clear that Wanting is not the same as Liking. This distinction is in direct discord with other theories of addiction that hypothesize it is drug Liking and the desire to experience the pleasurable subjective effects of drugs which serves as the primary motivation for drug-seeking and drug-use behavior (Robinson & Berridge, 1993).
Tibboel et al. (2015) have reviewed the literature that demonstrates the separability of Wanting and Liking. Because Liking is not a focus of the current dissertation study, this literature is not reviewed here. However, both animal and human studies have provided evidence that Wanting can occur without Liking, and Liking can occur without Wanting (Tibboel, De Houwer, & Van Bockstaele, 2015).
Most pertinent to the current dissertation is the idea that the Wanting system is more important in addiction than the Liking system, and the notion that Wanting is associated with attention to drug cues. As Berridge (2009) notes:
“….incentive salience transforms the mere sensory shape, smell or sound into an attractive and attention-riveting incentive. Once attributed, the incentive percept becomes difficult to avoid noticing, the eyes naturally move toward the incentive, it captures the gaze and becomes motivationally attractive, and the rest of the body may well follow to obtain it”
As can be seen, IST gives attentional bias a prominent role. However, IST does not describe in detail the proposed causal relationships between attentional bias and craving. Rather, attentional bias and craving were viewed by IST as two indices of a Wanting system. However, this idea was further developed by Franken (2003), and is reviewed in Chapter 4.
Neurobiology of Wanting and Liking
- Many theories postulate that dopamine underlies addition
- Wanting and liking as separate neurobiological constructs
- Explanation of each
- What repeated drug use does to shift levels of liking vs. wanting
IST 93 contains significant detail on the neurobiological changes or adaptations that occur after repeated drug use and contend it is these drug-induced changes of neural circuits that result in significant craving and relapse (Robinson & Berridge, 1993). IST 93 relied heavily on the neurobiological pathways associated with Liking and Wanting to explain addiction. A detailed account of the neurobiological processes underlying Liking and Wanting is beyond the scope of this manuscript. However, a brief review is presented below.
IST 93 noted that nearly all drugs of addiction activate the dopamine system and that the mesotelencephalic dopamine system is the underlying neural pathway that facilitates the incentive-sensitization observed in those with drug addiction (Robinson & Berridge, 1993). This view is a departure from previous theories of addiction that implicate dopamine as the neurotransmitter association with pleasure (liking) (citation). Although dopamine transmission in the mesolimbic system had historically been associated with pleasure (liking), IST 93 conceptualized dopamine transmission as mediating the process of attribution of incentive salience (described earlier). The mesotelencepahalic dopamine system has been further indicated as the route of for increase wanting in studies that blunted or blocked dopamine transmission. Sensitization did not occur in lab animals that were also given a dopamine agonist which indicates that dopamine is key to sensitization (Robinson & Berridge, 1993). With continued drug use, dopamine transmission is increased over time (sensitization), and this process underlies sensitization of incentive salience.
The role of incentive salience has been further studied since being discussed in IST 93. A more recent review of Liking and Wanting provided an update to the how specific brain regions interact to affect Liking and Wanting separately (Berridge, 2009). This nuanced understanding of the role of brain structures only became possible after IST 93 was published due to the development of more sensitive brain measurement technology. Berridge discussed how the mesocorticolimic or mesolimbic dopamine system is the major underlying neural pathway involved in incentive salience. The dopamine neurons of this system originate in the midbrain and extend to the nucleus accumbens, striatum, amygdala, and prefrontal cortex (Berridge, 2009). Studies have shown that manipulation of dopamine activity within the mesolimbic system is associated with increased Wanting without increased Liking. In addition to sensitization occurring from a large number of drug administrations, it is possible for enduring and permanent sensitization to occur as a result of drug binges (Berridge, 2009).
As previously mentioned, Wanting and Liking can occur separately which is supported by neurophysiological research. Research has indicated that specific “hotspots” within the limbic system and nucleus accumbens are specifically orirented to generating Liking, but not wanting (Berridge, 2009). Generating liking is a much more specific process than generating Wanting. Research indicates that Liking requires opioid stimulation of specific “hotspots” whereas Wanting can be stimulated more by more diffuse stimulation of the multiple limbic structures by dopamine (Berridge, 2009).
