Relapse Prevention

An Overview of Marlatt’s Cognitive-Behavioral Model

Mary E. Larimer, Ph.D., Rebekka S. Palmer, and G. Alan Marlatt, Ph.D.

Relapse prevention (RP) is an important component of alcoholism treatment. The RP model proposed by Marlatt and Gordon suggests that both immediate determinants (e.g., high-risk situations, coping skills, outcome expectancies, and the abstinence violation effect) and covert antecedents (e.g., lifestyle factors and urges and cravings) can contribute to relapse. The RP model also incorporates numerous specific and global intervention strategies that allow therapist and client to address each step of the relapse process. Specific interventions include identifying specific high-risk situations for each client and enhancing the client’s skills for coping with those situations, increasing the client’s self-efficacy, eliminating myths regarding alcohol’s effects, managing lapses, and restructuring the client’s perceptions of the relapse process. Global strategies comprise balancing the client’s lifestyle and helping him or her develop positive , employing control techniques and urge- management techniques, and developing relapse road maps. Several studies have provided theoretical and practical support for the RP model. KEY WORDS:AODD (alcohol and other drug dependence) relapse; relapse prevention; treatment model; ; behavior therapy; risk factors; coping skills; self efficacy; expectancy; AOD (alcohol and other drug) abstinence; lifestyle; AOD craving; intervention; alcohol cue; reliability (research methods); validity (research methods); literature review

elapse, or the return to heavy cess, a series of events that unfold over alcohol use following a period of time (Annis 1986; Litman et al. 1979; MARY E. LARIMER, PH.D., is a research Rabstinence or moderate use, occurs Marlatt and Gordon 1985). According assistant professor of psychology, REBEKKA S. in many drinkers who have undergone to these models, the relapse process PALMER is a graduate student in clinical alcoholism treatment. Traditional alco- begins prior to the first posttreatment psychology, and G. ALAN MARLATT, holism treatment approaches often alcohol use and continues after the initial PH.D., is a professor of psychology at the conceptualize relapse as an end-state, a use. This conceptualization provides Addictive Behaviors Research Center, negative outcome equivalent to treatment a broader conceptual framework for Department of Psychology, University of failure. Thus, this perspective considers intervening in the relapse process to Washington, Seattle, Washington. only a dichotomous treatment out- prevent or reduce relapse episodes and come—that is, a person is either absti- thereby improve treatment outcome. Preparation of this manuscript was sup- nent or relapsed. In contrast, several This article presents one influential ported by National Institute on Alcohol models of relapse that are based on model of the antecedents of relapse and Abuse and Alcoholism grants R3A–AA– social-cognitive or behavioral theories the treatment measures that can be 05591 to G. Alan Marlatt and 5RO1– emphasize relapse as a transitional pro- taken to prevent or limit relapse after AA–10772–03 to Mary E. Larimer.

Vol. 23, No. 2, 1999 151 treatment completion. This relapse pre- description of the development, theo- response to these situations, as well as vention (RP) model, which was devel- retical underpinnings, and treatment backward to examine the lifestyle fac- oped by Marlatt and Gordon (1985) and components of the RP model, see Dimeff tors that increase the drinker’s exposure which has been widely used in recent and Marlatt 1998; Marlatt 1996; to high-risk situations. Based on this years, has been the focus of considerable Marlatt and Gordon 1985). A central careful examination of the relapse pro- research. This article reviews various aspect of the model is the detailed clas- cess, the therapist then devises strate- immediate and covert triggers of relapse sification (i.e., taxonomy) of factors or gies to target weaknesses in the client’s proposed by the RP model, as well as situations that can precipitate or con- cognitive and behavioral repertoire and numerous specific and general interven- tribute to relapse episodes. In general, the thereby reduce the risk of relapse. tion strategies that may help patients RP model posits that those factors fall avoid and cope with relapse-inducing into two categories: immediate determi- situations. The article also presents nants (e.g., high-risk situations, a per- Immediate Determinants of Relapse studies that have provided support for son’s coping skills, outcome expectancies, the validity of the RP model. and the abstinence violation effect) and High-Risk Situations. A central con- covert antecedents (e.g., lifestyle imbal- cept of the RP model postulates that ances and urges and cravings). high-risk situations frequently serve as Overview of the RP Model Treatment approaches based on the the immediate precipitators of initial RP model begin with an assessment of alcohol use after abstinence (see figure 1). Marlatt and Gordon’s (1985) RP model the environmental and emotional char- According to the model, a person who is based on social-cognitive psychology acteristics of situations that are potentially has initiated a behavior change, such as and incorporates both a conceptual associated with relapse (i.e., high-risk alcohol abstinence, should begin expe- model of relapse and a set of cognitive situations). After identifying those riencing increased self-efficacy or mas- and behavioral strategies to prevent or characteristics, the therapist works for- tery over his or her behavior, which limit relapse episodes (for a detailed ward by analyzing the individual drinker’s should grow as he or she continues to

Decreased Effective coping Increased probability of response self-efficacy relapse

High-risk situation

Decreased self-efficacy Abstinence and violation Ineffective Lapse (initial Increased positive effect and coping use of probability outcome perceived response alcohol) expectancies positive of relapse for effects of effects of alcohol alcohol

Figure 1 The cognitive-behavioral model of the relapse process posits a central role for high-risk situations and for the drinker’s response to those situations. People with effective coping responses have confidence that they can cope with the situation (i.e., increased self-efficacy), thereby reducing the probability of a relapse. Conversely, people with ineffective coping responses will experience decreased self-efficacy, which, together with the expectation that alcohol use will have a positive effect (i.e., positive outcome expectancies), can result in an initial lapse. This lapse, in turn, can result in feelings of guilt and failure (i.e., an abstinence violation effect). The abstinence violation effect, along with positive outcome expectancies, can increase the probability of a relapse.

