Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 http://www.substanceabusepolicy.com/content/6/1/17

REVIEW Open Access Relapse prevention for addictive behaviors Christian S Hendershot1,2*, Katie Witkiewitz3, William H George4 and G Alan Marlatt4

Abstract The Relapse Prevention (RP) model has been a mainstay of addictions theory and treatment since its introduction three decades ago. This paper provides an overview and update of RP for addictive behaviors with a focus on developments over the last decade (2000-2010). Major treatment outcome studies and meta-analyses are summarized, as are selected empirical findings relevant to the tenets of the RP model. Notable advances in RP in the last decade include the introduction of a reformulated cognitive-behavioral model of relapse, the application of advanced statistical methods to model relapse in large randomized trials, and the development of mindfulness- based relapse prevention. We also review the emergent literature on genetic correlates of relapse following pharmacological and behavioral treatments. The continued influence of RP is evidenced by its integration in most cognitive-behavioral substance use interventions. However, the tendency to subsume RP within other treatment modalities has posed a barrier to systematic evaluation of the RP model. Overall, RP remains an influential cognitive-behavioral framework that can inform both theoretical and clinical approaches to understanding and facilitating behavior change. Keywords: Alcohol, cognitive-behavioral skills training, continuing care, drug use, psychosocial intervention, sub- stance use treatment

Introduction historical and theoretical foundations of the RP model Relapse poses a fundamental barrier to the treatment of and a brief summary of clinical intervention strategies. addictive behaviors by representing the modal outcome Next, we review the major theoretical, methodological of behavior change efforts [1-3]. For instance, twelve- and applied developments related to RP in the last dec- month relapse rates following alcohol or tobacco cessa- ade. Specific emphasis is placed on the reformulated tion attempts generally range from 80-95% [1,4] and evi- cognitive-behavioral model of relapse [8] as a basis for dence suggests comparable relapse trajectories across hypothesizing and studying dynamic aspects of the various classes of substance use [1,5,6]. Preventing relapse process. In reviewing empirical findings we focus relapse or minimizing its extent is therefore a prerequi- on major treatment outcome studies, meta-analyses, and site for any attempt to facilitate successful, long-term selected results that coincide with underlying tenets of changes in addictive behaviors. the RP model. We conclude by noting critiques of the Relapse prevention (RP) is a tertiary intervention strat- RP model and summarizing current and future direc- egy for reducing the likelihood and severity of relapse tions in studying and preventing relapse. following the cessation or reduction of problematic This paper extends recent reviews of the RP literature behaviors. Three decades since its introduction [7], the [1,8-10] in several ways. Most notably, we provide a RP model remains an influential cognitive-behavioral recent update of the RP literature by focusing primarily approach in the treatment and study of addictions. The on studies conducted within the last decade. We also aim of this paper is to provide readers with an update provide updated reviews of research areas that have on empirical and applied developments related to RP, seen notable growth in the last few years; in particular, with a primary focus on events spanning the last decade the application of advanced statistical modeling techni- (2000-2010). We begin with a concise overview of the ques to large treatment outcome datasets and the devel- opment of mindfulness-based relapse prevention. * Correspondence: [email protected] Additionally, we review the nascent but rapidly growing 1Centre for Addiction and Mental Health, 33 Russell St., Toronto, ON, M5S 2S1, Canada literature on genetic predictors of relapse following sub- Full list of author information is available at the end of the article stance use interventions. In focusing exclusively on

© 2011 Hendershot et al; licensee BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 2 of 17 http://www.substanceabusepolicy.com/content/6/1/17

addictive behaviors (for which the RP model was initially and a set of treatment strategies designed to limit conceived) we forego a discussion of RP as it relates to relapse likelihood and severity. Because detailed various other behavioral domains (e.g., sexual offending, accounts of the model’s historical background and theo- depression, diet and exercise) and refer readers to other retical underpinnings have been published elsewhere (e. sources for updates on the growing range of RP applica- g., [16,22,23]), we limit the current discussion to a con- tions [8,11]. cise review of the model’s history, core concepts and clinical applications. Definitions of relapse and relapse prevention Based on the cognitive-behavioral model of relapse, The terms “relapse” and “relapse prevention” have seen RP was initially conceived as an outgrowth and augmen- evolving definitions, complicating efforts to review and tation of traditional behavioral approaches to studying evaluate the relevant literature. Definitions of relapse are and treating addictions. The evolution of cognitive-beha- varied, ranging from a dichotomous treatment outcome vioral theories of substance use brought notable changes to an ongoing, transitional process [8,12,13]. Overall, a in the conceptualization of relapse, many of which large volume of research has yielded no consensus departed from traditional (e.g., disease-based) models of operational definition of the term [14,15]. For present addiction. For instance, whereas traditional models often purposes we define relapse as a setback that occurs dur- attribute relapse to endogenous factors like cravings or ing the behavior change process, such that progress withdrawal–construed as symptoms of an underlying toward the initiation or maintenance of a behavior disease state–cognitive-behavioral theories emphasize change goal (e.g., abstinence from drug use) is inter- contextual factors (e.g., environmental stimuli and cog- rupted by a reversion to the target behavior. We also nitive processes) as proximal relapse antecedents. Cogni- take the perspective that relapse is best conceptualized tive-behavioral theories also diverged from disease as a dynamic, ongoing process rather than a discrete or models in rejecting the notion of relapse as a dichoto- terminal event (e.g., [1,8,10]). mous outcome. Rather than being viewed as a state or Definitions of RP have also evolved considerably, due endpoint signaling treatment failure, relapse is consid- largely to the increasingly broad adoption of RP ered a fluctuating process that begins prior to and approaches in various treatment contexts. Though the extends beyond the return to the target behavior [8,24]. phrase “relapse prevention” was initially coined to denote From this standpoint, an initial return to the target a specific clinical intervention program [7,16], RP strate- behavior after a period of volitional abstinence (a lapse) gies are now integral to most psychosocial treatments for is seen not as a dead end, but as a fork in the road. substance use [17], including many of the most widely dis- While a lapse might prompt a full-blown relapse, seminated interventions (e.g., [18-20]). The National Reg- another possible outcome is that the problem behavior istry of Evidence-based Programs and Practices, is corrected and the desired behavior re-instantiated–an maintained by the U.S. Substance Abuse and Mental event referred to as prolapse. A critical implication is Health Services Administration (SAMHSA), includes list- that rather than signaling a failure in the behavior ings for numerous empirically supported interventions change process, lapses can be considered temporary set- with “relapse prevention” as a descriptor or primary treat- backs that present opportunities for new learning to ment objective (http://www.nrepp.samhsa.gov). Thus, RP occur. In viewing relapse as a common (albeit undesir- has in many ways evolved into an umbrella term encom- able) event, emphasizing contextual antecedents over passing most skills-based treatments that emphasize cog- internal causes, and distinguishing relapse from treat- nitive-behavioral skills building and coping responses. ment failure, the RP model introduced a comprehensive, While attesting to the broad influence of the RP model, flexible and optimistic alternative to traditional the diffuse application of RP approaches also tends to approaches. complicate efforts to define RP-based treatments and eval- Marlatt’s original RP model is depicted in Figure 1. A uate their overall efficacy (e.g., [21]). In the present review basic assumption is that relapse events are immediately we emphasize Marlatt’s RP model [7,16] and its more preceded by a high-risk situation, broadly defined as any recent iteration [8] when discussing the theoretical basis context that confers vulnerability for engaging in the of RP. By necessity, our literature review also includes stu- target behavior. Examples of high-risk contexts include dies that do not explicitly espouse the RP model, but that emotional or cognitive states (e.g., negative affect, are relevant nonetheless to its predictions. diminished self-efficacy), environmental contingencies (e.g., conditioned drug cues), or physiological states (e. Marlatt’s relapse prevention model: Historical foundations g., acute withdrawal). Although some high-risk situa- and overview tions appear nearly universal across addictive behaviors The RP model developed by Marlatt [7,16] provides (e.g., negative affect; [25]), high-risk situations are likely both a conceptual framework for understanding relapse to vary across behaviors, across individuals, and within Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 3 of 17 http://www.substanceabusepolicy.com/content/6/1/17

Figure 1 Original cognitive-behavioral model of relapse (Marlatt & Gordon, 1985) the same individual over time [10]. Whether a high-risk expectancies, readiness to change, and concomitant fac- situation culminates in a lapse depends largely on the tors that could complicate treatment (e.g., comorbid dis- individual’s capacity to enact an effective coping orders, neuropsychological deficits). Using high-risk response–defined as any cognitive or behavioral com- situations as a starting point, the clinician works back- pensatory strategy that reduces the likelihood of lapsing. ward to identify immediate precipitants and distal life- Central to the RP model is the role of cognitive factors style factors related to relapse, and forward to evaluate in determining relapse liability. For example, successful coping responses [16,24]. Ideally, this approach helps navigation of high-risk situations may increase self-effi- clients to recognize high-risk situations as discriminative cacy (one’s perceived capacity to cope with an impending stimuli signaling relapse risk, as well as to identify cog- situation or task; [26]), in turn decreasing relapse prob- nitive and behavioral strategies to obviate these situa- ability. Conversely, a return to the target behavior can tions or minimize their impact. Examples of specific undermine self-efficacy, increasing the risk of future intervention strategies include enhancing self-efficacy (e. lapses. Outcome expectancies (anticipated effects of sub- g., by setting achievable behavioral goals) and eliminat- stance use; [27]) also figure prominently in the RP model. ing myths and placebo effects (e.g., by challenging mis- Additionally, attitudes or beliefs about the causes and perceptions about the effects of substance use). meaning of a lapse may influence whether a full relapse The client’s appraisal of lapses also serves as a pivotal ensues. Viewing a lapse as a personal failure may lead to intervention point in that these reactions can determine feelings of guilt and abandonment of the behavior change whether a lapse escalates or desists. Establishing lapse goal [24]. This reaction, termed the Abstinence Violation management plans can aid the client in self-correcting Effect (AVE; [16]), is considered more likely when one soon after a slip, and cognitive restructuring can help holds a dichotomous view of relapse and/or neglects to clients to re-frame the meaning of the event and mini- consider situational explanations for lapsing. In sum, the mize the AVE [24]. A final emphasis in the RP approach RP framework emphasizes high-risk contexts, coping is the global intervention of lifestyle balancing, designed responses, self-efficacy, affect, expectancies and the AVE to target more pervasive factors that can function as as primary relapse antecedents. relapse antecedents. For example, clients can be encour- Implicit in the RP approach is that the initiation and aged to increase their engagement in rewarding or maintenance of behavior change represent separate pro- stress-reducing activities into their daily routine. Success cesses governed by unique contingencies [12,28]. Thus, in these areas may enhance self-efficacy, in turn redu- specific cognitive and behavioral strategies are often cing relapse risk. Overall, the RP model is characterized necessary to maintain initial treatment gains and mini- by a highly ideographic treatment approach, a contrast mize relapse likelihood following initial behavior change. to the “one size fits all” approach typical of certain tradi- RP strategies fall into two broad categories: specific tional treatments. Moreover, an emphasis on post-treat- intervention techniques, often designed to help the ment maintenance renders RP a useful adjunct to patient anticipate and cope with high-risk situations, various treatment modalities (e.g., cognitive-behavioral, and global self-control approaches, intended to reduce twelve step programs, pharmacotherapy), irrespective of relapse risk by promoting positive lifestyle change. An the strategies used to enact initial behavior change. essential starting point in treatment is a thorough assessment of the client’s substance use patterns, high- Developments in Relapse Prevention: 2000-2010 risk situations and coping skills. Other important assess- The last decade has seen numerous developments in the ment targets include the client’s self-efficacy, outcome RP literature, including the publication of Relapse Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 4 of 17 http://www.substanceabusepolicy.com/content/6/1/17

