Alcoholics Anonymous Outcomes and Benefits

These keywords were added by machine and not by the authors. This process is experimental and the keywords may be updated as the learning algorithm improves.

Introduction

Strong opinions, both pro and con, have been voiced about AA. It has been argued that AA is the most effective method to arrest alcoholism (e.g., Snyder, 1980). In contrast, it has also been argued that AA is helpful to only 5% of the people who choose to affiliate with the organization (Bufe, 1998). McCrady and Miller (1993) suggested that 1 in 10 Americans will attend a twelve-step meeting in their lifetime, but Bufe (1998) asserted that a majority of individuals who seek relief from alcohol-related problems by attending AA are coerced to do so, with fewer than 1 in 30 remaining in AA after 1 year. Finally, a majority of outpatient and inpatient alcohol treatment programs in the United States routinely include referral to AA, with one survey indicating that 79% of all Veteran Affairs substance abuse programs in the United States make such referrals (Humphreys et al., 1999). Mandated AA attendance, however, has been successfully challenged as unconstitutional in the United States because of AA’s heavy emphasis upon spirituality although the Supreme Court has demonstrated unusual leniency in upholding this decision, e.g., levied a one-dollar fine. Ironically, then, while AA formally eschews public controversy, it has been a lightening rod for conflict among professionals and laypersons regarding the treatment of alcoholism.

The study of the effectiveness of AA has a long and checkered history, with the first empirical paper on the effectiveness of AA appearing in 1945. By the early 1990s there were about 117 empirical papers on AA (Emrick et al., 1993), and today the number of peer-reviewed papers has steadily increased into the hundreds. In spite of this intense empirical focus on AA-related processes and benefits, however, substantial controversy remains about the basic usefulness of the organization to aid problematic drinkers. Not in dispute, however, alcoholics regard AA to be one of the most important and accessible resources for alcohol problems. We know, for example, that a majority of substance abusers who present for treatment will have had some AA exposure (Fiorentine & Hillhouse, 2000) and that between 56% (Humphreys et al., 1999) and 75% (Tonigan, 2001) of adults attending treatment will attend AA afterwards, regardless of the therapeutic orientation of the treatment program.

Figure 1.

Frequency of empirical AA publications 1950–2000.

A fundamental thesis in this chapter is that there have been three eras of AA investigations and that the vast empirical AA literature can be best understood and applied by highlighting the underlying themes and assumptions held by investigators within each era. Figure 1 shows the frequency of empirical AA publications 1950–2000, and it is offered to show the accelerating nature of AA-related research. We anticipate that the number of empirical AA investigations will continue to increase. For this reason, a secondary yet important objective of this chapter is to educate readers how to evaluate the credibility and generalizability of past and future published AA investigations.

It is important to distinguish between community-based AA—the focus of this chapter—and formal twelve-step therapy. While these two resources share some common objectives, e.g., endorsing prescribed AA behaviors like attending AA meetings, reading core AA literature, and working the twelve steps, the differences between AA and twelve-step therapy are substantial in content, structure, and process (Emrick et al., 1993). With this said, nearly all studies investigating community-based AA processes and outcomes do so from the perspective of recruiting alcohol-dependent persons entering formal substance abuse treatment. While this self-selection bias appears, at first glance, significant, naturalistic studies are beginning to document that considerable bidirectional migration occurs between twelve-step therapy and community-based AA (Timko et al., 2000; Fiorentine & Hillhouse, 2000; Bogenschutz & Tonigan, 2007). It appears, then, that findings based on treatment-seeking samples may offer more generalizability than originally thought.

This chapter is intended to provide a general overview of the effectiveness of AA. The primary goal is to provide readers with an evidence-based review of what is currently understood about AA-related outcomes and to do so by placing AA-related investigations into historical context. At the outset, it is important to stress that space limitations prevent a full and comprehensive description of the many AA-outcome-related studies. This chapter thus selectively samples from the large pool of AA studies with the intention of providing the clearest exemplars of studies holding specific assumptions about what constitutes AA exposure and AA-related outcome. To achieve this objective, this chapter is organized into four sections. Section 2 defines and describes the themes and assumptions of the three eras of AA research. Emphasis is placed on elucidating how era-specific assumptions influenced answers to the basic question, is AA effective? Section 3 details what is currently known about the effectiveness of AA. Here, findings will be divided according to the time era that generated them and secondary measures of AA-related outcome. The chapter ends with a brief summary of key points in the chapter.