NOTE: This model also applies to users of drugs other than alcohol.

152 Alcohol Research & Health Relapse Prevention: Marlatt’s Cognitive-Behavioral Model

maintain the change. Certain situations (i.e., using “willpower” to limit con- an initial lapse) and a return to uncon- or events, however, can pose a threat sumption), and nonspecific cravings trolled drinking or abandonment of to the person’s sense of control and, also were identified as high-risk situ- the abstinence goal (i.e., a full-blown consequently, precipitate a relapse cri- ations that could precipitate relapse. relapse). Although research with vari- sis. Based on research on precipitants of ous addictive behaviors has indicated relapse in alcoholics who had received Coping. Although the RP model con- that a lapse greatly increases the risk of inpatient treatment, Marlatt (1996) siders the high-risk situation the imme- eventual relapse, the progression from categorized the emotional, environmen- diate relapse trigger, it is actually the lapse to relapse is not inevitable. tal, and interpersonal characteristics of person’s response to the situation that Marlatt and Gordon (1980, 1985) relapse-inducing situations described determines whether he or she will expe- have described a type of reaction by the by study participants. According to this rience a lapse (i.e., begin using alco- drinker to a lapse called the abstinence taxonomy, several types of situations hol). A person’s coping behavior in a violation effect, which may influence can play a role in relapse episodes, as high-risk situation is a particularly criti- whether a lapse leads to relapse. This follows: cal determinant of the likely outcome. reaction focuses on the drinker’s emo- Thus, a person who can execute effec- tional response to an initial lapse and • Negative emotional states, such as tive coping strategies (e.g., a behavioral on the causes to which he or she attributes anger, anxiety, depression, frustra- strategy, such as leaving the situation, the lapse. People who attribute the lapse tion, and boredom, which are also or a cognitive strategy, such as positive to their own personal failure are likely referred to as intrapersonal high-risk self-talk) is less likely to relapse com- to experience guilt and negative emo- situations, are associated with the pared with a person lacking those skills. tions that can, in turn, lead to increased highest rate of relapse (Marlatt and Moreover, people who have coped suc- drinking as a further attempt to avoid Gordon 1985). These emotional cessfully with high-risk situations are or escape the feelings of guilt or failure. states may be caused by primarily assumed to experience a heightened Furthermore, people who attribute the intrapersonal perceptions of certain sense of self-efficacy (i.e., a personal lapse to stable, global, internal factors situations (e.g., feeling bored or lonely perception of mastery over the specific beyond their control (e.g., “I have no after coming home from work to an risky situation) (Bandura 1977; Marlatt willpower and will never be able to stop empty house) or by reactions to envi- et al. 1995, 1999; Marlatt and Gordon drinking”) are more likely to abandon ronmental events (e.g., feeling angry 1985). Conversely, people with low the abstinence attempt (and experience about an impending layoff at work). self-efficacy perceive themselves as lack- a full-blown relapse) than are people ing the or ability to resist who attribute the lapse to their inability • Situations that involve another person drinking in high-risk situations. to cope effectively with a specific high- or a group of people (i.e., interper- risk situation. In contrast to the former sonal high-risk situations), particu- Outcome Expectancies. Research group of people, the latter group real- larly interpersonal conflict (e.g., an among college students has shown that izes that one needs to “learn from one’s argument with a family member), those who drink the most tend to have mistakes” and, thus, they may develop also result in negative emotions and higher expectations regarding the posi- more effective ways to cope with simi- can precipitate relapse. In fact, intra- tive effects of alcohol (i.e., outcome lar trigger situations in the future. personal negative emotional states and expectancies) and may anticipate only interpersonal conflict situations served the immediate positive effects while as triggers for more than one-half ignoring or discounting the potential Covert Antecedents of High-Risk of all relapse episodes in Marlatt’s negative consequences of excessive Situations (1996) analysis. drinking (Carey 1995). Such positive Although high-risk situations can be outcome expectancies may become conceptualized as the immediate deter- • Social pressure, including both direct particularly salient in high-risk situa- minants of relapse episodes, a number verbal or nonverbal persuasion and tions, when the person expects alcohol of less obvious factors also influence the indirect pressure (e.g., being around use to help him or her cope with nega- relapse process. These covert antecedents other people who are drinking), con- tive emotions or conflict (i.e., when include lifestyle factors, such as overall tributed to more than 20 percent of drinking serves as “self-medication”). stress level, as well as cognitive factors relapse episodes in Marlatt’s (1996) In these situations, the drinker focuses that may serve to “set up” a relapse, study. primarily on the anticipation of imme- such as rationalization, denial, and a diate gratification, such as stress reduc- desire for immediate gratification (i.e., • Positive emotional states (e.g., cele- tion, neglecting possible delayed nega- urges and cravings) (see figure 2). brations), exposure to alcohol- tive consequences. These factors can increase a person’s related stimuli or cues (e.g., seeing vulnerability to relapse both by increas- an advertisement for an alcoholic The Abstinence Violation Effect. A crit- ing his or her exposure to high-risk sit- beverage or passing by one’s favorite ical difference exists between the first uations and by decreasing motivation bar), testing one’s personal control violation of the abstinence goal (i.e., to resist drinking in high-risk situations.