Prevention, Second Edition [29] and its companion text, are examples of distal variables that could influence Assessment of Addictive Behaviors, Second Edition [30]. relapse liability a priori. Tonic processes also include The following sections provide an overview of major cognitive factors that show relative stability over time, theoretical, empirical and applied advances related to RP such as drug-related outcome expectancies, global self- over the last decade. efficacy, and personal beliefs about abstinence or relapse. Whereas tonic processes may dictate initial sus- The reformulated cognitive-behavioral model of relapse ceptibility to relapse, its occurrence is determined lar- Efforts to develop, test and refine theoretical models are gely by phasic responses–proximal or transient factors critical to enhancing the understanding and prevention that serve to actuate (or prevent) a lapse. Phasic of relapse [1,2,14]. A major development in this respect responses include cognitive and affective processes that was the reformulation of Marlatt’scognitive-behavioral can fluctuate across time and contexts–such as urges/ relapse model to place greater emphasis on dynamic cravings, mood, or transient changes in outcome expec- relapse processes [8]. Whereas most theories presume tancies, self-efficacy, or motivation. Additionally, linear relationships among constructs, the reformulated momentary coping responses can serve as phasic events model (Figure 2) views relapse as a complex, nonlinear that may determine whether a high-risk situation culmi- process in which various factors act jointly and interac- nates in a lapse. Substance use and its immediate conse- tively to affect relapse timing and severity. Similar to the quences (e.g., impaired decision-making, the AVE) are original RP model, the dynamic model centers on the additional phasic processes that are set into motion high-risk situation. Against this backdrop, both tonic once a lapse occurs. Thus, whereas tonic processes can (stable) and phasic (transient) influences interact to determine who is vulnerable for relapse, phasic pro- determine relapse likelihood. Tonic processes include cesses determine when relapse occurs [8,31]. distal risks–stable background factors that determine an A key feature of the dynamic model is its emphasis on individual’s “set point” or initial threshold for relapse the complex interplay between tonic and phasic pro- [8,31]. Personality, genetic or familial risk factors, drug cesses. As indicated in Figure 2, distal risks may influ- sensitivity/metabolism and physical withdrawal profiles ence relapse either directly or indirectly (via phasic

Figure 2 Revised cognitive-behavioral model of relapse (Witkiewitz & Marlatt, 2004) Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 5 of 17 http://www.substanceabusepolicy.com/content/6/1/17

processes). The model also predicts feedback loops Systematic reviews and large-scale treatment among hypothesized constructs. For instance, the return outcome studies to substance use can have reciprocal effects on the same The first comprehensive review of RP treatment out- cognitive or affective factors (motivation, mood, self-effi- come studies was Carroll’s [35] descriptive account of cacy) that contributed to the lapse. Lapses may also 24 interventions focusing on substance use. This review evoke physiological (e.g., alleviation of withdrawal) and/ found consistent support for the superiority of RP over or cognitive (e.g., the AVE) responses that in turn deter- no treatment, inconsistent support for its superiority mine whether use escalates or desists. The dynamic over discussion control conditions, and consistent sup- model further emphasizes the importance of nonlinear port that RP was equally efficacious to other active relationships and timing/sequencing of events. For treatments. Carroll concluded that RP, though not con- instance, in a high-risk context, a slight and momentary sistently superior to other active treatments, showed drop in self-efficacy could have a disproportionate particular promise in three areas: reducing relapse sever- impact on other relapse antecedents (negative affect, ity, enhancing durability of treatment gains, and match- expectancies) [8]. Furthermore, the strength of proximal ing treatment strategies to client characteristics. RP also influences on relapse may vary based on distal risk fac- showed delayed emergence effects in some studies, sug- tors, with these relationships becoming increasingly gesting that it may outperform other treatments during nonlinear as distal risk increases [31]. For example, one the maintenance stage of behavior change [35]. couldimagineasituationwherebyaclientwhoisrela- Subsequently, a meta-analysis evaluated 26 RP treat- tively committed to abstinence from alcohol encounters ment outcome studies totaling 9,504 participants [36]. a neighbor who invites the client into his home for a The authors examined two primary outcomes (substance drink. Feeling somewhat uncomfortable with the offer use and psychosocial functioning) and several treatment the client might experience a slight decrease in self-effi- moderators. Effect sizes indicated that RP was generally cacy, which cascades into positive outcome expectancies successful in reducing substance use (r =.14)and about the potential effects of having a drink as well as improving psychosocial functioning (r = .48), consistent feelings of shame or guilt about saying no to his neigh- with its purpose as both a specific and global interven- bor’soffer.Importantly,thisclientmightnothaveever tion approach. Moderation analyses suggested that RP considered such an invitation as a high-risk situation, was consistently efficacious across treatment modalities yet various contextual factors may interact to predict a (individual vs. group) and settings (inpatient vs. outpati- lapse. ent). RP was most effective for reducing alcohol and The dynamic model of relapse assumes that relapse polysubstance use and less effective for tobacco and can take the form of sudden and unexpected returns to cocaine use–a contrast to Carroll’s[35]findingofcom- the target behavior. This concurs not only with clinical parable efficacy across drug classes. In addition, RP was observations, but also with contemporary learning mod- more effective when delivered in conjunction with phar- els stipulating that recently modified behavior is inher- macotherapy, when compared to wait-list (vs. active) ently unstable and easily swayed by context [32]. While comparison conditions, and when outcomes were maintaining its footing in cognitive-behavioral theory, assessed soon after treatment. Though some findings the revised model also draws from nonlinear dynamical were considered tentative due to sample sizes, the systems theory (NDST) and catastrophe theory, both authors concluded that RP was broadly efficacious [36]. approaches for understanding the operation of complex McCrady [37] conducted a comprehensive review of systems [10,33]. Detailed discussions of relapse in rela- 62 alcohol treatment outcome studies comprising 13 tion to NDST and catastrophe theory are available else- psychosocial approaches. Two approaches–RP and brief where [10,31,34]. intervention–qualified as empirically validated treat- ments based on established criteria. Interestingly, Miller Empirical findings relevant to the RP model and Wilbourne’s [21] review of clinical trials, which The empirical literature on relapse in addictions has evaluated the efficacy of 46 different alcohol treatments, grown substantially over the past decade. Because the ranked “relapse prevention” as 35th out of 46 treatments volume and scope of this work precludes an exhaustive based on methodological quality and treatment effect review, the following section summarizes a select body sizes. However, many of the treatments ranked in the of findings reflective of the literature and relevant to RP top 10 (including brief interventions, social skills train- theory. The studies reviewed focus primarily on alcohol ing, community reinforcement, behavior contracting, and tobacco cessation, however, it should be noted that behavioral marital therapy, and self-monitoring) incor- RP principles have been applied to an increasing range porate RP components. These two reviews highlighted of addictive behaviors [10,11]. the increasing difficulty of classifying interventions as Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 6 of 17 http://www.substanceabusepolicy.com/content/6/1/17

specifically constituting RP, given that many treatments there was general support for the efficacy of pharmaco- for substance use disorders (e.g., cognitive behavioral logical treatments [40]. treatment (CBT)) are based on the cognitive behavioral Recently, Magill and Ray [41] conducted a meta-analy- model of relapse developed for RP [16]. One of the key sis of 53 controlled trials of CBT for substance use dis- distinctions between CBT and RP in the field is that the orders. As noted by the authors, the CBT studies term “CBT” is more often used to describe stand-alone evaluated in their review were based primarily on the primary treatments that are based on the cognitive- RP model [29]. Overall, the results were consistent with behavioral model, whereas RP is more often used to the review conducted by Irvin and colleagues, in that describe aftercare treatment. Given that CBT is often the authors concluded that 58% of individuals who used as a stand-alone treatment it may include addi- received CBT had better outcomes than those in com- tional components that are not always provided in RP. parison conditions. In contrast with the findings of Irvin For example, the CBT intervention developed in Project and colleagues [36], Magill and Ray [41] found that MATCH[18](describedbelow)equatedtoRPwith CBT was most effective for individuals with marijuana respect to the core sessions, but it also included elective use disorders. sessions that are not typically a focus in RP (e.g., job- seeking skills, family involvement). Recent findings in support of RP model An increasing number of large-scale trials have components allowed for statistically powerful evaluations of psycho- The following section reviews selected empirical findings social interventions for alcohol use. Project MATCH that support or coincide with tenets of the RP model. [18] evaluated the efficacy of three interventions–Moti- Sections are organized in accordance with major model vational Enhancement Therapy (MET), Twelve-Step constructs. Because the scope of this literature precludes Facilitation (TSF), and Cognitive Behavioral Therapy an exhaustive review, we highlight select findings that (CBT)–for treating alcohol dependence. The CBT inter- are relevant to the main tenets of the RP model, in par- vention was a skills-based treatment containing elements ticular those that coincide with predictions of the refor- of RP. Spanning nine data collection sites and following mulated model of relapse. over 1700 participants for up to three years, Project MATCH was the largest psychotherapy trial conducted Self-efficacy to that point. Multiple matching hypotheses were pro- Self-efficacy (SE), the perceived ability to enact a given posed in evaluating differential treatment efficacy as a behavior in a specified context [26], is a principal deter- function of theoretically relevant client attributes. Pri- minant of health behavior according to social-cognitive mary analyses supported only one of sixteen matching theories. In fact, some theories view SE as the final com- hypotheses: outpatients lower in psychiatric severity mon pathway to relapse [42]. Although SE is proposed fared better in TSF than in CBT during the year follow- as a fluctuating and dynamic construct [26], most stu- ing treatment [18]. Although primary analyses provided dies rely on static measures of SE, preventing evaluation relatively little support for tailoring alcohol treatments of within-person changes over time or contexts [43]. based on specific client attributes, matching effects have Shiffman, Gwaltney and colleagues have used ecological been identified in subsequent analyses (described in momentary assessment (EMA; [44]) to examine tem- more detail later). poral variations in SE in relation to smoking relapse. Since 2005 an ongoing Cochrane review has evaluated Findings from these studies suggested that participants’ RP for smoking cessation [38,39]. As of 2009, meta-ana- SE was lower on the day before a lapse, and that lower lyses had found no support for the efficacy of skills- SE in the days following a lapse in turn predicted pro- based RP approaches in preventing relapse to smoking gression to relapse [43,45]. One study [46] reported [38]. However, a recent re-analysis of these trials yielded increases in daily SE during abstinent intervals, perhaps different results [40]. The re-analysis stratified beha- indicating mounting confidence as treatment goals were vioral interventions based on specific intervention con- maintained [45]. tent while also imposing stricter analytic criteria Inthefirststudytoexaminerelapseinrelationto regarding the length of follow-up assessments. In these phasic changes in SE [46], researchers reported results analyses, CBT/RP-based self-help interventions showed that appear consistent with the dynamic model of a significant overall effect in increasing long-term absti- relapse. During a smoking cessation attempt, partici- nence (pooled OR: 1.52, 95% CI: 1.15 - 2.01, based on 3 pants reported on SE, negative affect and urges at ran- studies) and group counseling showed significant short- dom intervals. Findings indicated nonlinear relationships term efficacy (pooled OR: 2.55, 95% CI: 1.58 - 4.11, between SE and urges, such that momentary SE based on 2 studies). There was limited evidence for the decreased linearly as urges increased but dropped efficacy of other specific behavioral treatments, although abruptly as urges peaked. Moreover, this finding Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 7 of 17 http://www.substanceabusepolicy.com/content/6/1/17