Three Eras of AA-Related Outcome Studies

Our assumptions about the nature of a phenomena shape how we frame our research questions about the phenomena. Well understood, then, our assumptions can strongly influence, in advance, the answers to our questions. Shifts in our assumptions (Kuhn, 1962) rarely occur through sudden consensus on a new set of basic assumptions, however. It is important to note, therefore, that while the three eras of AA-outcome studies are placed in temporal sequence, the exact years of each era are arbitrary and probably irrelevant. The first era, 1945 to circa 1988, largely eschewed AA-related processes and defined AA-related outcome in categorical terms, i.e., abstinence versus non-abstinence. The four underlying research assumptions in this era are as follows: (1) the “dose” of AA was fixed and invariant across meetings regardless of meeting type, size, and membership characteristics; (2) AA group social dynamics or context did not influence the generation, transmission, or the reception of the AA “dose”; (3) AA-related social context itself did not account for drinking outcome, directly or indirectly; and (4) the importance of the AA “dose” was temporally invariant (e.g., “unit benefit” of one AA meeting was the same for AA members regardless of membership longevity). Clearly, the foundational phase of AA investigations considered AA to be a “black box,” one that was relatively unimportant to understand when explaining AA outcome and benefit. With some notable exceptions (Vaillant, 1983), studies of AA-related outcome in this era were typically cross-sectional in design, yielding correlational findings that were often (and inappropriately) used to infer causal relationships.

A number of factors converged in the late 1980s to mobilize a re-evaluation of how AA exposure was defined and how AA-related outcome would be measured. Beyond the scope of this chapter, two important factors catalyzing this movement were increased commitment by the National Institute on Alcohol Abuse and Alcoholism (NIAAA) to fund AA-related research and the formation of a cadre of investigators disenchanted with the poor fit between assumptions guiding past AA research and observations of AA-related processes. This new era lasted between 1989 and 2001, and its relatively short duration can be traced to its important but limited objectives. Specifically, enlargement of the focus of AA studies to include AA-related processes necessarily involved the development of psychometrically valid measures of new constructs. Black box models of AA relied exclusively on counts of AA meeting attendance. New psychometrically valid measures were therefore required to describe the plethora of prescribed AA-related behaviors, beliefs, and practices. Related, interest turned to how, if at all, constructs representing AA-related processes changed over time, changed in their relationship to one another, and, finally, how AA processes predicted subsequent substance use. Tacitly, then, longitudinal designs became increasingly desirable and feasible with NIAAA support.

While assumptions about AA and its processes were not explicit in this second phase of AA research (and in many ways sustained earlier assumptions), what defined this era was the strong emphasis on scientific rigor and measurement, hence establishing the legitimacy of AA-related research to the larger field of addiction research. While phase 1 established a waster shed of correlational AA-outcome research, it was phase 2 that provided the necessary and firm foundation that established that, for many problem drinkers, AA produced positive outcomes. Here, AA outcome became defined recognizing the multidimensional nature of alcohol use. Measures depicting intensity of drinking, frequency of drinking, days elapsed between drinking, consequences associated with drinking, and so forth all came to be applied in describing how, if at all, AA-related benefit may occur and for whom.

Phase 3, 2002 to present, is taking the next logical step by investigating optimal and suboptimal response profiles to AA exposure and to discern how and why AA-related benefit occurs for some people but not others. This most recent trend in AA-outcome studies offers high clinical yield, and several chapters in this section are devoted to presenting current findings on the moderating and mediating effects surrounding AA participation. For this reason, AA-related studies focusing on these important issues will not be discussed in this chapter. Perhaps most important to this movement is the assumption that behavior change (hence outcome) occurs in a dynamic context and is itself dynamic. This emphasis on ecological validity is an important contribution, one made possible because of advanced statistical techniques such as multilevel analysis (e.g., Raudenbush & Bryk, 2002) and structural equation modeling (e.g., Byrne, 1994) that model the trajectory of individual change over time.