In addition to research on “hedonic hotspots,” other research has indicated there are functional differences between the “core” and the “shell” of the nucleus accumbens. Microinjections of opioids on the on the “shell” produced increases in Liking while microinjections in the “core” did not increase Liking (Pecina & Berridge, 2000). This result further supports the idea that liking is controlled by specific brain structrues and regions.
In a more recent review, Berridge (2009) elaborated further and wrote that “different brain mechanisms seem to mediate wanting and liking for the same reward.” Berridge (2009) noted that there is a “hedonic hotspot” in the nucleus accumbens that is part of the Liking pathway but that other regions in the nucleus accumbens are part of the Wanting pathway (Tibboel et al., 2015). Relatedly, some research has drawn a distinction between the “core” and the “shell” of the nucleus accumbens. In an early paper, Pecina and Berridge (2000) identified a “liking hotspot” on the shell of the nucleus accumbens that appeared to increase “liking” behavior. A similar reaction was not found when testing areas in the core of the nucleus accumbens. The main idea emerging from this area of research is that the mesolimbic dopamine system is associated with Wanting, whereas Liking is mediated by activity in more restricted subcortical “hedonic hotspots”. In addition, Tibboel et el. (2015) note that opioid transmission is more closely associated with Liking than Wanting
In the following sections, research examining the relationship between stress/negative affect and craving will be reviewed. Although stress and negative affect are not the central focus of the current dissertation, this review is provided for the following reasons. First, stress and negative affect have been identified as major factors contributing to relapse (Shiffman & Waters, 2004), and therefore they are important to review. Second, stress and negative affect were examined in a preliminary study that is germane to the current study, as described later (Lammers, 2016).
Heckman and colleagues completed a meta-analysis of 27 laboratory studies to determine what effect negative affect and positive affect had on subsequent craving to smoke (Heckman et al., 2013). The results of this meta-analysis showed that a negative affect manipulation increased post-manipulation craving with a medium effect size (Heckman et al., 2013). Manipulations of positive affect did not increase post-manipulation craving. This meta-analysis is important in that it supports theoretical models and participant self-report that negative affect is associated with an increase in craving (Heckman et al., 2013).
Despite the results of the meta-analysis reported by Heckman et al. (2013) it should be noted that not all studies revealed that a manipulation of negative affect increased craving. Shiffman and colleagues (2013) utilized different cue sets to determine the effect on subsequent craving and smoking behavior. In this study, the negative affect cue set did not significantly increase craving (Shiffman et al., 2013){Shiffman, 2013 #52}. This could be due to the relatively short time period between negative cue exposure and ad lib smoking opportunity.
Craving can be manipulated by use of a cue exposure (imaginal or in vivo) paradigm. Participants have been exposed to drug cues (that elicit craving) and neutral cues (that elicit less craving). The effect of this manipulation on stress and negative affect is then assessed (Drobes & Tiffany, 1997). Consistent with previous studies, participants reported significantly greater craving when exposed to explicit urge cues, regardless of scenario type (Drobes & Tiffany, 1997). There were no significant differences in subsequent craving between imaginal and in vivo scenarios. This study also demonstrated that manipulation of craving in a laboratory setting increased negative affect and decreased positive affect (Drobes & Tiffany, 1997). Thus, there is evidence that stress can increase craving, and that craving can increase stress.
A number of studies have also examined relationships between stress/negative affect and craving in the field using EMA. EMA will be described in more detail in the following chapter, but a few key findings relating to stress/negative affect will be reviewed here. Shiffman et al. (1996) examined negative affect at random assessments, temptation assessments, and lapse assessments in smokers attempting to quit smoking. Participants were asked to report their level of negative affect as it was just prior to the temptation. Higher levels of urge to smoke and craving were found prior to temptation assessments compared to random assessments (Shiffman, Paty, et al., 1996). More pertinent to the current paper, negative affect ratings were higher just prior to temptation episodes than at random assessments. Negative affect was highest just prior to lapse assessments (Shiffman, Paty, et al., 1996).