appeared attributable to individual differences in base- the first study to examine how daily fluctuations in line (tonic) levels of SE. When urge and negative affect expectancies predict relapse [45], researchers assessed were low, individuals with low, intermediate or high positive outcome expectancies for smoking (POEs) baseline SE were similar in their momentary SE ratings. among participants during a tobacco cessation attempt. However, these groups’ momentary ratings diverged sig- Lower POEs on the quit day were associated with nificantly at high levels of urges and negative affect, greater abstinence likelihood, and POEs decreased in the such that those with low baseline SE had large drops in days following the quit day. Lapses were associated with momentary SE in the face of increasingly challenging higher POEs on the preceding day, and in the days fol- situations. These findings support that higher distal risk lowing a lapse those who avoided a full relapse showed can result in bifurcations (divergent patterns) of beha- decreases in POEs whereas those who relapsed did not vior as the level of proximal risk factors increase, consis- [45]. These results suggest that outcome expectancies tent with predictions from nonlinear dynamic systems might play a role in predicting relapse as both a tonic theory [31]. and phasic risk factor. A recent meta-analysis evaluated the association of SE Expectancy research has recently started examining with smoking relapse [47]. The review included 54 stu- the influences of implicit cognitive processes, generally dies that assessed prospective associations of SE and defined as those operating automatically or outside con- smoking during a quit attempt. A major finding con- scious awareness [54,55]. Recent reviews provide a con- cerned differential effect sizes based on the timing of SE vincing rationale for the putative role of implicit assessments: the negative association of SE with likeli- processes in addictive behaviors and relapse [54,56,57]. hood of future smoking represented a small effect (d = Implicit measures of alcohol-related cognitions can dis- -.21) when SE was assessed prior to the quit attempt, criminate among light and heavy drinkers [58] and pre- but a medium effect (d = -.47) when SE was assessed dict drinking above and beyond explicit measures [59]. after the quit day. The authors concluded that, given the One study found that smokers’ attentional bias to centrality of SE to most cognitive-behavioral models of tobacco cues predicted early lapses during a quit relapse, the association of SE with cessation was weaker attempt, but this relationship was not evident among than would be expected (i.e., SE accounted for roughly people receiving nicotine replacement therapy, who 2% of the variance in treatment outcome following showed reduced attention to cues [60]. initial abstinence). The findings also suggested that SE Initial evidence suggests that implicit measures of should ideally be measured after the cessation attempt, expectancies are correlated with relapse outcomes, as and that controlling for concurrent smoking is critical demonstrated in one study of heroin users [61]. In when examining SE in relation to prospective relapse another recent study, researchers trained participants in [47]. Finally, in analyses from a cross-national study of attentional bias modification (ABM) during inpatient the natural history of smoking cessation, researchers treatment for alcohol dependence and measured relapse examined self-efficacy in relation to relapse rates across over the course of three months post-treatment [62]. an extended time period [48,49]. Results indicated that Relative to a control condition, ABM resulted in signifi- self-efficacy increased with cumulative abstinence and cantly improved ability to disengage from alcohol- correlated negatively with urges, consistent with RP the- related stimuli during attentional bias tasks. While inci- ory. Also, higher self-efficacy consistently predicted dence of relapse did not differ between groups, the lower relapse rates across time and partly mediated the ABM group showed a significantly longer time to first association of perceived benefits of smoking with relapse heavy drinking day compared to the control group. events [48,49]. Additionally, the intervention had no effect on subjec- tive measures of craving, suggesting the possibility that Outcome expectancies intervention effects may have been specific to implicit Outcome expectancies (anticipated outcomes of a given cognitive processes [62]. Overall, research on implicit behavior or situation) are central to the RP model and cognitions stands to enhance understanding of dynamic have been studied extensively in the domain of alcohol relapse processes and could ultimately aid in predicting use [27]. In theory, expectancies are shaped by various lapses during high-risk situations. tonic risk factors (e.g., environment, culture, personality, genetics) and mediate these antecedent influences on Withdrawal drinking [27]. Research supports that expectancies could Withdrawal tendencies can develop early in the course partly mediate influences such as personality factors of addiction [25] and symptom profiles can vary based [50], genetic variations [51,52], and negative affect [53] on stable intra-individual factors [63], suggesting the on drinking. Outcome expectancies could also be involvement of tonic processes. Despite serving as a involved as a phasic response to situational factors. In chief diagnostic criterion, withdrawal often does not Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 8 of 17 http://www.substanceabusepolicy.com/content/6/1/17

predict relapse, perhaps partly explaining its de-empha- who reported higher negative affect or increased nega- sis in contemporary motivational models of addiction tive affect over time had the highest probability of heavy [64]. However, recent studies show that withdrawal pro- and frequent drinking following treatment, and had a files are complex, multi-faceted and idiosyncratic, and near-zero probability of transitioning to moderate drink- that in the context of fine-grained analyses withdrawal ing. Heavier and more frequent alcohol use predicted a indeed can predict relapse [64,65]. Such findings have greater probability of high negative affect and increased contributed to renewed interest in negative reinforce- negative affect over time. ment models of drug use [63]. KnowledgeabouttheroleofNAindrinkingbehavior Although withdrawal is usually viewed as a physiologi- has benefited from daily process studies in which parti- cal process, recent theory emphasizes the importance of cipants provide regular reports of mood and drinking. behavioral withdrawal processes [66]. Whereas physiolo- Such studies have shown that both positive and negative gical withdrawal symptoms tend to abate in the days or moods show close temporal links to alcohol use [73]. weeks following drug cessation, the unavailability of a One study [74] found evidence suggesting a feedback conditioned behavioral coping response (e.g., the ritual cycle of mood and drinking whereby elevated daily levels of drug administration) may leave the former user ill- of NA predicted alcohol use, which in turn predicted equipped to cope with ongoing stressors, thus exacer- spikes in NA. These findings were moderated by gender, bating and/or prolonging symptoms [66]. Current theory social context, and time of week. Other studies have and research indicate that physiological components of similarly found that relationships between daily events drug withdrawal may be motivationally inert, with the and/or mood and drinking can vary based on intraindi- core motivational constituent of withdrawal being nega- vidual or situational factors [73], suggesting dynamic tive affect [25,66]. Thus, examining withdrawal in rela- interplay between these influences. tion to relapse may only prove useful to the extent that negative affect is assessed adequately [64]. Self-control and coping responses Strengthening coping skills is a goal of virtually all cog- Negative affect nitive-behavioral interventions for substance use [75]. A large literature attests to the role of negative affect Several studies have used EMA to examine coping (NA) in the etiology and maintenance of addictive beha- responses in real time. One study [76] found that viors. NA is consistently cited as a relapse trigger in ret- momentary coping differentiated smoking lapses from rospective reports (e.g., [67,68]), although participants temptations, such that coping responses were reported might sometimes misattribute lapses to negative mood in 91% of successful resists vs. 24% of lapses. Shiffman states[15]. In one study, individuals who were unable to and colleagues [68] found that restorative coping follow- sustain a smoking cessation attempt for more than 24 ing a smoking lapse decreased the likelihood of a second hours (compared to those with a sustained quit attempt) lapse the same day. Exactly how coping responses reported greater depressive symptoms and NA in reduce the likelihood of lapsing remains unclear. One response to stress and displayed less perseverance dur- study found that momentary coping reduced urges ing experimental stress inductions [69]. Supporting the among smokers, suggesting a possible mechanism [76]. dynamic influence of NA on relapse, Shiffman and Some studies find that the number of coping responses Waters [70] found that smoking lapses were not asso- is more predictive of lapses than the specific type of ciated with NA in the preceding days, but were asso- coping used [76,77]. However, despite findings that cop- ciated with rising NA in the hours leading up to a lapse. ing can prevent lapses there is scant evidence to show Evidence further suggests that negative affect can pro- that skills-based interventions in fact lead to improved mote positive outcome expectancies [53] or undermine coping [75]. situational self-efficacy [71], outcomes which could in Some researchers propose that the self-control turn promote a lapse. Moreover, Baker and colleagues required to maintain behavior change strains motiva- propose that high levels of negative affect can interfere tional resources, and that this “fatigue” can undermine with controlled cognitive processes, such that adaptive subsequent self-control efforts [78]. Consistent with this coping and decision-making may be undermined as idea, EMA studies have shown that social drinkers negative affect peaks [25]. Witkiewitz and Villarroel [72] report greater alcohol consumption and violations of found that drinking rates following treatment were sig- self-imposed drinking limits on days when self-control nificantly associated with current and prior changes in demands are high [79]. Limit violations were predictive negative affect and changes in negative affect were sig- of responses consistent with the AVE the following day, nificantly associated with current and prior changes in and greater distress about violations in turn predicted drinking state (effect size range = 0.13 (small) to 0.33 greater drinking [80]. Findings also suggested that these (medium)). Overall, the results showed that individuals relationships varied based on individual differences, Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 9 of 17 http://www.substanceabusepolicy.com/content/6/1/17