Two additional and related assumptions are beginning to emerge in phase 3 of AA-outcome research. First is the recognition that alcohol abuse and dependence is a chronic disorder that may extend for decades. This perspective necessarily redefines the AA experience in the life of problematic drinkers, thus recognizing that the process of individuals moving into, through, and out of AA may recapitulate itself many times over the course of an individual’s lifetime. Second, and related, “outcome” is becoming more broadly defined, including such behaviors as good citizenship and quality of life. While not specifically relevant to substance use, such behaviors, attitudes, and beliefs may offset substance use relapse and hence be important AA-related outcomes.

Empirical-Based Review of AA Effectiveness

Early AA Studies

Figure 2 displays the relationship between AA attendance and abstinence reported in 33 studies, all of which were conducted between 1945 and 1990. Overall, this combined literature indicates that a moderate and positive association was present between frequency of AA attendance and increased abstinence (r _w = .21). Importantly, using meta-analytic techniques it is also shown that the variability in the strength of the association reported between the 33 studies was the result of sampling error. Stressing this latter point, the variability in study findings about the association of interest reflected differences in measures, follow-up intervals, sample characteristics, and statistical techniques, not differences in the magnitude of benefit associated with AA exposure.

Figure 2.

Correlations between frequency of AA meeting attendance and measures of abstinence in 33 studies.

Some caution should be exercised in the interpretation of the overall positive relationship between AA exposure and increased abstinence shown in Figure 2. Described by Emrick et al. (1993), nearly half of the studies (47.9%) in this initial period of AA investigations did not report any reliability information, for example, and about 65% used a cross-sectional or posttest only design. Generally, intact groups were studied (68%), and the use of collaterals to verify self-report was uncommon (31%) and the use of biological markers of substance use was even rarer (13%). Finally, the finding of a positive relationship between AA and abstinence for many, but not all, of the studies in Figure 2 was made within the context of posttreatment functioning. How, if at all, the association may be moderated by formal treatment was largely ignored during this era.

It is instructive to review a typical AA study from this genre. McCown (1989) investigated the relationship between impulsivity, empathy, AA exposure, and abstinence. Here, mutual-help exposure was defined by a composite measure reflecting number of hours per week attending meetings, working with others, sponsoring and being sponsored, and involvement in “community outreach” activities. The study used a cross-sectional design wherein 150 questionnaires were distributed at three twelve-step clubs. Sampling procedures were not described, although it appeared that two twelve-step members distributed and collected the surveys. Only 65% of the questionnaires included sufficient data to be included in the analyses, and no effort was made to corroborate self-reported abstinence. The author reported that empathy and hours per week in mutual-help activities was significantly and positively related, r = .24, p < .01. McCown concluded that either people who are more empathic get involved in prescribed twelve-step activities or getting involved in twelve-step activities produces increased empathy. No sample information, however, was reported in the article about the average length of abstinence of the AA members, their longevity in twelve-step programs, or even basic demographics, e.g., gender. As a result, the extent of sample self-selection cannot be determined, and it is also problematic to discern which alternative provided by McCown for the positive relationship between twelve-step exposure and empathy is the more likely.

A few randomized studies including direct assignment to AA were conducted in the early phase of AA research (e.g., Ditman et al., 1967, Brandsma, Maultsby, & Welsh, 1980). Ditman et al., for example, randomized 301 adults to one of three groups: no treatment, Alcoholics Anonymous, and “alcoholic” treatment. All participants were chronic alcohol offenders defined as two drunken arrests in the past three months or three arrests for drunken behavior in the past year. A bench warrant was issued for non-compliant participants, and they reported to the court 6 months after randomization. Recidivism, not drinking, was the primary outcome measure to assess between-group differences. Recidivism rates did not differ at 12 months between the three groups, and Ditman et al. concluded that forced referral to AA or treatment was ineffective relative to the no-treatment control group. This study, one of the few that randomized to AA, is often cited as demonstrating that AA is ineffective. Strengths of the study include the RCT design, an 80% follow-up rate, and an outcome measure that was verifiable through court records. Obvious weaknesses include an insensitive outcome measure and poor monitoring of treatment compliance and AA attendance (for all groups). For a review of 21 RCT studies that included AA referral—in all cases coerced—see Kownacki and Shadish (1999).