In a second study researchers compared individuals who were successful in a quit attempt to individuals who subsequently relapse back to smoking in an effort to identify how certain characteristics of temptation lead to relapse (Shiffman, Gnys, et al., 1996). The data collected from these two types of assessments were compared between those participants who maintained abstinence (maintainers) and those who lapsed back to smoking (lapsers). The results of this study show that temptations assessments do not significantly differ between groups in terms of setting, frequency, intensity, and affect (Shiffman, Gnys, et al., 1996). Similar to previous findings, data showed that temptations, regardless of groups, were more likely to occur when consuming alcohol or coffee, when exposed to smoking cues, and when negative affect was significantly elevated (Shiffman, Gnys, et al., 1996).
Recent EMA studies have also manipulated cue type (either neutral or smoking) and then measured changes in craving and changes in mood at the same time point. A cue reactivity EMA (CREMA) study by Warthen and Tiffany found significant increases in craving following exposure to smoking-related cues but not neutral cures (2009). There were also significant increases in negative mood and significant decreases in positive mood following exposure to smoking-related cues with the photographs having a larger magnitude of change compared to imagery scripts (Warthen & Tiffany, 2009).
A similar CREMA design was used to investigate how craving and mood were influenced by presenting smoking and neutral cues utilizing photographic and in-vivo presentation methods (Wray, Godleski, & Tiffany, 2011). In line with the previous study, smoking cues elicited higher craving than neutral cues for both presentation types (in-vivo an photographic), and negative mood was significantly worse following exposure to smoking cues (Wray et al., 2011).
Lammers (2016) described another EMA study that investigated the temporal aspect of how stress and negative affect are related to temptation episodes. This study will be reviewed in the ‘preliminary studies’ section.
In general, both laboratory and EMA studies have provided support for the hypotheses that stress/negative affect increases craving (or elicits temptations) and that craving increases stress/negative affect.
The following chapter describes EMA methodology in much further detail, including its use for the assessment of attentional bias in the field.
CHAPTER 3: ECOLOGICAL MOMENTARY ASSESSMENT
Ecological momentary assessment can be best explained by defining each of its terms. “Ecological” refers to the setting in which participants complete the assessments. EMA studies are “ecologically valid” because participants complete the assessments in their natural environments rather than under strictly controlled laboratory settings. The term “momentary” refers to the fact that participants respond to how they are feeling “at that moment”. In other words, participants are not asked to think about an earlier event but rather are probed with how they are feelings at the time the assessment is administered. The addition of repeated sampling produces rich datasets with greater external validity compared to laboratory studies (Shiffman, Stone, & Hufford, 2008)
The tools available for EMA studies have developed over the years and consist of simple self-report journals and substance use logs, electronic diaries, personal desktop assistants (PDAs), and smartphones (Shiffman et al., 2008). EMA has been previously used to identify variables related to temptations and relapse in studies of addiction. EMA is an appropriate study design when investigating addiction and relapse due its ability to capture the episodic nature of drug administration, contextual information related to drug taking, as well as changes throughout the day.
In sum, EMA studies are a useful extension of laboratory studies because EMA studies have greater external validity, allow for larger number of sampling points, and allow researchers to investigate the temporal nature of how variables are related (Shiffman et al., 2008). Shiffman, Stone, and Hufford (2008) review the history and suitability of EMA as a type of study design and the relevant background information is contained below.
There are two main assessment schedules that allow for the participant to complete multiple measures outside of the laboratory setting (Shiffman et al., 2008). EMA studies follow different schedules of assessment administration depending on the hypothesis and goals of research (Shiffman, 2009). Some studies utilize an event-based schedule of reporting where the participants are instructed to complete an assessment each time a previously defined event occurs (Shiffman et al., 2008). These events could include incidents of seeing another person use their preferred substance, when a person experience a particular mood, or after a meal is eaten (Shiffman, 2009). Event-based assessments provide valuable information regarding the immediate internal and external factors associated with drug use or abstinence.