suggesting the interplay of static and dynamic factors in hypothesis, whereby females with low baseline motiva- AVE responses. Evidence further suggests that practicing tion and males with lower levels of alcohol dependence routine acts of self-control can reduce short-term inci- and low baseline motivation who received MET as an dence of relapse. For instance, Muraven [81] conducted aftercare treatment had better outcomes than those who a study in which participants were randomly assigned to were assigned to receive CBT as an aftercare treatment. practice small acts self-control acts on a daily basis for two weeks prior to a smoking cessation attempt. Com- Genetic influences on treatment response and relapse pared to a control group, those who practiced self-con- The last decade has seen a marked increase in the num- trol showed significantly longer time until relapse in the ber of human molecular genetic studies in medical and following month. behavioral research, due largely to rapid technological advances in genotyping platforms, decreasing cost of Emerging topics in relapse and relapse molecular analyses, and the advent of genome-wide prevention association studies (GWAS). Not surprisingly, molecular Using nonlinear methods to model relapse genetic approaches have increasingly been incorporated A key contribution of the reformulated relapse model is in treatment outcome studies, allowing novel opportu- to highlight the need for non-traditional assessment and nities to study biological influences on relapse. Given analytic approaches to better understand relapse. Most the rapid growth in this area, we allocate a portion of studies of relapse rely on statistical methods that assume this review to discussing initial evidence for genetic continuous linear relationships, but these methods may associations with relapse. Specifically, we focus on be inadequate for studying a behavior characterized by recent, representative findings from studies evaluating discontinuity and abrupt changes [33]. Consistent with candidate single nucleotide polymorphisms (SNPs) as the tenets of the reformulated RP model, several studies moderators of response to substance use interventions. suggest advantages of nonlinear statistical approaches It is important to note that these studies were not for studying relapse. designed to evaluate specific components of the RP In one study, researchers used catastrophe models to model, nor do these studies explicitly espouse the RP examine proximal and distal predictors of post-treat- model. Also, many studies have focused solely on phar- ment drinking among individuals with alcohol use disor- macological interventions, and are therefore not directly ders [31]. Catastrophe models accounted for more than related to the RP model. However, we review these find- double the amount of variance in drinking than that ings in order to illustrate the scope of initial efforts to predicted by linear models. Similar results have been include genetic predictors in treatment studies that found using the much larger Project MATCH dataset examine relapse as a clinical outcome. These findings [33]. Two additional recent analyses of the MATCH may be informative for researchers who wish to incor- dataset showed that nonlinear approaches can detect porate genetic variables in future studies of relapse and processes that may go unobserved in the context of lin- relapse prevention. ear models. Witkiewitz and colleagues [34,82] used cata- Broadly speaking, there are at least three primary con- strophe modeling and latent growth mixture modeling texts in which genetic variation could influence liability to re-assess two of the matching hypotheses that were for relapse during or following treatment. First, in the not supported in the original study–that individuals low context of pharmacotherapy interventions, relevant in baseline self-efficacy would respond more favorably genetic variations can impact drug pharmacokinetics or to cognitive-behavioral therapy (CBT) than motivational pharmacodynamics, thereby moderating treatment enhancement therapy (MET) and that individuals low in response (pharmacogenetics). Second, the likelihood of baseline motivation would respond more favorably to abstinence following a behavioral or pharmacological MET than CBT [18]. In the first study [34], catastrophe intervention can be moderated by genetic influences on models provided the best fit to the data, and latent metabolic processes, receptor activity/expression, and/or growth analyses confirmed the predicted interaction: fre- incentive value specific to the addictive substance in quent drinkers with low initial self-efficacy had better question. For instance, SNPs with functional implica- outcomes in CBT than in MET, while those high in tions for relevant neurotransmitter or metabolic path- self-efficacy fared better in MET. Similarly, a second ways can influence the reward value of marijuana (e.g., study [82] found that individuals in the outpatient arm FAAH; CNR1); nicotine (e.g., CYP2A6, CHRNB2, of Project MATCH with low motivation to change at CHRNA4); and alcohol (ALDH2, ADH1B), while others baseline who were assigned to MET had better out- show potential for influencing the incentive value of comes than those assigned to CBT. The authors also multiple drugs (e.g., ANKK1; DRD4; OPRM1). Third, found a treatment by gender by alcohol dependence variants implicated in broad traits relevant for addictive severity interaction in support of the matching behaviors–for instance, executive cognitive functioning Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 10 of 17 http://www.substanceabusepolicy.com/content/6/1/17

(e.g., COMT) or externalizing traits (e.g., GABRA2, were not evident in participants receiving NTX and DRD4)–could influence relapse proneness via general CBI). A smaller placebo controlled study has also found neurobehavioral mechanisms, irrespective of drug class evidence for better responses to NTX among Asp40 car- or treatment modality. As summarized below, prelimin- riers [94]. The Asp40 variant has further been linked to ary empirical support exists for each of these intermediate phenotypes that could influence relapse possibilities. proneness, including hedonic responses to alcohol [95], Genetic influences on relapse have been studied most increased neural responses to alcohol primes [96], extensively in the context of pharmacogenetics, with the greater craving in response to alcohol use [97] and bulk of studies focusing on nicotine dependence (for increased dopamine release in the ventral striatum dur- recent reviews see [83,84]). Several candidate poly- ing alcohol challenge [98]. One study found that the morphisms have been examined in response to smoking Asp40 allele predicted cue-elicited craving among indivi- cessation treatments, especially nicotine replacement duals low in baseline craving but not those high in therapy (NRT) and bupropion [84]. The catechol-O- initial craving, suggesting that tonic craving could inter- methyltransferase (COMT) Val158Met polymorphism, act with genotype to predict phasic responses to drug established as predicting variability in prefrontal dopa- cues [97]. mine levels, has been evaluated in relation to smoking Findings concerning possible genetic moderators of cessation in several studies. Independent trials of NRT response to acamprosate have been reported [99], but have found cessation rates to differ based on COMT are preliminary. Additionally, other findings suggest the genotype [85-87]. A polymorphism in the nicotinic acet- influence of a DRD4 variable number of tandem repeats ylcholine ß2 receptor gene (CHRNB2) has been asso- (VNTR) polymorphism on response to olanzapine, a ciated with length of abstinence and withdrawal dopamine antagonist that has been studied as an experi- symptoms during bupropion treatment [88] and with mental treatment for alcohol problems. Olanzapine was relapse rates and ability to quit on the target day during found to reduce alcohol-related craving those with the NRT [89]. One bupropion trial found that DRD2 varia- long-repeat VNTR (DRD4 L), but not individuals with tions predicted withdrawal symptoms, medication the short-repeat version (DRD4 S; [100,101]). Further, a response and time to relapse [90]. In a study of the mu- randomized trial of olanzapine led to significantly opioid receptor (OPRM1) Asn40/Asp40 variant during improved drinking outcomes in DRD4 L but not DRD4 NRT, those with the Asp40 variant had higher rates of S individuals [100]. abstinence and reduced negative affect compared to There is also preliminary evidence for the possibility Asn40 individuals [91]. Additionally, post-hoc analyses of genetic influences on response to psychosocial inter- indicated that Asp40 carriers were more likely to regain ventions, including those incorporating RP strategies. In abstinence following a lapse, suggesting a possible role a secondary analysis of the Project MATCH data, of the genotype in predicting prolapse. researchers evaluated posttreatment drinking outcomes The most promising pharmacogenetic evidence in in relation to a GABRA2 variant previously implicated alcohol interventions concerns the OPRM1 A118G poly- in the risk for alcohol dependence [102]. Analyses morphism as a moderator of clinical response to nal- included MATCH participants of European descent who trexone (NTX). An initial retrospective analysis of NTX provided a genetic sample (n = 812). Those carrying the trials found that OPRM1 influenced treatment response, high-risk GABRA2 allele showed a significantly such that individuals with the Asp40 variant (G allele) increased likelihood of relapse following treatment, receiving NTX had a longer time until the first heavy including a twofold increase in the likelihood of heavy drinking day and were half as likely to relapse compared drinking. Furthermore, GABRA2 interacted with treat- to those homozygous for the Asn40 variant (A allele) ment condition to influence drinking outcomes. Among [92]. This finding was later extended in the COMBINE those with the high-risk genotype, drinking behavior did study, such that G carriers showed a greater proportion not appear to be modified by treatment, with outcomes of days abstinent and a lower proportion of heavy drink- being similar regardless of treatment condition. How- ing days compared in response to NTX versus placebo, ever, treatment differences emerged in the low-risk gen- whereas participants homozygous for the A allele did otype group, such that TSF produced the best not show a significant medication response [93]. More- outcomes, followed by MET [102]. In another psychoso- over, 87.1% of G allele carriers who received NTX were cial treatment study, researchers in Poland examined classified as having a good clinical outcome at study genetic moderators of relapse following inpatient alcohol endpoint, versus 54.5% of Asn40 homozygotes who treatment [103]. Results showed that polymorphisms in received NTX. (Moderating effects of OPRM1 were spe- BDNF (Val66Met) and COMT (Val158Met) significantly cific to participants receiving medication management predicted relapse probability. Overall, evidence for without the cognitive-behavioral intervention [CBI] and genetic moderation effects in psychosocial trials are Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 11 of 17 http://www.substanceabusepolicy.com/content/6/1/17

consistent with the notion that variants with broad [115], whereby clients are taught to view urges as analo- implications for neurotransmitter function, cognitive gous to an ocean wave that rises, crests, and diminishes. function, and/or externalizing traits can potentially Rather than being overwhelmed by the wave, the goal is influence relapse proneness. In the absence of a plausi- to “surf” its crest, attending to thoughts and sensations ble biological mechanism for differential response to as the urge peaks and subsides. specific psychosocial treatments (e.g., MET vs. CBT) as Results of a preliminary nonrandomized trial supported a function of genotype, the most parsimonious interpre- the potential utility of MBRP for reducing substance use. tation of these findings is that some variants will impose In this study incarcerated individuals were offered the greater risk for relapse following any quit attempt, chance to participate in an intensive 10-day course in regardless of treatment availability or modality. Vipassana meditation (VM). Those participating in VM Findings from numerous non-treatment studies are were compared to a treatment as usual (TAU) group on also relevant to the possibility of genetic influences on measures of post-incarceration substance use and psy- relapse processes. For instance, genetic factors could chosocial functioning. Relative to the TAU group, the influence relapse in part via drug-specific cognitive pro- VM group reported significantly lower levels of substance cesses. Recent studies have reported genetic associations use and alcohol-related consequences and improved psy- with alcohol-related cognitions, including alcohol expec- chosocial functioning at follow-up [116]. tancies, drinking refusal self-efficacy, drinking motives, More recently, a randomized controlled trial com- and implicit measures of alcohol-related motivation pared an eight-week MBRP course to treatment as usual [51,52,104-108]. Overall, the body of research on genetic (TAU), which consisted of 12-step-based process- influences on relapse and related processes is nascent oriented discussion and psychoeducation groups [117]. and virtually all findings require replication. Consistent The majority of MBRP participants (86%) engaged in with the broader literature, it can be anticipated that meditation practices immediately posttreatment and most genetic associations with relapse outcomes will be 54% continued practice for at least 4 months posttreat- small in magnitude and potentially difficult to replicate. ment (M = 4.74 days/week, up to 30 min/day). Com- Nonetheless, initial studies have yielded intriguing pared to TAU, MBRP participants reported significantly results. It is inevitable that the next decade will see reduced craving, and increased acceptance and mindful exponential growth in this area, including greater use of awareness over the 4-month follow-up period, consistent genome-wide analyses of treatment response [109] and with the core goals of MBRP. Over the course of treat- efforts to evaluate the clinical utility and cost effective- ment, MBRP evinced fewer days of use compared to ness of tailoring treatments based on pharmacogenetics. TAU(MBRP:M=.06days,TAU:M=2.57days). Finally, an intriguing direction is to evaluate whether These differences persisted at 2-month follow-up (2.08 providing clients with personalized genetic information days for MBRP vs. 5.43 days for TAU). Secondary ana- can facilitate reductions in substance use or improve lyses [118] showed that compared to TAU, MBRP parti- treatment adherence [110,111]. cipants evinced a decreased relation between depressive symptoms and craving following treatment. This Mindfulness-based relapse prevention attenuation was related to subsequent decreases in alco- In terms of clinical applications of RP, the most notable hol and other drug use, suggesting MBRP led to development in the last decade has been the emergence decreased craving in response to negative affect, thereby and increasing application of Mindfulness-Based Relapse lessening the need to alleviate affective discomfort with Prevention (MBRP) for addictive behaviors [112,113]. alcohol and other drug use. Furthermore, individuals Given supportive data for the efficacy of mindfulness- with moderate depression in the MBRP group had a sig- based interventions in other behavioral domains, espe- nificantly lower probability of substance use, fewer cially in prevention of relapse of major depression [114], drinks per drinking day, and fewer drinks per day than there is increasing interest in MBRP for addictive beha- individuals with moderate depression in TAU. A larger, viors. The merger of mindfulness and cognitive-beha- randomized trial comparing MBRP to TAU and RP is vioral approaches is appealing from both theoretical and currently underway at the University of Washington to practical standpoints [115] and MBRP is a potentially evaluate whether the addition of mindfulness to the effective and cost-efficient adjunct to CBT-based treat- standard RP treatment leads to better substance use ments. In contrast to the cognitive restructuring strate- outcomes following treatment. As is the case in other gies typical of traditional CBT, MBRP stresses clinical domains [114], interest in MBRP for substance nonjudgmental attention to thoughts or urges. From use disorders is increasing rapidly. Results of additional this standpoint, urges/cravings are labeled as transient randomized controlled trials will be important for events that need not be acted upon reflexively. This informing its broader application for various addictive approach is exemplified by the “urge surfing” technique behaviors. Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 12 of 17 http://www.substanceabusepolicy.com/content/6/1/17