Maturing AA Studies 1989–2001

Numerous longitudinal and well-designed studies using psychometrically strong measures approached the question of AA-related outcome in the second generation of AA research (e.g., PMRG, 1997, 1998; Ouimette, Moos, & Finney, 1998; Humphreys, Huebsch, Finney, & Moos, 1999; McKeller, Stewart, & Humphreys, 2003 Timko et al., 2000; Fiorentine & Hillhouse, 2000; Tonigan, 2001; Connors, Tonigan, & Miller, 2001). Generally, but not always, these investigations were conducted within the context of larger clinical trials investigating the effectiveness of twelve-step therapy in relation to other, more research supported, therapeutic orientations, e.g., cognitive behavioral therapy. Noteworthy especially, more elaborate and reliable measures of AA exposure, drinking behavior, and secondary outcomes began to be used in this era. Measurement of AA exposure, for instance, now routinely included AA program, e.g., reading core literature and progress in working the twelve steps (e.g., Tonigan et al., 2001; Gilbert, 1991), and AA fellowship dimensions, e.g., seeking support from other AA members and having and being a sponsor (Kaskutas, Bond, & Humphreys, 2002;Tonigan, Connors, & Miller, 1996; Morgenstern et al., 19XX). In tandem, interviewer-based and self-report measures of daily drinking behavior became more detailed and critically evaluated, now taking into account frequency, intensity, and duration of abstinent and drinking days (Sobell & Sobell, 2004). Secondary measures of outcome also gained momentum in application and included such dimensions as adverse alcohol-related consequences (e.g., Miller et al., 1995), ordinal measures of clinical outcomes (e.g., Zweben & Cisler, 1996), and purpose in life (Crumbaugh, & Henrion, 1988). Combined, the foundation was rapidly established to identify what AA practices, if any, produced changes, if any, in specific dimensions of drinking and life functioning.

Two thorny issues remained in declaring causal relationships between AA exposure and drinking, issues only partially addressed in earlier AA research. First, AA exposure was, by definition, self-selective. This aspect of “voting with the feet” plagued causal inferences about AA-related benefit. In response, in phase 2, AA-outcome studies began to statistically control for individual state and trait variables that may confound the study of AA-related benefit, e.g., motivation for change (McKeller et al., 2003; Kelly, Stout, Zywiak, & Schneider, 2006), alcohol impairment (Timko et al., 2000), and psychiatric problem severity (e.g., Connors et al., 2001). Second, the demonstration of causality required, at a minimum, appropriate temporal ordering of AA exposure and drinking. For this reason, studies during this era increasingly adopted longitudinal designs with frequent and extended follow-ups (Moos & Moos, 2006; Tonigan et al., 2001; Fiorentine & Hillhouse, 2000; Ouimette, Moos, & Finney, 1998).

With this background, what was concluded about outcomes associated with AA exposure? Looking first at AA meeting attendance alone, Tonigan (2001) reported in a multisite clinical trial with 1,726 outpatient and aftercare clients that AA attendance for the first 3 months after treatment was correlated, on average, with percent days abstinent for months 9–12 posttreatment, r = .25. With correction for measurement attenuation, this overall relationship increased to r = .31. Variability in the magnitude of this relationship between the 11 sites ranged from .14 to .33, but such variation reflected sampling error and not differences in the magnitude of AA benefit.

Ouimette et al. (1998), this time after statistically controlling for the effects of prior inpatient substance abuse treatment and Axis I diagnosis, reported similar findings about the benefit of AA attendance. Specifically, among 3,018 patients recruited at 15 VA inpatient treatment programs, frequency of AA meetings attended during and after treatment and 12-month abstinence was significantly and positively correlated, r = .34. What of AA outcome without prior or concomitant treatment? Timko et al. (2000) conducted an 8-year naturalistic study that monitored the help-seeking behaviors of 466 problem drinkers. Focusing on those individuals who sought no help (n = 78) or sought help only from AA (n = 66), strong evidence was provided about AA benefit. Specifically, at 1-year follow-up, 47.5% of the AA group reported total abstinence while 19.6% of the no-help-seeking group reported complete abstinence. This advantage was manifest at all follow-ups, and at the 8-year follow-up, 48.5% of the AA only group reported abstinence while only 25.6% of the no-help group reported complete abstinence. While the three studies reviewed had very different aims and sampled from different populations of substance abusers, they consistently demonstrated that AA is beneficial when outcome is defined using measures reflecting abstinence, as either a continuous or a categorical measure.