Vol. 23, No. 2, 1999 153 In many cases, initial lapses occur in up for situations with overwhelmingly nent drinker to purchase a bottle of high-risk situations that are completely high risk. These choices have been liquor “just in case guests stop by.” unexpected and for which the drinker termed “apparently irrelevant decisions” Marlatt and Gordon (1985) have is often unprepared. In relapse “set ups,” (AIDs), because they may not be overtly hypothesized that such decisions may however, it may be possible to identify recognized as related to relapse but nev- enable a person to experience the a series of covert decisions or choices, ertheless help move the person closer to immediate positive effects of drinking each of them seemingly inconsequential, the brink of relapse. One example of while disavowing personal responsibil- which in combination set the person such an AID is the decision by an absti- ity for the lapse episode (“How could

Self-monitoring and behavior assessment Revised (e.g., situational decision competency matrix test)

Increasing lifestyle balance (e.g., developing positive Efficacy-enhancing Lapse management Stimulus control addictions [jogging, Relapse road maps strategies (e.g., viewing (e.g., contract to limit techniques (e.g., meditation, "body time"] (i.e., analysis of change process as alcohol use, reminder removing all and substitute indulgences high-risk situations skills acquisition, cards with instructions items associated [e.g., recreational activities, and the available breaking down overall on how to cope with with alcohol use) massage]) choices) task into subtasks) a lapse)

Decreased self- Initial Urges Rationalization, Lack of efficacy; Abstinence Lifestyle Desire for High-risk substance and denial, and coping positive violation imbalance indulgence situation use (i.e., craving AIDs response outcome effect expectancies lapse)

Urge management Avoidance Coping-skills Cognitive (e.g., coping strategies training (e.g., restructuring (e.g., imagery, such as relaxation training, considering lapse a "urge surfing") stress management, mistake, not a and assertiveness result of personal training) and relapse failure) rehearsal

Education about warning signals of Analysis of Eliminating myths relapse (e.g., AIDs, relapse fantasies and placebo effects positive expectancies, and descriptions (e.g., education about lifestyle imbalances) of past relapses immediate vs. delayed effects of alcohol, use of decision matrix)

Figure 2 Covert antecedents and immediate determinants of relapse and intervention strategies for identifying and preventing or avoiding those determinants. Lifestyle balance is an important aspect of preventing relapse. If stressors are not balanced by sufficient stress management strategies, the client is more likely to use alcohol in an attempt to gain some relief or escape from stress. This reaction typically leads to a desire for indulgence that often develops into cravings and urges. Two cognitive mechanisms that contribute to the covert planning of a relapse episode—rationalization and denial—as well as apparently irrelevant decisions (AIDs) can help precipitate high-risk situations, which are the central determinants of a relapse. People who lack adequate coping skills for handling these situations experience reduced confidence in their ability to cope (i.e., decreased self-efficacy). Moreover, these people often have positive expectations regarding the effects of alcohol (i.e., outcome expectancies). These factors can lead to initial alcohol use (i.e., a lapse), which can induce an abstinence violation effect that, in turn, influences the risk of progressing to a full relapse. Self-monitoring, behavior assessment, analyses of relapse fantasies, and descriptions of past relapses can help identify a person’s high-risk situations. Specific intervention strategies (e.g., skills training, relapse rehearsal, education, and ) and general strategies (e.g., relaxation training, stress management, efficacy-enhancing imagery, contracts to limit the extent of alcohol use, and reminder cards) can help reduce the impact of relapse determinants. Shaded boxes indicate steps in the relapse process and intervention measures that are specific to each client and his or her ability to cope with alcohol-related situa- tions. White boxes indicate steps in the relapse process and intervention strategies that are related to the client’s general lifestyle and coping skills. High-risk situations are related to both the client’s general and specific coping abilities.