Critiques of the RP Model Mechanisms of treatment effects Following the initial introduction of the RP model in the Elucidating the “active ingredients” of CBT treatments 1980s, its widespread application largely outpaced efforts remains an important and challenging goal. Consistent to systematically validate the model and test its underly- with the RP model, changes in coping skills, self-efficacy ing assumptions. Given this limitation, the National Insti- and/or outcome expectancies are the primary putative tutes on Alcohol Abuse and Alcoholism (NIAAA) mechanisms by which CBT-based interventions work sponsored the Relapse Replication and Extension Project [126]. However, few studies support these presumptions. (RREP), a multi-site study aiming to test the reliability One study, in which substance-abusing individuals were and validity of Marlatt’s original relapse taxonomy. randomly assigned to RP or twelve-step (TS) treatments, Efforts to evaluate the validity [119] and predictive valid- found that RP participants showed increased self-effi- ity[120]ofthetaxonomyfailedtogeneratesupportive cacy, which accounted for unique variance in outcomes data. It was noted that in focusing on Marlatt’s relapse [69]. In a recent study, Witkiewitz and colleagues taxonomy the RREP did not comprehensive evaluation of (under review) found that individuals in the combined the full RP model [121]. Nevertheless, these studies were behavioral intervention of the COMBINE study who useful in identifying limitations and qualifications of the received drink-refusal skills training as part of the beha- RP taxonomy and generated valuable suggestions [121]. vioral intervention had significantly better outcomes The recently introduced dynamic model of relapse [8] than those who did not receive the drink-refusal skills takes many of the RREP criticisms into account. Addi- training, particularly African American clients [127]. tionally, the revised model has generated enthusiasm Further, there was strong support that increases in self- among researchers and clinicians who have observed efficacy following drink-refusal skills training was the these processes in their data and their clients [122,123]. primary mechanism of change. In another study examin- Still, some have criticized the model for not emphasiz- ing the behavioral intervention arm of the COMBINE ing interpersonal factors as proximal or phasic influ- study [128], individuals who received a skills training ences [122,123]. Other critiques include that nonlinear module focused on coping with craving and urges had dynamic systems approaches are not readily applicable significantly better drinking outcomes via decreases in to clinical interventions [124], and that the theory and negative mood and craving that occurred after receiving statistical methods underlying these approaches are eso- the module. teric for many researchers and clinicians [14]. Rather Despite findings like these, many studies of treatment than signaling weaknesses of the model, these issues mechanisms have failed to show that theoretical media- could simply reflect methodological challenges that tors account for salutary effects of CBT-based interven- researchers must overcome in order to better under- tions. Also, many studies that have examined potential stand dynamic aspects of behavior [45]. Ecological mediators of outcomes have not provided a rigorous momentary assessment [44], either via electronic device test [129] of mechanisms of change. These results sug- or interactive voice response methodology, could pro- gest that researchers should strive to consider alternative vide the data necessary to fully test the dynamic model mechanisms, improve assessment methods and/or revise of relapse. Ideally, assessments of coping, interpersonal theories about how CBT-based interventions work stress, self-efficacy, craving, mood, and other proximal [77,130]. factors could be collected multiple times per day over the course of several months, and combined with a Continued empirical evaluation of the RP model thorough pre-treatment assessment battery of distal risk As the foregoing review suggests, validation of the refor- factors. Future research with a data set that includes mulated RP model will likely progress slowly at first multiple measures of risk factors over multiple days because researchers are only beginning to evaluate could also take advantage of innovative modeling tools dynamic relapse processes. Currently, the dynamic that were designed for estimating nonlinear time-varying model can be viewed as a hypothetical, theory-driven dynamics [125]. framework that awaits empirical evaluation. Testing the model’s components will require that researchers avail Directions for Future Research themselves of innovative assessment techniques (such as Considering the numerous developments related to RP EMA) and pursue cross-disciplinary collaboration in over the last decade, empirical and clinical extensions of order to integrate appropriate statistical methods. Irre- the RP model will undoubtedly continue to evolve. In spective of study design, greater integration of distal and addition to the recent advances outlined above, we high- proximal variables will aid in modeling the interplay of light selected areas that are especially likely to see tonic and phasic influences on relapse outcomes. As was growth over the next several years. the case for Marlatt’s original RP model, efforts are Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 13 of 17 http://www.substanceabusepolicy.com/content/6/1/17

needed to systematically evaluate specific theoretical undoubtedly see a significant escalation in the number components of the reformulated model [1]. of studies using fMRI to predict response to psychoso- cial and pharmacological treatments. In this context, a Integrating implicit cognition and neurocognition in critical question will concern the predictive and clinical relapse models utility of brain-based measures with respect to predict- Historically, cognitive processes have been central to the ing treatment outcome. RP model [8]. In the last several years increasing emphasis has been placed on “dual process” models of Conclusions and Policy Implications addiction, which hypothesize that distinct (but related) Relapse prevention is a cognitive-behavioral approach cognitive networks, each reflective of specific neural designed to help individuals anticipate and cope with pathways, act to influence substance use behavior. setbacks during the behavior change process. The broad According to these models, the relative balance between aim of RP, to reduce the incidence and severity of controlled (explicit) and automatic (implicit) cognitive relapse, subsumes two basic goals: to minimize the networks is influential in guiding drug-related decision impact of high-risk situations by increasing awareness making [54,55]. Dual process accounts of addictive and building coping skills, and to limit relapse prone- behaviors [56,57] are likely to be useful for generating ness by promoting a healthy and balanced lifestyle. Over hypotheses about dynamic relapse processes and the past decade RP principles have been incorporated explaining variance in relapse, including episodes of sud- across an increasing array of behavior domains, with den divergence from abstinence to relapse. Implicit cog- addictive behaviors continuing to represent the primary nitive processes are also being examined as an application. intervention target, with some potentially promising As outlined in this review, the last decade has seen results [62]. notable developments in the RP literature, including sig- Related work has also stressed the importance of base- nificant expansion of empirical work with relevance to line levels of neurocognitive functioning (for example as the RP model. Overall, many basic tenets of the RP measured by tasks assessing response inhibition and model have received support and findings regarding its working memory; [56]) as predicting the likelihood of clinical effectiveness have generally been supportive. RP drug use in response to environmental cues. The study modules are standard to virtually all psychosocial inter- of implicit cognition and neurocognition in models of ventions for substance use [17] and an increasing num- relapse would likely require integration of distal neuro- ber of self-help manuals are available to assist both cognitive factors (e.g., baseline performance in cognitive therapists and clients. RP strategies can now be dissemi- tasks) in the context of treatment outcomes studies or nated using simple but effective methods; for instance, EMA paradigms. Additionally, lab-based studies will be mail-delivered RP booklets are shown to reduce smok- needed to capture dynamic processes involving cogni- ing relapse [135,136]. As noted earlier, the broad influ- tive/neurocognitive influences on lapse-related ence of RP is also evidenced by the current clinical phenomena. vernacular, as “relapse prevention” has evolved into an umbrella term synonymous with most cognitive-beha- Evaluating neural markers of relapse liability vioral skills-based interventions addressing high-risk The use of functional magnetic resonance imaging situations and coping responses. While attesting to the (fMRI) techniques in addictions research has increased influence and durability of the RP model, the tendency dramatically in the last decade [131] and many of these to subsume RP within various treatment modalities can studies have been instrumental in providing initial evi- also complicate efforts to systematically evaluate inter- dence on neural correlates of substance use and relapse. vention effects across studies (e.g., [21]). In one study of treatment-seeking methamphetamine Although many developments over the last decade users [132], researchers examined fMRI activation dur- encourage confidence in the RP model, additional ing a decision-making task and obtained information on research is needed to test its predictions, limitations and relapse over one year later. Based on activation patterns applicability. In particular, given recent theoretical revi- in several cortical regions they were able to correctly sions to the RP model, as well as the tendency for dif- identify 17 of 18 participants who relapsed and 20 of 22 fuse application of RP principles across different who did not. Functional imaging is increasingly being treatment modalities, there is an ongoing need to evalu- incorporated in treatment outcome studies (e.g., [133]) ate and characterize specific theoretical mechanisms of and there are increasing efforts to use imaging treatment effects. approaches to predict relapse [134]. While the overall In the current review we have noted several areas for number of studies examining neural correlates of relapse future research, including examining dynamic models of remains small at present, the coming years will treatment outcomes, extensions of RP to include Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 14 of 17 http://www.substanceabusepolicy.com/content/6/1/17