The other type of assessment schedule is time-based. Time-based assessment can occur at either fixed intervals (e.g. every 4 hours after waking and until bedtime) or, more commonly, at random times (e.g. 4 assessments randomly distributed during waking hours) (Shiffman et al., 2008). Compared to event-based assessments, time-based assessments gather information regarding the changes in subjective and objective factors that occur throughout the day. This information is useful when determining how subtle changes in either mood or cognition are related to subsequent behaviors and outcomes (Shiffman, 2009). Studies can also utilize both event-based and time-based assessments within the same study. This approach was taken in the current study.
Craving and relapse are common elements in models of addiction but can be challenging to study using traditional laboratory research methods. EMA methods allow for closer examination of factors that contribute to relapse by minimizing the reliance on retrospective recall and by assessing participants multiple times in their natural environments (Serre, Fatseas, Swendsen, & Auriacombe, 2015; Shiffman, 2009). EMA methods allow for researchers to gather information on numerous factors related to drug use and addiction including mood, stress, affect, craving, and environmental information (Shiffman, 2009). Research conducted with separate groups of crack-cocaine, opiate, and heroine addicted participants resulted in sufficient compliance to the study protocols and demonstrated the utility of employing EMA methods and handheld technology to collect data on illicit drug use (Shiffman, 2009).
A major concern of any EMA study is whether participants are able to adhere to the study protocol and whether the data gathered during assessments is valid and reliable. This becomes even more of a concern when working with individuals with drug addiction who may have generally chaotic lifestyles (Shiffman, 2009).
Researchers studying individuals with alcohol disorder found that participants reported completing time-based assessments 81 percent of the time when alerted to by a wrist-worn alarm (Litt, Cooney, & Morse, 1998). However, information gathered during the study debriefing found that 70 percent of the participants admitted to beginning the assessment when alerted but finishing the assessment at a later time. These authors utilized paper-and-pencil assessment record forms which allowed for participants to deviate from study procedures (Litt et al., 1998).
Shiffman and colleagues utilized handheld computers which recorded accurate date and time information that the assessment was completed but also require the participants to complete the entire assessment in the same sitting. Compliance was found to be 91 percent and the data was collected as intended due to using computers rather than paper and pencil forms (Shiffman et al., 1994). A meta-analysis of studies utilizing electronic diaries yielded compliance rates above 90 percent and that participants significantly preferred electronic diaries over paper-and-pencil forms (Hufford & Shields).
Other studies involving crack-cocaine, opiate, and heroine addicted participants yielded acceptable compliance when utilizing hand held computers or other elcronic devices to present and record assessments (Shiffman, 2009). In sum, compliance and validity are issues that can be mitigate by employing specific methodologies (such as handheld computers or electronic devices).
(Hufford, 2007; Stone, Shiffman, Atienza, & Nebeling, 2007)
Use of electronic diaries or handheld computers have shown high compliance rates as outlined above. These compliance rates typically looked at assessments overall or time-based assessments. The data are less clear with how compliant participants are with reporting event-based assessments. It is difficult to verify if participants are correctly completing event-based assessment. One study found that participants rarely completed temptation assessments when instructed to complete an assessment when they felt the urge to drink (Litt et al., 1998). However, this study focused on alcohol use (vs. cigarette smoking) and the participants reported confusion with protocol instructions. This resulted in the participants failing to initiate assessments when experiencing an increased urge to drink (Litt et al., 1998).
The low compliance and poor protocol adherence in the above study is inconsistent with studies of smokers that found participants were able to recognize periods of increased urge and properly initiate an event-based assessment (Shiffman et al., 1994).
Some studies use multiple sources of information in an attempt to verify compliance. For example, smoking studies can compare smoking logs with biochemical measures of smoking, such as carbon monoxide or cotinine, to bolster validity (Shiffman et al., 2008). Verifying compliance is even more difficult when measuring psychological processes such as mood and affect.