154 Alcohol Research & Health Relapse Prevention: Marlatt’s Cognitive-Behavioral Model

anyone expect me not to drink when RP Intervention Strategies relapse episodes and relapse dreams or there’s a bottle of liquor in the house?”). fantasies in order to identify situations The RP model includes a variety of in which the client has or might have Lifestyle Factors. Marlatt and Gordon cognitive and behavioral approaches difficulty coping. Several self-report (1985) have proposed that the covert designed to target each step in the relapse questionnaires also can help assess the antecedent most strongly related to process (see figure 2). These approaches situations in which clients have been relapse risk involves the degree of balance include specific intervention strategies prone to drinking heavily in the past in the person’s life between perceived that focus on the immediate determinants as well as the clients’ self-efficacy for external demands (i.e., “shoulds”) and of relapse as well as global self-manage- resisting future drinking in these situa- internally fulfilling or enjoyable activi- ment strategies that focus on the covert tions (Annis and Davis 1988; Annis ties (i.e., “wants”). A person whose life antecedents of relapse. Both the specific 1982a). Furthermore, clients who have is full of demands may experience a and global strategies fall into three main not yet initiated abstinence are encour- constant sense of stress, which not only categories: skills training, cognitive aged to self-monitor their drinking can generate negative emotional states, restructuring, and lifestyle balancing. behavior—for example, by maintaining thereby creating high-risk situations, an ongoing record of the situations, but also enhances the person’s desire for emotions, and interpersonal factors asso- pleasure and his or her rationalization Specific Intervention Strategies ciated with drinking or urges to drink. that indulgence is justified (“I owe myself The goal of the specific intervention Such a record allows clients to become a drink”). In the absence of other non- strategies—identifying and coping with more aware of the immediate precipitants drinking pleasurable activities, the person high-risk situations, enhancing self- of drinking. Even in clients who have may view drinking as the only means efficacy, eliminating myths and placebo already become abstinent, self-monitor- of obtaining pleasure or escaping pain. effects, lapse management, and cogni- ing can still be used to assess situations tive restructuring—is to teach clients to in which urges are more prevalent. Urges and Cravings. The desire for imme- anticipate the possibility of relapse and Once a person’s high-risk situations diate gratification can take many forms, to recognize and cope with high-risk have been identified, two types of and some people may experience it as a situations. These strategies also focus intervention strategies can be used to craving or urge to use alcohol. Although on enhancing the client’s awareness of lessen the risks posed by those situations. many researchers and clinicians consider cognitive, emotional, and behavioral The first strategy involves teaching the urges and cravings primarily physiolog- reactions in order to prevent a lapse client to recognize the warning signals ical states, the RP model proposes that from escalating into a relapse. The first associated with imminent danger—that both urges and cravings are precipitated step in this process is to teach clients is, the cues indicating that the client is by psychological or environmental stim- the RP model and to give them a “big about to enter a high-risk situation. uli. Ongoing cravings, in turn, may picture” view of the relapse process. Such warning signals to be recognized erode the client’s commitment to main- For example, the therapist can use the may include, for example, AIDs, stress taining abstinence as his or her desire metaphor of behavior change as a jour- and lack of lifestyle balance, and strong for immediate gratification increases. ney that includes both easy and diffi- positive expectances about drinking. As This process may lead to a relapse set- cult stretches of highway and for which a result of identifying those warning up or increase the client’s vulnerability various “road signs” (e.g., “warning sig- signals, the client may be able to take to unanticipated high-risk situations. nals”) are available to provide guidance. some evasive action (e.g., escape from Although they are often used inter- According to this metaphor, learning to the situation) or possibly avoid the changeably, the terms “urges” and “crav- anticipate and plan for high-risk situa- high-risk situation entirely. ings” can be associated with distinct tions during recovery from alcoholism The second strategy, which is possi- meanings. Thus, Marlatt and Gordon is equivalent to having a good road map, bly the most important aspect of RP, (1985) have defined an urge as a rela- a well-equipped tool box, a full tank of involves evaluating the client’s existing tively sudden impulse to engage in an gas, and a spare tire in good condition motivation and ability to cope with act such as alcohol consumption, whereas for the journey. specific high-risk situations and then craving is defined as the subjective desire helping the client learn more effective to experience the effects or consequences Identifying and Coping With High- coping skills. Relevant coping skills can of such an act. Nevertheless, the same Risk Situations. To anticipate and plan be behavioral or cognitive in nature processes may mediate both urges and accordingly for high-risk situations, the and can include both strategies to cope cravings. Two such processes have been person first must identify the situations 1 proposed: (1) conditioning elicited by in which he or she may experience dif- 1Classical or Pavlovian conditioning occurs when an stimuli associated with past gratifica- ficulty coping and/or an increased desire originally neutral stimulus (e.g., the sight of a beer tion and (2) cognitive processes associ- to drink. These situations can be identi- bottle) is repeatedly paired with a stimulus (e.g., alco- ated with anticipated gratification (i.e., fied using a variety of assessment strate- hol consumption) that induces a certain physiological response. After the two stimuli have been paired the expectancies for the immediate gies. For example, the therapist can repeatedly, the neutral stimulus becomes a conditioned pleasurable effects of alcohol). interview the client about past lapses or stimulus that elicits the same physiological response.