Replicated in several studies, measures of AA commitment and participation are stronger predictors of later abstinence relative to counting prior AA meeting attendance. Clearly, attending and becoming involved in AA are distinct, and engagement in the AA program and fellowship has been found to be associated with the strongest associations with later abstinence. In an early study evaluating the posttreatment drinking status of a sample of inpatients, for example, Montgomery et al. (1995) reported that AA attendance and a composite measure reflecting AA involvement predicted total alcohol consumed at 7-month follow-up to be r = –.23, n.s. and r = –.44, p < .05, respectively. Of some import, extent of involvement in AA remained significantly and negatively associated with total amount of alcohol consumed even after statistically controlling for frequency of AA meeting attendance. This finding is clinically important, and it appears to replicate across twelve-step sister programs. Weiss et al. (2005), for example, recruited and investigated the mutual-help activities of 336 cocaine-dependent adults. Here, twelve-step meeting attendance (AA and NA) for months 1–3 was not predictive of days cocaine use or the ASI drug composite score collected months 4–6. In contrast, extent of investment in twelve-step prescribed activities, e.g., reading twelve-step literature, speaking with a sponsor, step work, and speaking at meetings, was significantly related with later reductions in days cocaine use and ASI drug use scores, with more investment being associated with more positive outcomes. Interestingly, the distinction between the twelve-step investors and non-investors was stark, and readily and clinically observable. In particular, the high investors (52.9% of the sample) reported 5.5 days per week of twelve-step-related activities while low investors reported 0.5 days per week of twelve-step activities.

More recently, using structural equation modeling to reflect AA participation as a latent construct that included multiple dimensions, e.g., meeting attendance, reading AA literature, number of AA friends, working the twelve steps, Connors et al. (2001) and McKeller et al. (2003) arrived at similar conclusions. First, to the surprise of few, both investigative teams found that the experiencing of AA is multidimensional and that there were strong statistical justifications to depict the experience as such, e.g., both teams found empirically good measurement models reflecting AA participation. Second, both teams reported that the latent construct representing AA participation was a strong and significant predictor of positive AA outcome. Specifically, in the McKeller study, AA participation at 1 year was significantly predictive of reductions in alcohol-related problems at the 2-year follow-up, while Connors et al. (2001) reported that AA participation at 9-month follow-up was significantly predictive of increased abstinence at 12 months posttreatment and that this positive benefit was similar for clients who had previously attended outpatient and aftercare treatments.

At this point, it would be easy to infer that AA attendance is, in relative terms, unimportant in predicting positive outcome. This impression would be erroneous. The relationship between attendance and investment in AA is complex, and the nature of this relationship appears to vary according to one’s current trajectory of AA exposure. Specifically, contrary to conventional wisdom, it seems that attendance and commitment to AA are not linearly related during early AA affiliation (Tonigan, Connors, & Miller, 2003). Among aftercare and outpatient clients, for example, immediately after treatment, AA attendance and commitment were strongly and positively related to a threshold of meeting attendance, after which increased attendance was not associated with parallel increases in doing prescribed AA-related behaviors. In this particular study, diminishing returns on AA investment occurred when AA attendance occurred on more than 66% of the available days. Evidence also suggests that the relationship between AA attendance and commitment may be moderated by AA member characteristics. Unlike adults, for instance, frequency of AA meeting attendance among a large number of adolescent substance abusers (n = 2, 317) receiving treatment was a significant, strong, and positive predictor of complete abstinence at 6- and 12-month follow-ups (Hsieh et al., 1998). In yet another adolescent inpatient sample (n = 74), Kelly et al. (2000) reported that frequency of twelve-step meeting attendance actually outperformed a composite measure of twelve-step participation in predicting days abstinence 4–6 months after treatment.