mindfulness and/or self-control training, research on the aftercare services. It is also important that policy makers mechanisms of change following successful treatment and funding entities support initiatives to evaluate RP outcomes, the role of genetic influences as potential and other established interventions in the context of moderators of treatment outcomes, and neurocognitive continuing care models. In general, more research on and neurobiological examinations of the relapse process the acquisition and long-term retention of specific RP using tests of implicit cognition and advanced neuroi- skills is necessary to better understand which RP skills maging techniques. In addition to these areas, which will be most useful in long-term and aftercare treat- already have initial empirical data, we predict that we ments for addictions. could learn significantly more about the relapse process using experimental manipulation to test specific aspects of the cognitive-behavioral model of relapse. For exam- Author details 1Centre for Addiction and Mental Health, 33 Russell St., Toronto, ON, M5S ple, it has been shown that self-efficacy for abstinence 2S1, Canada. 2Department of Psychiatry, University of Toronto, 250 College can be manipulated [137]. Thus, one could test whether St., Toronto, ON M5T 1R8, Canada. 3Department of Psychology, Washington increasing self-efficacy in an experimental design is State University, 14204 NE Salmon Creek Ave, Vancouver, WA, 98686, USA. 4Department of Psychology, University of Washington, Box 351525, Seattle, related to better treatment outcomes. Similarly, self-reg- WA 98195, USA. ulation ability, outcome expectancies, and the abstinence violation effect could all be experimentally manipulated, Authors’ contributions CH wrote the manuscript with contributions from KW. BG and GAM assisted which could eventually lead to further refinements of in conceptualizing the paper and provided critical review. All authors read RP strategies. and approved the manuscript. Ultimately, individuals who are struggling with beha- Competing interests vior change often find that making the initial change is The authors declare that they have no competing interests. not as difficult as maintaining behavior changes over time. Many therapies (both behavioral and pharmacolo- Received: 19 May 2011 Accepted: 19 July 2011 Published: 19 July 2011 gical) have been developed to help individuals cease or References reduce addictive behaviors and it is critical to refine 1. Brandon TH, Vidrine JI, Litvin EB: Relapse and relapse prevention. Annu Rev strategies for helping individuals maintain treatment Clin Psychol 2007, 3:257-284. goals. As noted by McLellan [138] and others [124], it is 2. Orleans CT: Promoting the maintenance of health behavior change: Recommendations for the next generation of research and practice. imperative that policy makers support adoption of treat- Health Psychol 2000, 19:76-83. ments that incorporate a continuing care approach, such 3. Polivy J, Herman CP: If at first you don’t succeed: False hopes of self- that addictions treatment is considered from a chronic change. Am Psychol 2002, 57:677-689. 4. Miller WR, Westerberg VS, Harris RJ, Tonigan JS: What predicts relapse? (rather than acute) care perspective. Broad implementa- Prospective testing of antecedent models. Addiction 1996, 91(Suppl): tion of a continuing care approach will require policy S155-172. change at numerous levels, including the adoption of 5. Hunt WA, Barnett LW, Branch LG: Relapse rates in addiction programs. J Clin Psychol 1971, 27:455-456. long-term patient-based and provider-based strategies 6. Kirshenbaum AP, Olsen DM, Bickel WK: A quantitative review of the and contingencies to optimize and sustain treatment ubiquitous relapse curve. J Subst Abuse Treat 2009, 36:8-17. outcomes [139,140]. 7. Marlatt GA, Gordon JR: Determinants of relapse: Implications for the maintenance of behavior change. In Behavioral Medicine: Changing health In support of continuing care approaches the United lifestyles. Edited by: Davidson PO, Davidson SM. New York: Brunner/Mazel; States Office of National Drug Control Policy recently 1980:410-452. published the 2010 National Drug Control Strategy in 8. Witkiewitz K, Marlatt GA: Relapse prevention for alcohol and drug problems: That was Zen, this is Tao. Am Psychol 2004, 59:224-235. the United States [141], which includes strategies to 9. Hendershot CS, Marlatt GA, George WH: Relapse prevention and the integrate treatment for substance use disorders into the maintenance of optimal health. In The Handbook of Health Behavior mainstream health care system and to expand support Change. Volume 2007.. 3 edition. Edited by: Shumaker S, Ockene JK, Riekert K. New York: Springer Publishing Co; . for continuing care efforts. One critical goal will be to 10. Witkiewitz K, Marlatt GA: Therapist’s Guide to Evidence-Based Relapse integrate empirically supported substance use interven- Prevention London: Academic Press; 2007. tions in the context of continuing care models of treat- 11. Marlatt GA, Witkiewitz K: In Relapse Prevention for Alcohol and Drug Problems.. 2 edition. Edited by: Marlatt G Alan, Donovan Dennis M. Relapse ment delivery, which in many cases requires adapting prevention: Maintenance strategies in the treatment of addictive behaviors; existing treatments to facilitate sustained delivery [140]. 2005:1-44, 2005.. Given its focus on long-term maintenance of treatment 12. Brownell KD, Marlatt GA, Lichtenstein E, Wilson GT: Understanding and preventing relapse. Am Psychol 1986, 41:765-782. gains, RP is a behavioral intervention that is particularly 13. Miller WR: Section I. theoretical perspectives on relapse: What is a well suited for implementation in continuing care con- relapse? Fifty ways to leave the wagon. Addiction 1996, 91:S15-S27. texts. Many treatment centers already provide RP as a 14. Maisto SA, Connors GJ: Relapse in the addictive behaviors: Integration and future directions. Clin Psychol Rev 2006, 26:229-231. routine component of aftercare programs. However, it is 15. Piasecki TM: Relapse to smoking. Clin Psychol Rev 2006, 26:196-215. imperative that insurance providers and funding entities 16. Marlatt GA, Gordon JR: Relapse prevention: Maintenance strategies in the support these efforts by providing financial support for treatment of addictive behaviors New York: Guilford Press; 1985. Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 15 of 17 http://www.substanceabusepolicy.com/content/6/1/17

17. McGovern MP, Wrisley BR, Drake RE: Relapse of substance use disorder 44. Stone AA, Shiffman S: Ecological momentary assessment (EMA) in and its prevention among persons with co-occurring disorders. Psychiatr behavorial medicine. Ann Behav Med 1994, 16:199-202. Serv 2005, 56:1270-1273. 45. Gwaltney CJ, Shiffman S, Balabanis MH, Paty JA: Dynamic Self-Efficacy and 18. Project MATCH Research Group: Matching alcoholism treatments to client Outcome Expectancies: Prediction of Smoking Lapse and Relapse. J heterogeneity: Project MATCH Posttreatment drinking outcomes. J Stud Abnorm Psychol 2005, 114:661-675. Alcohol 1997, 58:7-29. 46. Gwaltney CJ, Shiffman S, Sayette MA: Situational Correlates of Abstinence 19. COMBINE Study Research Group: Testing combined pharmacotherapies Self-Efficacy. J Abnorm Psychol 2005, 114:649-660. and behavioral interventions in alcohol dependence: rationale and 47. Gwaltney CJ, Metrik J, Kahler CW, Shiffman S: Self-efficacy and smoking methods. Alcohol Clin Exp Res 2003, 27:1107-1122. cessation: a meta-analysis. Psychol Addict Behav 2009, 23:56-66. 20. Rawson RA, Shoptaw SJ, Obert JL, McCann MJ, Hasson AL, Marinelli- 48. Herd N, Borland R: The natural history of quitting smoking: findings from Casey PJ, Brethen PR, Ling W: An intensive outpatient approach for the International Tobacco Control (ITC) Four Country Survey. Addiction cocaine abuse treatment. The Matrix model. J Subst Abuse Treat 1995, 2009, 104:2075-2087. 12:117-127. 49. Herd N, Borland R, Hyland A: Predictors of smoking relapse by duration 21. Miller WR, Wilbourne PL: Mesa Grande: A methodological analysis of of abstinence: findings from the International Tobacco Control (ITC) clinical trials of treatments for alcohol use disorders. Addiction 2002, Four Country Survey. Addiction 2009, 104:2088-2099. 97:265-277. 50. Katz EC, Fromme K, D’Amico EJ: Effects of outcome expectancies and 22. Cummings C, Gordon JR, Marlatt GA: Relapse: Strategies of prevention personality on young adults’ illicit drug use, heavy drinking, and risky and prediction. The addictive behaviors: Treatment of alcoholism, drug abuse, sexual behavior. Cognit Ther Res 2000, 24:1-22. smoking, and obesity Oxford: Pergamon; 1980, 291-321. 51. Hendershot CS, Neighbors C, George WH, McCarthy DM, Wall TL, Liang T, 23. Dimeff LA, Marlatt GA: Preventing relapse and maintaining change in Larimer ME: ALDH2, ADH1B and alcohol expectancies: integrating genetic addictive behaviors. Clin Psychol Sci Prac 1998, 5:513-525. and learning perspectives. Psychol Addict Behav 2009, 23:452-463. 24. Larimer ME, Palmer RS, Marlatt GA: Relapse prevention: An overview of 52. McCarthy DM, Wall TL, Brown SA, Carr LG: Integrating biological and Marlatt’s cognitive-behavioral model. Alcohol Res Health 1999, 23:151-160. behavioral factors in alcohol use risk: The role of ALDH2 status and 25. Baker TB, Piper ME, McCarthy DE, Majeskie MR, Fiore MC: Addiction alcohol expectancies in a sample of Asian Americans. Exp Clin Motivation Reformulated: An Affective Processing Model of Negative Psychopharmacol 2000, 8:168-175. Reinforcement. Psychol Rev 2004, 111:33-51. 53. Cohen LM, McCarthy DM, Brown SA, Myers MG: Negative affect combines 26. Bandura A: Self-efficacy: The exercise of control New York: Freeman; 1997. with smoking outcome expectancies to predict smoking behavior over 27. Goldman MS, Darkes J, Del Boca FK: In Expectancy mediation of time. Psychol Addict Behav 2002, 16:91-97. biopsychosocial risk for alcohol use and alcoholism. Edited by: Kirsch, Irving. 54. Stacy AW, Wiers RW: Implicit cognition and addiction: a tool for Washington, DC, US: American Psychological Association.; 1999; explaining paradoxical behavior. Annu Rev Clin Psychol 2010, 6:551-575. 1999:233-262, How expectancies shape experience. 55. Wiers RW, Stacy AW: Handbook of implicit cognition and addiction 28. Rothman AJ: Toward a theory-based analysis of behavioral maintenance. Handbook of implicit cognition and addiction. Thousand Oaks, CA: Sage Health Psychol 2000, 19:64-69. Publications; 2006, 2006.. 29. Marlatt GA, Donovan DM: Relapse prevention: Maintenance strategies in the 56. Field M, Wiers RW, Christiansen P, Fillmore MT, Verster JC: Acute alcohol treatment of addictive behaviors. 2 edition. New York: Guilford; 2005. effects on inhibitory control and implicit cognition: implications for loss 30. Donovan DM, Marlatt GA: Assessment of addictive behaviors. 2 edition. New of control over drinking. Alcohol Clin Exp Res 2010, 34:1346-1352. York: Guilford; 2005. 57. Wiers RW, Bartholow BD, van den Wildenberg E, Thush C, Engels RC, 31. Hufford MR, Witkiewitz K, Shields AL, Kodya S, Caruso JC: Relapse as a Sher KJ, Grenard J, Ames SL, Stacy AW: Automatic and controlled nonlinear dynamic system: Application to patients with alcohol use processes and the development of addictive behaviors in adolescents: A disorders. J Abnorm Psychol 2003, 112:219-227. review and a model. Pharmacol Biochem Behav 2007, 86:263-283. 32. Bouton ME: A learning theory perspective on lapse, relapse, and the 58. Kramer DA, Goldman MS: Using a modified Stroop task to implicitly maintenance of behavior change. Health Psychol 2000, 19:57-63. discern the cognitive organization of alcohol expectancies. J Abnorm 33. Witkiewitz K, Marlatt GA: Modeling the complexity of post-treatment Psychol 2003, 112:171-175. drinking: it’s a rocky road to relapse. Clin Psychol Rev 2007, 27:724-738. 59. McCarthy DM, Thompsen DM: Implicit and Explicit Measures of Alcohol 34. Witkiewitz K, van der Maas HLJ, Hufford MR, Marlatt GA: Nonnormality and and Smoking Cognitions. Psychol Addict Behav 2006, 20:436-444. divergence in posttreatment alcohol use: Reexamining the Project 60. Waters AJ, Shiffman S, Sayette MA, Paty JA, Gwaltney CJ, Balabanis MH: MATCH data “another way.”. J Abnorm Psychol 2007, 116:378-394. Attentional bias predicts outcome in smoking cessation. Health Psychol 35. Carroll KM: Relapse prevention as a psychosocial treatment: A review of 2003, 22:378-387. controlled clinical trials. Exp Clin Psychopharmacol 1996, 4:46-54. 61. Marissen MA, Franken IH, Waters AJ, Blanken P, van den Brink W, 36. Irvin JE, Bowers CA, Dunn ME, Wang MC: Efficacy of relapse prevention: A Hendriks VM: Attentional bias predicts heroin relapse following meta-analytic review. J Consult Clin Psychol 1999, 67:563-570. treatment. Addiction 2006, 101:1306-1312. 37. McCrady BS: Alcohol use disorders and the Division 12 Task Force of the 62. Schoenmakers TM, de Bruin M, Lux IF, Goertz AG, Van Kerkhof DH, American Psychological Association. Psychol Addict Behav 2000, Wiers RW: Clinical effectiveness of attentional bias modification training 14:267-276. in abstinent alcoholic patients. Drug Alcohol Depend 2010, 109:30-36. 38. Hajek P, Stead LF, West R, Jarvis M, Lancaster T: Relapse prevention 63. Piasecki TM, Jorenby DE, Smith SS, Fiore MC, Baker TB: Smoking interventions for smoking cessation. Cochrane Database Syst Rev 2009, Withdrawal Dynamics: III. Correlates of Withdrawal Heterogeneity. Exp CD003999. Clin Psychopharmacol 2003, 11:276-285. 39. Lancaster T, Hajek P, Stead LF, West R, Jarvis MJ: Prevention of relapse 64. Piasecki TM, Niaura R, Shadel WG, Abrams D, Goldstein M, Fiore MC, after quitting smoking: a systematic review of trials. Arch Intern Med Baker TB: Smoking withdrawal dynamics in unaided quitters. J Abnorm 2006, 166:828-835. Psychol 2000, 109:74-86. 40. Agboola S, McNeill A, Coleman T, Leonardi Bee J: A systematic review of 65. Piasecki TM, Fiore MC, Baker TB: Profiles in discouragement: Two studies the effectiveness of smoking relapse prevention interventions for of variability in the time course of smoking withdrawal symptoms. J abstinent smokers. Addiction 2010, 105:1362-1380. Abnorm Psychol 1998, 107:238-251, 2000. 41. Magill M, Ray LA: Cognitive-behavioral treatment with adult alcohol and 66. Baker TB, Japuntich SJ, Hogle JM, McCarthy DE, Curtin JJ: Pharmacologic illicit drug users: a meta-analysis of randomized controlled trials. J Stud and Behavioral Withdrawal from Addictive Drugs. Curr Dir Psychol Sci Alcohol Drugs 2009, 70:516-527. 2006, 15:232-236. 42. Niaura R: Cognitive social learning and related perspectives on drug 67. Marlatt GA: Craving for alcohol, loss of control, and relapse: A cognitive- craving. Addiction 2000, 95(Suppl 2):S155-163. behavioral analysis. In Alcoholism: New directions in behavioral research and 43. Shiffman S, Balabanis MH, Paty JA, Engberg J, Gwaltney CJ, Liu KS, Gnys M, treatment. Edited by: Nathan PE, Marlatt GA, Loberg T. New York: Plenum; Hickcox M, Paton SM: Dynamic effects of self-efficacy on smoking lapse 1978:271-314. and relapse. Health Psychol 2000, 19:315-323. Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 16 of 17 http://www.substanceabusepolicy.com/content/6/1/17