Vol. 23, No. 2, 1999 155 with specific high-risk situations (e.g., manageable subtasks that can be client to consider both the immediate and refusing drinks in social situations and addressed one at a time (Bandura 1977). the delayed consequences of drinking. assertive communication skills) and Thus, instead of on a distant Even when alcohol’s perceived positive general strategies that can improve cop- end goal (e.g., maintaining lifelong effects are based on actual drug effects, ing with various situations (e.g., medi- abstinence), the client is encouraged to often only the immediate effects are tation, anger management, and positive set smaller, more manageable goals, positive (e.g., euphoria), whereas the self-talk). such as coping with an upcoming high- delayed effects are negative (e.g., sleepi- Assessing a client’s existing coping risk situation or making it through the ness), particularly at higher alcohol skills can be a challenging task. Question- day without a lapse. Because an increase doses. Asking clients questions designed naires such as the situational confidence in self-efficacy is closely tied to achieving to assess expectancies for both immedi- test (Annis 1982b) can assess the amount preset goals, successful mastery of these ate and delayed consequences of drink- of self-efficacy a person has in coping individual smaller tasks is the best strat- ing versus not drinking (i.e., using a with drinking-risk situations. Those egy to enhance feelings of self-mastery. decision matrix) (see table, p. 157) often measures do not necessarily indicate, Therapists also can enhance self- can be useful in both eliciting and however, whether a client is actually efficacy by providing clients with feed- modifying expectancies. With such a able or willing to use his or her coping back concerning their performance on matrix, the client can juxtapose his or skills in a high-risk situation. To increase other new tasks, even those that appear her own list of the delayed negative the likelihood that a client can and will unrelated to alcohol use. In general, consequences with the expected posi- utilize his or her skills when the need success in accomplishing even simple tive effects. arises, the therapist can use approaches tasks (e.g., showing up for appointments such as role plays and the development on time) can greatly enhance a client’s Lapse Management. Despite precau- and modeling of specific coping plans for feelings of self-efficacy. This success can tions and preparations, many clients managing potential high-risk situations. then motivate the client’s effort to committed to abstinence will experi- change his or her pattern of alcohol use ence a lapse after initiating abstinence. Enhancing Self-Efficacy. Another and increase the client’s confidence that Lapse-management strategies focus on approach to preventing relapse and he or she will be able to successfully halting the lapse and combating the promoting behavioral change is the use master the skills needed to change. abstinence violation effect to prevent of efficacy-enhancement procedures— an uncontrolled relapse episode. Lapse that is, strategies designed to increase a Eliminating Myths and Placebo Effects. management includes contracting with client’s sense of mastery and of being Counteracting the drinker’s mispercep- the client to limit the extent of use, to able to handle difficult situations with- tions about alcohol’s effects is an contact the therapist as soon as possible out lapsing. One of the most important important part of relapse prevention. after the lapse, and to evaluate the situ- efficacy-enhancing strategies employed To accomplish this goal, the therapist ation for clues to the factors that trig- in RP is the emphasis on collaboration first elicits the client’s positive expecta- gered the lapse. Often, the therapist between the client and therapist instead tions about alcohol’s effects using either provides the client with simple written of a more typical “top down” doctor- standardized questionnaires or clinical instructions to refer to in the event of a patient relationship. In the RP model, interviews. Positive expectancies regard- lapse. These instructions reiterate the the client is encouraged to adopt the role ing alcohol’s effects often are based on importance of stopping alcohol con- of colleague and to become an objective myths or placebo effects of alcohol (i.e., sumption and (safely) leaving the lapse- observer of his or her own behavior. In effects that occur because the drinker inducing situation. Lapse management developing a sense of objectivity, the client expects them to, not because alcohol is presented to clients as an “emergency is better able to view his or her alcohol causes the appropriate physiological preparedness” kit for their “journey” to use as an addictive behavior and may be changes). In particular, considerable abstinence. Many clients may never need more able to accept greater responsibility research has demonstrated that alcohol’s to use their lapse-management plan, both for the drinking behavior and for perceived positive effects on social behav- but adequate preparation can greatly the effort to change that behavior. Clients ior are often mediated by placebo effects, lessen the harm if a lapse does occur. are taught that changing a habit is a resulting from both expectations (i.e., process of skill acquisition rather than a “set”) and the environment (i.e., “set- Cognitive Restructuring. Cognitive test of one’s willpower. As the client gains ting”) in which drinking takes place restructuring, or reframing, is used new skills and feels successful in imple- (Marlatt and Rohsenow 1981). Sub- throughout the RP treatment process menting them, he or she can view the sequently, the therapist can address to assist clients in modifying their attri- process of change as similar to other each expectancy, using cognitive restruc- butions for and perceptions of the situations that require the acquisition turing (which is discussed later in this relapse process. In particular, cognitive of a new skill. section) and education about research restructuring is a critical component of Another efficacy-enhancing strategy findings. The therapist also can use interventions to lessen the abstinence involves breaking down the overall task examples from the client’s own experi- violation effect. Thus, clients are taught of behavior change into smaller, more ence to dispel myths and encourage the to reframe their perception of lapses—

156 Alcohol Research & Health Relapse Prevention: Marlatt’s Cognitive-Behavioral Model

to view them not as failures or indicators strategies are designed to modify the and friends) have gradually been replaced of a lack of willpower but as mistakes client’s lifestyle to increase balance as by drinking as a source of entertain- or errors in learning that signal the need well as to identify and cope with covert ment and gratification. Therefore, for increased planning to cope more antecedents of relapse (i.e., early warn- one global self-management strategy effectively in similar situations in the ing signals, cognitive distortions, and involves encouraging clients to pursue future. This perspective considers lapses relapse set-ups). again those previously satisfying, non- key learning opportunities resulting from drinking recreational activities. In addi- an interaction between coping and sit- Balanced Lifestyle and Positive tion, specific cognitive-behavioral skills uational determinants, both of which . Assessing lifestyle factors training approaches, such as relaxation can be modified in the future. This associated with increased stress and training, stress-management, and time reframing of lapse episodes can help decreased lifestyle balance is an important management, can be used to help clients decrease the clients’ tendency to view first step in teaching global self-manage- achieve greater lifestyle balance. lapses as the result of a personal failing ment strategies. This assessment can be Helping the client to develop “posi- or moral weakness and remove the self- accomplished through approaches in tive addictions” (Glaser 1976)—that is, fulfilling prophecy that a lapse will which clients self-monitor their daily activities (e.g., meditation, exercise, or inevitably lead to relapse. activities, identifying each activity as a yoga) that have long-term positive “want,” “should,” or combination of effects on mood, health, and coping— both. Clients also can complete stan- is another way to enhance lifestyle bal- Global Lifestyle Self-Control dardized assessment measures, such as ance. Self-efficacy often increases as a Strategies the Daily Hassles and Uplifts Scale result of developing positive addictions, Although specific intervention strategies (Delongis et al. 1982), to evaluate the largely caused by the experience of suc- can address the immediate determi- degree to which they perceive their life cessfully acquiring new skills by per- nants of relapse, it is also important to stressors to be balanced by pleasurable forming the activity. modify individual lifestyle factors and life events. covert antecedents that can increase Many clients report that activities Stimulus-Control Techniques. exposure or reduce resistance to high- they once found pleasurable (e.g., hob- Although achieving a more balanced risk situations. Global self-control bies and social interactions with family lifestyle may reduce the risk of cravings