Psychological sciences often rely on self-monitoring to gather data since many variables of interest cannot be directly measured. In order for data obtained in EMA studies to be valid, it is important that participants’ thoughts, behavior, or emotions are not impacted by the experience of completing assessments A review of self-monitoring literature has found variables that can influence data accuracy. Relevant to this study is the finding that level of motivation for change can impact behavior when participants are required to self-monitor the negative behavior. For example, self-monitoring reduced smoking in the group that was motivated to quit but not in the unmotivated group (Lipinski, Black, Nelson, & Ciminero, 1975). However, if self-monitoring were in and of itself the cause of behavior change, it would have been observed in both groups. Rather, it was necessary for participants to possess motivation for change for self-monitoring to be impact smoking.
In another study of smokers not motivated to quit, self-monitoring of cigarette consumption only decreased smoking by 0.3 cigarettes per day while CO levels remained the same (Shiffman et al., 2002). Reduction in cigarette smoking reached statistical significant but does equate to clinically significant reductions in consumption.
Another concern is that participants might react to the burden associated with completing multiple assessments throughout the day as a result of the alarm beeping, cognitive resources expended completing the assessments, or other factors related to assessment completion. In a chronic pain study, the number of assessments completed per day was manipulated in order to understand how completing an assessment impacts a participant’s responses (Stone et al., 2003). Participants were randomly assigned to groups based on the number of assessments completed each day. Researchers did not find significant differences in self-reported pain level based on the (Stone et al., 2003).
A recent study manipulated the number of EMA assessments completed each day by cigarette smokers undergoing a cessation attempt to determine if assessment grequency (low-1/day vs. high- 6/day) affected short-term abstinence, long-term abstinence, and several participant reported subjective variables (McCarthy, Minami, Yeh, & Bold, 2015). Significant effects were not found on either type of abstinence but completing more random assessments was associate with reduced craving, anxiety, anger, hunger, and positive affect (McCarthy et al., 2015).
This finding could indicate that EMA is not appropriate methodology to employ in cessation studies because completing the assessments changes the participant’ subjective experiences. Shiffman provided a response to the McCarthy study outlining why these findings are not detrimental to EMA studies of cigarette smoking. Perhaps most importantly, EMA assessment frequency did not influence behavior (Shiffman, 2015). Both initial cessation rates and abstinence at 12-week follow-up were not impacted by number of assessments (McCarthy et al., 2015). Shiffman acknowledges that differences in mean values of subjective variables appears troubling for EMA methodology. However, EMA studies investigate the relationship between variables, rather than the absolute values or averages of variables. While the average may have been reduced, there is no indication that the relationship between the variables has been impacted by assessment frequency (Shiffman, 2015).
It is theoretically possible that the manner in which assessments are presented (or initiated) might impact participants’ responses. The reliability and validity of the data would be detrimentally impacted if self-initiated assessments and random assessments exerted different influences on participants. Furthermore, random assessments are prompted by the electronic device and do not require forethought or awareness of particular situational aspects like event-based and self-initiated assessments. Whereas a random assessment is started when the device alerts the participant, researchers cannot truly know the exact precipitating factors that prompted a self-initiated assessment. Clearly defining and training participants when to initiate an assessment is helpful, but it is still possible that they did not initiate an assessment as instructed due to other reasons (distracted by other factors, unable to in immediate environment, chose not to, etc).
There are methodological changes that could take place if a significant difference in influence occurs. For instance, a study could be composed of only one type of assessment– either self-initiated assessments or random assessments. That way, assessments would be equally impacted by any reactivity that occurs due to assessment type. To date, no studies have examined if self-initiated assessments and random assessments influence participants differently and it is not currently considered to be a major impediment to conducting EMA studies.
EMA has been widely used to study smoking cessation. An illustrative study was conducted by (Shiffman & Waters, 2004). Participants attempting to quit were instructed to complete both event based and random assessments. The event based assessments were to be initiated during both periods of significantly heightened temptation as previous defined and incidents where they lapsed. Of particular note, the researchers included a measure of negative affect that was administered during every assessment and measured negative affect as well as perceived stress (Shiffman & Waters, 2004). The focus of this study was to clarify the immediate precipitants of lapse with a focus on stress and negative affect. Results showed that negative affect increased in the hours preceding a subset of lapses (those for which the participants endorsed stress or bad mood as a trigger for the lapse). In contrast, daily ratings of negative affect or stress were not prospectively associated with relapse.