68. Shiffman S, Hickcox M, Paty JA, Gnys M, Kassel JD, Richards TJ: Progression transporter and receptor polymorphisms on smoking cessation in a from a smoking lapse to relapse: Prediction from abstinence violation bupropion clinical trial. Health Psychol 2003, 22:541-548. effects, nicotine dependence, and lapse characteristics. J Consult Clin 91. Lerman C, Wileyto EP, Patterson F, Rukstalis M, Audrain-McGovern J, Psychol 1996, 64:993-1002. Restine S, Shields PG, Kaufmann V, Redden D, Benowitz N, Berrettini WH: 69. Brown RA, Lejuez CW, Kahler CW, Strong DR: Distress tolerance and The functional mu opioid receptor (OPRM1) Asn40Asp variant predicts duration of past smoking cessation attempts. J Abnorm Psychol 2002, short-term response to nicotine replacement therapy in a clinical trial. 111:180-185. Pharmacogenomics J 2004, 4:184-192. 70. Shiffman S, Waters AJ: Negative Affect and Smoking Lapses: A 92. Oslin DW, Berrettini W, Kranzler HR, Pettinati H, Gelernter J, Volpicelli JR, Prospective Analysis. J Consult Clin Psychol 2004, 72:192-201. O’Brien CP: A Functional Polymorphism of the mu-Opioid receptor gene 71. Gwaltney CJ, Shiffman S, Normal GJ, Paty JA, Kassel JD, Gnys M, Hickcox M, is associated with naltrexone response in alcohol-dependent patients. Waters A, Balabanis M: Does smoking abstinence self-efficacy vary across Neuropsychopharmacology 2003, 28:1546-1552. situations? Identifying context-specificity within the Relapse Situation 93. Anton RF, Oroszi G, O’Malley S, Couper D, Swift R, Pettinati H, Goldman D: Efficacy Questionnaire. J Consult Clin Psychol 2001, 69:516-527. An evaluation of mu-opioid receptor (OPRM1) as a predictor of 72. Witkiewitz K, Villarroel NA: Dynamic association between negative affect naltrexone response in the treatment of alcohol dependence: Results and alcohol lapses following alcohol treatment. J Consult Clin Psychol from the Combined Pharmacotherapies and Behavioral Interventions for 2009, 77:633-644. Alcohol Dependence (COMBINE) study. Arch Gen Psychiatry 2008, 73. Armeli S, Tennen H, Affleck G, Kranzler HR: Does affect mediate the 65:135-144. association between daily events and alcohol use? J Stud Alcohol 2000, 94. Ray R, Jepson C, Wileyto EP, Dahl JP, Patterson F, Rukstalis M, Pinto A, 61:862-871. Berrettini W, Lerman C: Genetic variation in mu-opioid-receptor- 74. Hussong AM, Hicks RE, Levy SA, Curran PJ: Specifying the relations interacting proteins and smoking cessation in a nicotine replacement between affect and heavy alcohol use among young adults. J Abnorm therapy trial. Nicotine Tob Res 2007, 9:1237-1241. Psychol 2001, 110:449-461. 95. Ray LA, Hutchison KE: A polymorphism of the mu-opioid receptor gene 75. Morgenstern J, Longabaugh R: Cognitive-behavioral treatment for alcohol (OPRM1) and sensitivity to the effects of alcohol in humans. Alcohol Clin dependence: A review of evidence for its hypothesized mechanisms of Exp Res 2004, 28:1789-1795. action. Addiction 2000, 95:1475-1490. 96. Filbey FM, Ray L, Smolen A, Claus ED, Audette A, Hutchison KE: Differential 76. O’Connell KA, Hosein VL, Schwartz JE, Leibowitz RQ: How Does Coping neural response to alcohol priming and alcohol taste cues is associated Help People Resist Lapses During Smoking Cessation? Health Psychol with DRD4 VNTR and OPRM1 genotypes. Alcohol Clin Exp Res 2008, 2007, 26:77-84. 32:1113-1123. 77. Litt MD, Kadden RM, Stephens RS: Coping and Self-Efficacy in Marijuana 97. van den Wildenberg E, Wiers RW, Dessers J, Janssen RGJH, Lambrichs EH, Treatment: Results From the Marijuana Treatment Project. J Consult Clin Smeets HJM, van Breukelen GJP: A functional polymorphism of the mu- Psychol 2005, 73:1015-1025. opioid receptor gene (OPRM1) influences cue-induced craving for 78. Muraven M, Baumeister RF: Self-regulation and depletion of limited alcohol in male heavy drinkers. Alcohol Clin Exp Res 2007, 31:1-10. resources: Does self-control resemble a muscle? Psychol Bull 2000, 98. Ramchandani VA, Umhau J, Pavon FJ, Ruiz-Velasco V, Margas W, Sun H, 126:247-259. Damadzic R, Eskay R, Schoor M, Thorsell A, et al: A genetic determinant of 79. Muraven M, Collins RL, Shiffman S, Paty JA: Daily Fluctuations in Self- the striatal dopamine response to alcohol in men. Mol Psychiatry 2010. Control Demands and Alcohol Intake. Psychol Addict Behav 2005, 99. Ooteman W, Naassila M, Koeter MW, Verheul R, Schippers GM, Houchi H, 19:140-147. Daoust M, van den Brink W: Predicting the effect of naltrexone and 80. Muraven M, Collins RL, Morsheimer ET, Shiffman S, Paty JA: The Morning acamprosate in alcohol-dependent patients using genetic indicators. After: Limit Violations and the Self-Regulation of Alcohol Consumption. Addict Biol 2009, 14:328-337. Psychol Addict Behav 2005, 19:253-262. 100. Hutchison KE, Ray L, Sandman E, Rutter MC, Peters A, Davidson D, Swift R: 81. Muraven M: Practicing self-control lowers the risk of smoking lapse. The effect of olanzapine on craving and alcohol consumption. Psychol Addict Behav 2010, 24:446-452. Neuropsychopharmacology 2006, 31:1310-1317. 82. Witkiewitz K, Hartzler B, Donovan D: Matching motivation enhancement 101. Hutchison KE, Wooden A, Swift RM, Smolen A, McGeary J, Adler L, Paris L: treatment to client motivation: re-examining the Project MATCH Olanzapine reduces craving for alcohol: A DRD4 VNTR polymorphism by motivation matching hypothesis. Addiction 2010, 105:1403-1413. pharmacotherapy interaction. Neuropsychopharmacology 2003, 83. Ho MK, Goldman D, Heinz A, Kaprio J, Kreek MJ, Li MD, Munafo MR, 28:1882-1888. Tyndale RF: Breaking barriers in the genomics and pharmacogenetics of 102. Bauer LO, Covault J, Harel O, Das S, Gelernter J, Anton R, Kranzler HR: drug addiction. Clin Pharmacol Ther 2010, 88:779-791. Variation in GABRA2 predicts drinking behavior in project MATCH 84. Sturgess JE, George TP, Kennedy JL, Heinz A, Muller DJ: Pharmacogenetics subjects. Alcohol Clin Exp Res 2007, 31:1780-1787. of alcohol, nicotine and drug addiction treatments. Addict Biol 2011. 103. Wojnar M, Brower KJ, Strobbe S, Ilgen M, Matsumoto H, Nowosad I, 85. Colilla S, Lerman C, Shields PG, Jepson C, Rukstalis M, Berlin J, DeMichele A, Sliwerska E, Burmeister M: Association between Val66Met brain-derived Bunin G, Strom BL, Rebbeck TR: Association of catechol-O- neurotrophic factor (BDNF) gene polymorphism and post-treatment methyltransferase with smoking cessation in two independent studies relapse in alcohol dependence. Alcohol Clin Exp Res 2009, 33:693-702. of women. Pharmacogenet Genomics 2005, 15:393-398. 104. Hendershot CS, Lindgren KP, Liang T, Hutchison KE: COMT and ALDH2 86. Johnstone EC, Elliot KM, David SP, Murphy MF, Walton RT, Munafo MR: polymorphisms moderate associations of implicit drinking motives with Association of COMT Val108/158Met genotype with smoking cessation alcohol use. Addict Biol 2011. in a nicotine replacement therapy randomized trial. Cancer Epidemiol 105. Hendershot CS, Witkiewitz K, George WH, Wall TL, Otto JM, Liang T, Biomarkers Prev 2007, 16:1065-1069. Larimer ME: Evaluating a cognitive model of ALDH2 and drinking 87. Munafo MR, Johnstone EC, Guo B, Murphy MF, Aveyard P: Association of behavior. Alcohol Clin Exp Res 2011, 35:91-98. COMT Val108/158Met genotype with smoking cessation. Pharmacogenet 106. Miranda R, Ray L, Justus A, Meyerson LA, Knopik VS, McGeary J, Monti PM: Genomics 2008, 18:121-128. Initial evidence of an association between OPRM1 and adolescent 88. Conti DV, Lee W, Li D, Liu J, Van Den Berg D, Thomas PD, Bergen AW, alcohol misuse. Alcohol Clin Exp Res 2010, 34:112-122. Swan GE, Tyndale RF, Benowitz NL, Lerman C: Nicotinic acetylcholine 107. Wiers RW, Rinck M, Dictus M, van den Wildenberg E: Relatively strong receptor beta2 subunit gene implicated in a systems-based candidate automatic appetitive action-tendencies in male carriers of the OPRM1 G- gene study of smoking cessation. Hum Mol Genet 2008, 17:2834-2848. allele. Genes Brain Behav 2009, 8:101-106. 89. Perkins KA, Lerman C, Mercincavage M, Fonte CA, Briski JL: Nicotinic 108. Young RM, Lawford BR, Feeney GFX, Ritchie T, Noble EP: Alcohol-related acetylcholine receptor beta2 subunit (CHRNB2) gene and short-term expectancies are associated with the D2 dopamine receptor and GABA ability to quit smoking in response to nicotine patch. Cancer Epidemiol receptor beta3 subunit genes. Psychiatry Res 2004, 127:171-183. Biomarkers Prev 2009, 18:2608-2612. 109. Uhl GR, Liu QR, Drgon T, Johnson C, Walther D, Rose JE, David SP, Niaura R, 90. Lerman C, Shields PG, Wileyto EP, Audrain J, Hawk LH Jr, Pinto A, Lerman C: Molecular genetics of successful smoking cessation: Kucharski S, Krishnan S, Niaura R, Epstein LH: Effects of dopamine Hendershot et al. Substance Abuse Treatment, Prevention, and Policy 2011, 6:17 Page 17 of 17 http://www.substanceabusepolicy.com/content/6/1/17