An Example of a Decision Matrix for Alcohol Abstinence or Alcohol Use*

Immediate Consequences Delayed Consequences

Positive Negative Positive Negative

Remain Improved self- Frustration and Greater control Not able to Abstinent efficacy and anxiety, denied over one’s life, enjoy drinking self-esteem, pleasures of better health while watching family approval, drinking, unable and longevity, sports, bored better health, to go to bars, learn about one’sand depressed, more energy, anger at not self and others not able to remain save money being able to without being friends with heavy- and time, greater do what one intoxicated, more drinking buddies success at work wants without respect from others “paying the price”

Resume Automatic pleasure, Feel weak from Maintain friendships Possible loss of Alcohol Use reduced stress drinking, risk of with drinking family and job, and anxiety, not accidents and buddies, able deterioration of feel pain, not worry embarrassment, to drink while health and early about one’s problems, anger of wife watching sports, death, loss of able to enjoy sports and family, arrive not have to nondrinking or and drink with buddies late to or miss cope with wife light-drinking work, hangovers, and family by friends, ridicule waste money staying out drinking by others, low self-esteem

*In such a matrix, the client lists both the positive and negative immediate and delayed consequences of remaining abstinent versus resuming drinking. This list can facilitate the client’s decisionmaking process regarding his or her future alcohol consumption.

Vol. 23, No. 2, 1999 157 and urges to use alcohol, urges and anticipate and accept these reactions as help identify coping responses that can cravings might still result from expo- a “normal” conditioned response to an be used to avoid a lapse at each point sure to conditioned stimuli previously external stimulus. According to this in the interaction. associated with drinking. Stimulus- approach, the client should not identify control techniques are relatively simple with the urge or view it as an indica- but effective strategies that can be used tion of his or her “desire” to drink. Theoretical and Practical to decrease urges and cravings in response Instead, the client is taught to label the Support for the RP Model to such stimuli, particularly during the urge as an emotional or physiological early abstinence period. Simply stated, response to an external stimulus in his Several studies over the past two these techniques encourage the client or her environment that was previously decades have evaluated the reliability to remove all items directly associated associated with heavy drinking, similar and predictive validity2 of the RP model with alcohol use from his or her home, to Pavlov’s dog, which continued to as well as the efficacy of treatment office, and car. This includes eliminat- salivate at the sound of a bell that had techniques based on this model. One ing, at least temporarily, all alcohol previously signaled food. recent large-scale research effort assess- supplies, including those typically kept In one clinical intervention based on ing the RP model was the Relapse for “guests,” as well as packing away this approach, the client is taught to Replication and Extension Project wine or shot glasses, corkscrews, and visualize the urge or craving as a wave, (RREP), which was funded by the similar items. Clients who used to hide watching it rise and fall as an observer National Institute on Alcohol Abuse or stash alcoholic beverages should and not to be “wiped out” by it. This and Alcoholism (Lowman et al. 1996). make a concerted effort to remember imagery technique is known as “urge This collaborative research project eval- and remove alcohol from all possible surfing” and refers to conceptualizing uated the reliability of raters’ catego- hiding places, because these hidden or the urge or craving as a wave that crests rizations of high-risk situations using forgotten bottles can serve as a power- and then washes onto a beach. In so Marlatt’s taxonomy and assessed ful temptation when found “acciden- doing, the client learns that rather than whether a prior situation could predict tally” after a period of sobriety. building interminably until they become future lapse episodes. Other, more subtle items that may overwhelming, urges and cravings peak As described earlier in this article, serve as conditioned cues for drinking and subside rather quickly if they are not the original relapse taxonomy sought to may include the favorite living room acted on. The client is taught not to categorize the environmental or emo- easy chair or the music the client typi- struggle against the wave or give in tional stimuli associated with an initial cally listened to while unwinding in to it, thereby being “swept away” or return to drinking in order to enhance the evening with several of his or her “drowned” by the sensation, but to the long-term effectiveness of aversion favorite drinks. In these cases, a tempo- imagine “riding the wave” on a surf therapy. The resulting taxonomy con- rary change in seating or listening board. Like the conceptualization of tained three levels of categorization of habits may be helpful while the client urges and cravings as the result of an high-risk situations with increasing develops alternative coping strategies. external stimulus, this imagery fosters specificity to help clinicians obtain Similarly, certain social events or other detachment from the urges and crav- detailed information about the causes high-risk situations may have become ings and reinforces the temporary and underlying each relapse episode. In the associated with excessive drinking to external nature of these phenomena. RREP study, researchers from three study such an extent that they may induce sites were trained in coding relapse classically conditioned urges or crav- Relapse Road Maps. Finally, therapists episodes. The researchers then coded key, ings, particularly in the early stages of can assist clients with developing or baseline, relapse episodes3 described abstinence. Accordingly, approaches relapse road maps—that is, cognitive- by study participants entering treat- that provide the client with a range of behavioral analyses of high-risk situa- ment at the study sites. The study avoidance strategies for turning down tions that emphasize the different addressed three major issues, as follows: invitations, leaving risky situations, or choices available to clients for avoiding otherwise avoiding problematic places or coping with these situations as well • It determined the inter-rater reliabil- or events also can serve as stimulus- as their consequences. Such a “mapping ity of the relapse episode coding— control measures that may help prevent out” of the likely outcomes associated a lapse. with different choices along the way can be helpful in identifying AIDs. For 2The term “reliability” refers to the ability of a test or method to provide stable results (e.g., when dif- Urge-Management Techniques. Even example, if arguments with a former ferent patients are compared or different investigators with effective stimulus-control proce- spouse are a high-risk situation, the rate the same patient). The term “predictive valid- dures in place and an improved lifestyle therapist can help the client map out ity” refers to the ability of a test or method to pre- balance, most clients cannot completely several possible scenarios for interacting dict a certain outcome (e.g., relapse risk) accurately. avoid experiencing cravings or urges to with the ex-spouse, including the likeli- 3The key relapse episode was defined as the most drink. Therefore, an important aspect hood of precipitating an argument in recent use of alcohol following at least 4 days of of the RP model is to teach clients to each scenario. The therapist can then abstinence (Longabaugh et al. 1996).