Use of EMA to Study Smoking and Cognition
The development of small, personal electronic devices allows for cognitive measures to be administered in a participant’s normal environment outside of laboratory settings. Personal desktop assistant (PDAs) and cell phones are capable of signaling participants and presenting multiple types of cognitive assessments. A study by Shiffman and colleagues in 1995 illustrated that cognitive assessments can be administered on handheld devices. Dependent and non-dependent smokers completed cognitive and subjective measures at multiple prescheduled assessments in their normal environment (Shiffman, Paty, Gnys, Kassel, & Elash, 1995). The cognitive measures consisted of a simple reaction test and test of mental arithmetic. Results from this study found significant different on assessment performance between dependent and non-dependent smokers that were consistent with previous laboratory studies (Shiffman et al., 1995). The consistency of these findings indicates using PDAs to administer assessments in a field study is methodologically feasible.
In a study examining the feasibility of administering reaction times tests on PDAs in the field, smokers and non-smokers were recruited to conduct field assessments over a one-week period (Waters & Li, 2008). The participants completed subjective measures assessing affect, craving, and mood, measures assessing cigarette, alcohol, and coffee consumption, and contextual measures assessing environmental information. The Classic Stroop, Emotional-Stroop, and Smoking-Stroop tasks were also presented on PDAs. The rate of compliance indicated that EMA is a suitable method to gather ecologically valid cognitive data (Waters & Li, 2008). As described in later sections, numerous other studies have administered cognitive assessments on mobile devices (e.g., (Waters, Marhe, & Franken, 2012; Waters et al., 2014)).
The following chapter describes research that has examined the association between attentional bias and craving, both in the lab and during EMA.
INTRODUCTION TO ATTENTIONAL BIAS
Cognitive models of addiction often conceptualize addiction as a conflict between “automatic” (or “implicit”) processes, which generally increase the risk of drug taking/relapse, and “controlled” (or “explicit”) processes, which attempt to inhibit automatic processes or their output, and serve to decrease the risk of drug taking/relapse (see review in (Stacy & Wiers, 2010)). Automatic processes are relatively fast, parallel, and automatic processes that are outside of one’s conscious awareness (“system 1” in Kahneman’s terminology), whereas controlled processes are slow, reflexive, and serial (“system 2” in Kahneman’s terminology) (Kahneman, 2011). If automatic processes dominate, then drug use/relapse will occur. If controlled processes win out, then abstinence should be maintained. This conceptualization is referred to as “dual-process theory” because of the emphasis on two distinct types of process (Kahneman, 2011).
Two widely studied automatic processes in addiction are attentional bias and approach bias. Attentional bias is generally considered an automatic cognitive process in which a smoker attends to a smoking-related cue without making the conscious decision to attend to the cue (Field & Cox, 2008). Attentional bias is a focus of this dissertation, and assessments of attentional bias will be described in detail later. Another automatic process is “approach bias” which is the tendency to automatically approach drug-related cues (Wiers, Eberl, Rinck, Becker, & Lindenmeyer, 2011). Approach biases are measured with computerized tasks involving motor movements such as the Approach-Avoidance task (Wiers et al., 2013). There is evidence that approach bias is present in heavy smokers but not in ex-smokers. Additionally, approach biases in smokers are correlated with craving (Wiers et al., 2013).
THEORETICAL IMPORTANCE OF ATTENTIONAL BIAS
As noted earlier, IST 93 assigned an important role for attentional bias. Specifically, according to IST 93, attentional bias is an index of Wanting or sensitized incentive salience, which, according to IST, is the core process in addiction.
Although IST 93 assigned an important role to attentional bias, the authors did not provide a detailed account of the proposed causal relationships between attentional bias craving, and drug use. However, Franken (2003) built on Robinson and Berridge’s conceptualizations and provided a more detailed model of the inter-relationships between these measures. Most pertinent to the current study, Franken (2003) placed attentional bias and craving in a reciprocal feedback loop. He suggested that excessive attentional bias contributes to increases in craving which, in turn, contributes to excessive attentional bias (Franken, 2003).