convergent genome-wide association study results. Arch Gen Psychiatry 131. Volkow ND, Fowler JS, Wang GJ: The addicted human brain viewed in the 2008, 65:683-693. light of imaging studies: brain circuits and treatment strategies. 110. Hendershot CS, Otto JM, Collins SE, Liang T, Wall TL: Evaluation of a brief Neuropharmacology 2004, 47(Suppl 1):3-13. web-based genetic feedback intervention for reducing alcohol-related 132. Paulus MP, Tapert SF, Schuckit MA: Neural Activation Patterns of health risks associated with ALDH2. Ann Behav Med 2010, 40:77-88. Methamphetamine-Dependent Subjects During Decision Making Predict 111. Marteau TM, Munafo MR, Aveyard P, Hill C, Whitwell S, Willis TA, Relapse. Arch Gen Psychiatry 2005, 62:761-768. Crockett RA, Hollands GJ, Johnstone EC, Wright AJ, et al: Trial Protocol: 133. Mann K, Kiefer F, Smolka M, Gann H, Wellek S, Heinz A: Searching for Using genotype to tailor prescribing of nicotine replacement therapy: a responders to acamprosate and naltrexone in alcoholism treatment: randomised controlled trial assessing impact of communication upon rationale and design of the PREDICT study. Alcohol Clin Exp Res 2009, adherence. BMC Public Health 2010, 10:680. 33:674-683. 112. Witkiewitz K, Marlatt GA, Walker D: Mindfulness-Based Relapse Prevention 134. Durazzo TC, Gazdzinski S, Mon A, Meyerhoff DJ: Cortical perfusion in for Alcohol and Substance Use Disorders. J Cogn Psychother 2005, alcohol-dependent individuals during short-term abstinence: 19:211-228. Relationships to resumption of hazardous drinking after treatment. 113. Zgierska A, Rabago D, Chawla N, Kushner K, Koehler R, Marlatt A: Alcohol 2010, 44:201-210. Mindfulness meditation for substance use disorders: a systematic 135. Brandon TH, Collins BN, Juliano LM, Lazev AB: Preventing relapse among review. Subst Abus 2009, 30:266-294. former smokers: A comparison of minimal interventions through 114. Lau MA, Segal ZV: Mindfulness-based cognitive therapy as a relapse telephone and mail. J Consult Clin Psychol 2000, 68:103-113. prevention approach to depression. In Therapist’s Guide to Evidence-Based 136. Brandon TH, Meade CD, Herzog TA, Chirikos TN, Webb MS, Cantor AB: Relapse Prevention. Edited by: Witkiewitz K, Marlatt GA. London: Academic Efficacy and cost-effectiveness of a minimal intervention to prevent Press; 2007:. smoking relapse: Dismantling the effects of amount of content versus 115. Marlatt GA: Buddhist philosophy and the treatment of addictive contact. J Consult Clin Psychol 2004, 72:797-808. behavior. Cogn Behav Pract 2002, 9:44-49. 137. Shadel WG, Cervone D: Evaluating social-cognitive mechanisms that 116. Bowen S, Witkiewitz K, Dillworth TM, Chawla N, Simpson TL, Ostafin BD, regulate self-efficacy in response to provocative smoking cues: An Larimer ME, Blume AW, Parks GA, Marlatt GA: Mindfulness meditation and experimental investigation. Psychol Addict Behav 2006, 20:91-96. substance use in an incarcerated population. Psychol Addict Behav 2006, 138. McLellan AT: Have we evaluated addiction treatment correctly? 20:343-347. Implications from a chronic care perspective. Addiction 2002, 97:249-252. 117. Bowen S, Chawla N, Collins SE, Witkiewitz K, Hsu S, Grow J, Clifasefi S, 139. Humphreys K, McLellan AT: A policy-oriented review of strategies for Garner M, Douglass A, Larimer ME, Marlatt A: Mindfulness-based relapse improving the outcomes of services for substance use disorder prevention for substance use disorders: a pilot efficacy trial. Subst Abus patients*. Addiction 2011. 2009, 30:295-305. 140. McKay JR, Carise D, Dennis ML, Dupont R, Humphreys K, Kemp J, 118. Witkiewitz K, Bowen S: Depression, craving, and substance use following Reynolds D, White W, Armstrong R, Chalk M, et al: Extending the benefits a randomized trial of mindfulness-based relapse prevention. J Consult of addiction treatment: practical strategies for continuing care and Clin Psychol 2010, 78:362-374. recovery. J Subst Abuse Treat 2009, 36:127-130. 119. Maisto SA, Connors GJ, Zywiak WH: Replication and extension of Marlatt’s 141. Office of National Drug Control Policy: National Drug Control Strategy for taxonomy: Construct validation analyses on the Marlatt typology of 2010.[http://www.ondcp.gov]. relapse precipitants. Addiction 1996, 91:S89-S97. 120. Stout RL, Longabaugh R, Rubin A: Replication and extension of Marlatt’s doi:10.1186/1747-597X-6-17 taxonomy: Predictive validity of Marlatt’s relapse taxonomy versus a Cite this article as: Hendershot et al.: Relapse prevention for addictive more general relapse code. Addiction 1996, 91:S99-S110. behaviors. Substance Abuse Treatment, Prevention, and Policy 2011 6:17. 121. Donovan DM: Commentary on replications of Marlatt’s taxonomy: Marlatt’s classification of relapse precipitants: Is the emperor still wearing clothes? Addiction 1996, 91:S131-S137. 122. Hunter-Reel D, McCrady B, Hildebrandt T: Emphasizing interpersonal factors: an extension of the Witkiewitz and Marlatt relapse model. Addiction 2009, 104:1281-1290. 123. Stanton M: Relapse Prevention Needs More Emphasis on Interpersonal Factors. Am Psychol 2005, 60:340-341. 124. McKay JR, Franklin TR, Patapis N, Lynch KG: Conceptual, methodological, and analytical issues in the study of relapse. Clin Psychol Rev 2006, 26:109-127. 125. Walls TA, Schafer JL: Models for Intensive Longitudinal Data New York: Oxford University Press; 2006. 126. Longabaugh R, Donovan DM, Karno MP, McCrady BS, Morgenstern J, Tonigan JS: Active ingredients: How and why evidence-based alcohol behavioral treatment interventions work. Alcohol Clin Exp Res 2005, 29:235-247. 127. Witkiewitz K, Villarroel NA, Hartzler B, Donovan DM: Drinking outcomes following drink refusal skills training: Differential effects for African American and non-Hispanic White clients. Psychol Addict Behav 2011, Submit your next manuscript to BioMed Central 25:162-167. and take full advantage of: 128. Witkiewitz K, Bowen S, Donovan DM: Moderating effects of a craving intervention on the relation between negative mood and heavy drinking following treatment for alcohol dependence. J Consult Clin • Convenient online submission Psychol 2011, 79:54-63. • Thorough peer review 129. Kazdin AE, Nock MK: Delineating mechanisms of change in child and • No space constraints or color figure charges adolescent therapy: methodological issues and research recommendations. J Child Psychol Psychiatry 2003, 44:1116-1129. • Immediate publication on acceptance 130. Morgenstern J, Longabaugh R: Cognitive-behavioral treatment for alcohol • Inclusion in PubMed, CAS, Scopus and Google Scholar dependence: A review of evidence for its hypothesized mechanisms of • Research which is freely available for redistribution action. Addiction 2000, 95:1475-1490.

Submit your manuscript at www.biomedcentral.com/submit