158 Alcohol Research & Health Relapse Prevention: Marlatt’s Cognitive-Behavioral Model

that is, whether different researchers ing an initial lapse. Specifically, those • RP is associated with “delayed emer- coded a given relapse episode in an participants who had a greater belief in gence effects”—that is, significant identical or similar manner. the disease model of alcoholism and a effects favoring RP as compared higher commitment to absolute absti- with other treatment approaches are • It evaluated whether the key relapse nence (who were most likely to experi- often found only at later followup episodes predicted the types of relapse ence feelings of guilt over their lapse) points (i.e., 1 year or more after episodes that study participants were most likely to experience relapse treatment) (Carroll 1996). This reported after undergoing treatment in that study. In a recent review of the delayed effectiveness may result (Maisto et al. 1996; Stout et al. 1996). literature on relapse precipitants, Dimeff from the fact that it takes time to and Marlatt (1998) also concluded that learn new skills and that conse- • It extended research on the RP model considerable support exists for the notion quently RP effects become more beyond the taxonomy by evaluating that an abstinence violation effect can obvious as patients acquire addi- alternative methods for assessing precipitate a relapse. tional practice. high-risk situations as well as evalu- Several recent review articles and ating the relative contribution of meta-analyses have examined the effec- • Although RP has been applied with negative affect, abstinence violation tiveness of treatments based on the RP some success to various addictive effect, coping, and expectancies on model in preventing relapse (Dimeff behaviors, the effects of RP-based the likelihood of relapse. and Marlatt 1998; Rawson et al. 1993; approaches are greatest in the treat- Carroll 1996; Irvin et al. 1999). The ment of alcoholism or multiple drug The results reported in the RREP RP-based treatments included in those use (Irvin et al. 1999). study indicate that the original relapse analyses were delivered both as stand- taxonomy of the RP model has only alone treatments for initiating abstinence • Combining RP with medications moderate inter-rater reliability at the and as adjuncts to other treatment pro- (e.g., disulfiram or naltrexone) to highest level of specificity, although grams. Although the reviews differ in treat alcoholism leads to improved reliability of the more general categories their methodology and in their criteria outcomes as compared with either (e.g., negative affect and social pres- for including or excluding certain treat- RP or medication alone (Irvin et al. sure) was better. The model’s predictive ments, the conclusions regarding over- 1999). validity also was modest; however, the all effectiveness of the RP approach are definition of the key relapse episodes similar. The findings can be summa- utilized in these studies failed to clarify rized as follows: Summary whether these were voluntary change episodes or simply a return to drinking The RP model of relapse is centered following a short period of abstinence • The studies conducted to date tend around a detailed taxonomy of emo- that did not represent a serious attempt to support the effectiveness of cogni- tions, events, and situations that can to quit drinking. Therefore, the RREP tive-behavioral RP-based approaches precipitate both lapses and relapses to studies do not represent a good test of in reducing the frequency of relapse drinking. This taxonomy includes both the predictive validity of the taxonomy. episodes as well as the intensity of immediate relapse determinants and Nevertheless, the study provides rel- lapse and/or relapse episodes among covert antecedents, which indirectly atively good support for other aspects of people who resumed alcohol use increase a person’s vulnerability to relapse. the RP model. For example, Miller and after treatment (Irvin et al. 1999). Based on the classification of relapse colleagues (1996) found that although The effectiveness of RP was particu- determinants and high-risk situations mere exposure to specific high-risk larly great in studies that compared proposed in the RP model, numerous situations did not predict relapse, the relapse rates in patients before and treatment components have been manner in which people coped with after treatment or that compared developed that are aimed at helping the those situations strongly predicted sub- patients receiving RP-based treatment recovering alcoholic cope with high-risk sequent relapse or continued abstinence. with controls receiving no treatment. situations. The results of recent research, Furthermore, in that study the major- particularly the RREP study, likely will ity of relapse episodes after treatment • Despite its benefits, RP-based treat- lead to modifications of the original RP occurred during situations involving ment is not associated with higher model, particularly with regard to the negative emotional states, a finding abstinence rates compared with other assessment of high-risk situations as that has been replicated in other studies valid treatment approaches (Carroll well as the conceptualization of covert (Cooney et al. 1997; McKay 1999; 1996; Irvin et al. 1999). RP-based and immediate antecedents of relapse. Shiffman 1992). Finally, the results of treatment is, however, often associ- Overall, however, research findings Miller and colleagues (1996) support ated with lower drinking rates and support both the overall model of the the role of the abstinence violation effect fewer drinking problems among relapse process and the effectiveness in predicting which participants would patients who have experienced a of treatment strategies based on the experience a full-blown relapse follow- relapse (e.g., Chaney et al. 1978). model. ■

Vol. 23, No. 2, 1999 159 References DELONGIS, A.; COYNE, J.C.; DAKOF, G.; FOLKMAN, behavior change. In: Davidson, P.O., and Davidson, S.; AND LAZARUS, R.S. Relationship of daily hassles, S.M., eds. Behavioral Medicine: Changing Health ANNIS, H.M. Inventory of Drinking Situations. Toronto, uplifts, and major life events to health status. Health Lifestyles. New York: Brunner/Mazel, 1980. pp. Canada: Addiction Research Foundation, 1982a. Psychology 1(2):119–136, 1982. 410–452.