Thus, in this system, increases in attentional bias can lead to (cause) increases in craving. Increases in attentional bias may cause increases in craving due to increases in exposure to drug cues. Thus, an individual with a high attentional bias is more likely to be exposed to smoking cues, which in turn may provoke craving (see Figure 1). Conversely, according to Franken (2003), increases in craving (or feeling tempted to smoke) can lead to (cause) increases in attentional bias (see Figure 2). However, Franken (2003) did not elaborate on the underlying mechanisms for this causal relationship.
Franken’s (2003) model also predicts that there should be an association between attentional bias and drug use or relapse, partly mediated by craving. Some research has shown that attentional bias for drug cues has been found to prospectively predict outcome in addictive behavior (e.g., (Waters, Shiffman, Sayette, et al., 2003); (Janes, Pizzagalli, Richardt, de, et al., 2010); (Powell, Dawkins, West, Powell, & Pickering, 2010)). Thus, interventions that reduce attentional bias may improve treatment outcomes in the addictions.
Franken (2003) also notes that attentional bias for drug stimuli, in addition to increasing craving, may also contribute to an increase in drug-related cognitions. For example, users may retrieve memories of drug-taking episodes or ruminate about drug taking. These additional cognitive processes may interfere with working memory or executive function, and undermine cessation attempts. However, the assessment of other drug-related cognitions is beyond the scope of this study, which focuses solely on the association between attentional bias and craving.
Attentional bias has examined with multiple substances including alcohol, heroin, cocaine, and tobacco in both laboratory and field settings (Field, Munafo, & Franken, 2009). While attentional bias has been studied in addiction research, it is worth noting that attentional bias also been extensively investigated in the emergence and maintenance of clinical conditions including anxiety and depressive disorders (MacLeod, Mathews, & Tata, 1986).
Given the theoretical importance of attentional bias, it is important to understand how it is assessed. In fact, there are many methods used to assess cognitive biases including reaction time assessments, self-report assessments, the use of eye movements, and measures of brain activity. A comprehensive review of all measure types is outside the scope of this manuscript. The following is a brief review of the most commonly-used assessments.
Attentional bias has most commonly been assessed using reaction time tasks derived from experimental cognitive psychology. The two most widely used tasks have been the addiction Stroop task, which is used in the current study, and the visual probe task. These tasks are reviewed below.
In the original “classic” Stroop task (Stroop, 1935), participants are required to name the colors of congruent (e.g., GREEN written in green ink) and incongruent color words (e.g., RED written in green ink). Participants are typically much faster to color-name incongruent than congruent stimuli, due to the presence of the conflict in the latter. The addiction Stroop tasks involve presenting drug-related and neutral words in different color font (Cox, Fadardi, & Pothos, 2006). Participants are instructed to indicate the color of the word. Typically, addicted individuals, but not controls, are slower to color-name drug-related words than neutral words (Field & Cox, 2008). This difference in processing times is known as the addiction Stroop effect. Processing of the drug-related word interferes with the color-naming task in some way that response times are slower to color-name drug-related words. One advantage of the addiction Stroop is that it is simple and can be administered on both stand-alone computers as well as hand held devices such as PDAs or smartphones. This will be described in more detail later.
The visual probe or dot probe task is another common task used to assess attentional bias. This task was first developed to measure attentional bias resulting from emotional disorders such as anxiety (MacLeod et al., 1986) but can be used to measure attentional bias in addiction (Ehrman et al., 2002). To assess attentional bias in addiction, drug-related and neutral words (or pictures) are presented side-by-side on a computer screen, typically for 500 ms. The two pictures are then replaced by a single dot appearing in a location previously occupied by one of the pictures. Participants are required to indicate the location of the dot as quickly and as accurately as possible by pressing one of two response keys. Typically, addicted individuals are faster to indicate the location of a dot that replaces a drug-related stimulus than a neutral stimulus. This difference in response times is the attentional bias measure derived from the VP task (Ehrman et al., 2002). For example, in one of the earliest studies in smokers, current smokers exhibited an attentional bias towards smoking cues, whereas non-smokers did not (Ehrman et al., 2002). As noted later, the visual probe task has also been modified for the use of attentional retraining (attentional bias modification).