ANNIS, H.M. Situational Confidence Questionnaire. DIMEFF, L.A.; AND MARLATT, G.A. Preventing MARLATT, G.A.; AND GORDON, J.R., EDS. Relapse Toronto, Canada: Addiction Research Foundation, relapse and maintaining change in addictive behav- Prevention: Maintenance Strategies in the Treatment of 1982b. iors. Clinical Psychology: Science & Practice 5(4):513– Addictive Behaviors. New York: Guilford Press, 1985. 525, 1998. ANNIS, H.M. A relapse prevention model for treat- MARLATT, G.A., AND ROHSENOW, D.J. The think- GLASER, W. Positive Addiction. New York: Harper ment of alcoholics. In: Miller, W.R., and Heather, drink effect. Psychology Today 15:60–93, 1981. N., eds. Treating Addictive Behaviors: Processes of & Row, 1976. MARLATT, G.A.; BAER, J.S.; AND QUIGLEY, L.A. Change. New York: Plenum Press, 1986. pp. IRVIN, J.E.; BOWERS, C.A.; DUNN, M.E.; AND WANG, 407–433. M.C. Efficacy of relapse prevention: A meta- Self-efficacy and addictive behavior. In: Bandura, analytic review. Journal of Consulting and Clinical A., ed. Self-Efficacy in Changing Societies. New ANNIS, H.M.; AND DAVIS, C.S. Assessment of Psychology 67:563–570, 1999. York: University Press, 1995. pp. 289–315. expectancies. In: Donovan, D.M., and Marlatt, G.A., eds. Assessment of Addictive Behaviors: LITMAN, G.K.; EISER, J.R.; RAWSON, N.S.B.; AND MARLATT, G.A.; BARRETT, K., AND DALEY, D.C. Behavioral, Cognitive, and Physiological Procedures. OPPENHEIM, A.N. Differences in relapse precipi- Relapse prevention. The American Psychiatric Press New York: Guilford Press, 1988. pp. 198–213. tants and coping behaviours between alcohol relapsers Textbook of Substance Abuse Treatment. 2d ed. and survivors. Behaviour Research and Therapy 1999. pp. 393–407. BANDURA, A. Self-efficacy: Toward a unifying theory 17:89–94, 1979. of behavioral change. Psychological Review 84(2): MCKAY, J.R. Studies of factors in relapse to alcohol, 191–215, 1977. LONGABAUGH, R.; RUBIN, A.; STOUT, R.L.; ZWIAK, drug and nicotine use: A critical review of method- W.H.; AND LOWMAN, C. The reliablity of Marlatt’s ologies and findings. Journal of Studies on Alcohol CAREY, K.B. Alcohol-related expectancies predict taxonomy for classifying relapses. Addiction 60:566–576, 1999. quantity and frequency of heavy drinking among 91(suppl.): 73–88, 1996. college students. Psychology of Addictive Behaviors MILLER, W.R.; WESTERBERG, V.S.; HARRIS, R.J.; 9(4):236–241, 1995. LOWMAN, C.; ALLEN, J.; STOUT, R.L.; AND The AND TONIGAN, J.S. What predicts relapse? Relapse Research Group. Replication and extension Prospective testing of antecedent models. Addiction CARROLL, K.M. Relapse prevention as a psychoso- of Marlatt’s taxonomy of relapse precipitants: Overview 91(suppl):155–172, 1996. cial treatment: A review of controlled clinical trials. of procedures and results. Addiction 91(suppl.): Experimental and Clinical Psychopharmacology 4: 51–72, 1996. RAWSON, R.A.; OBERT, J.L.; MCCANN, M.J.; AND 46–54, 1996. MARINELLI-CASEY, P. Relapse prevention models MAISTO, S.A.; CONNORS, G.J.; AND ZYWIAK, W.H. for substance abuse treatment. 30(2): Construct validation analyses on the Marlatt typol- CHANEY, E.R.; O’LEARY, M.R.; AND MARLATT, 284–298, 1993. G.A. Skill training with alcoholics. Journal of ogy of relapse precipitants. Addiction 91(suppl): Consulting and Clinical Psychology 46:1092–1104, 89–98, 1996. SHIFFMAN, S. Relapse process and relapse preven- 1978. MARLATT, G.A. Taxonomy of high-risk situations tion in addictive behaviors. Behavior Therapist: for alcohol relapse: Evolution and development of a 9–11, 1992. COONEY, N.L.; LITT, M.D.; MORSE, P.A.; BAUER, cognitive-behavioral model. Addiction 91(suppl): L.O.; AND GUAPP, L. Alcohol cue reactivity, STOUT, R.L.; LONGABAUGH, R.; AND RUBIN, A. 37–49, 1996. negative-mood reactivity, and relapse in treated Predictive validity of Marlatt’s relapse taxonomy alcoholic men. Journal of Abnormal Psychology 106: MARLATT, G.A.; AND GORDON, J.R. Determinants versus a more general relapse code. Addiction 243–250, 1997. of relapse: Implications for the maintenance of 91(suppl):99–110, 1996.

160 Alcohol Research & Health