Secondary Measures of AA Outcome

The benefits of AA are less clear (or dramatic) when measures other than increased abstinence are considered the outcome index. Tonigan (2001) found that frequency of AA meeting attendance immediately after treatment was unrelated to later changes in alcohol-related problems at 12-month follow-up. The opposite finding was reported by McKeller et al. (2003) when, using a composite measure of AA participation, AA exposure did predict reductions in alcohol-related consequences at 2-year follow-up. It is not clear whether differences in these findings reflect the higher predictive value of a comprehensive measure of AA exposure or, alternatively, whether differences in declaring AA benefit were the result of different follow-up intervals.

Another secondary outcome measure of interest among AA researchers is psychological well-being. Core AA literature supports the prediction that emotional distress, fear of economic insecurity, and anxiety will fade as one proceeds through the prescribed twelve steps (e.g., AA promises). In fact, AA members make a clear distinction between abstinence and sobriety, with the basic belief being that sobriety is the higher objective and includes spiritual growth, serenity, and emotional well-being. In combing the results from 13 studies, all of which were completed before 1990, Emrick et al. (1993) concluded that frequency of AA attendance and psychological adjustment were modestly yet positively related, r = .25. Noteworthy, combining the results across these (mostly) cross-sectional studies involved collapsing measures of depression, anxiety, and so forth and thus may have underestimated the relationship of interest.

More recent longitudinal studies suggest that AA attendance and participation may have a more modest effect on psychological well-being than originally thought. In particular, as part of a larger study comparing older (age 55–77) and younger male alcohol inpatient treatment outcomes, Lemke and Moos (2003) reported that number of AA meetings in the first year after treatment was significantly albeit modestly related with psychological distress, r = –.11. Having an AA sponsor likewise was negatively and significantly related with 1-year self-reported psychological distress, r = –.12. If AA produces changes in psychological adjustment, then alcohol-dependent patients who attend both formal treatment and AA should report larger gains in psychological adjustment relative to problem drinkers attending formal treatment alone. Humphreys et al. (1994) tested this prediction using a sample of African American alcohol-dependent adults seeking public substance abuse treatment. Both groups reported significant and large reductions in psychological severity symptoms on the ASI at 1-year follow-up, but this reduction was unrelated to whether or not participants attended AA. Better drinking outcomes were, however, found for those participants who attended AA in addition to formal treatment relative to those participants who did not attend AA after treatment.

Related, investigators have also focused on whether, as a result of sustained AA exposure, one experiences an enhanced sense of life purpose. In a cross-sectional design that sampled 20 AA meetings, for example, Carroll (1993) concluded that life purpose increases with working step 11 (e.g., the practice of prayer and meditation) and longevity in AA or length of sobriety. The self-selective nature of this study, however, renders these findings suspect. Two longitudinal studies came to different conclusions about changes in life meaning among AA-exposed drinkers, and they used the same measure of life purpose developed by Crumbaugh and Henrion (1988). Montgomery et al. (1995) reported that a composite measure of AA involvement collected at 3 months predicted self-reported purpose in life at the 7-month follow-up (r = .48). This finding was based on the posttreatment functioning of an inpatient sample of alcoholics, and findings were derived from an admirable 82% follow-up rate. In contrast, Tonigan (2001) reported that purpose in life collected at a 12-month interview and frequency of AA attendance for the first 3 months after outpatient (N = 952) and aftercare (N = 774) treatment were not related (r = .04) and that this finding was consistent across 11 treatment sites in Project MATCH.

A secondary measure of tremendous importance for both the individual and the society is the influence, if any, of AA participation on tobacco use. Well known, tobacco use is three times higher among substance abuses relative to the general population, and substance abusers are significantly less likely to be successful in their quit efforts relative to non-substance abusing quitters (Bobo & Davis, 1993). Ironically, however, tobacco use probably presents a higher fatality risk factor for alcoholics than does alcohol use. Sadly, there are only a handful of investigations that focus on changes in tobacco use among AA members. Conceptually, the strong emphasis in AA on abstinence from mind-altering chemicals would lead one to predict large reductions in tobacco use among AA members (although, ironically, this insistence may also work against the use of effective nicotine replacement methods). Practically, however, the insistence on abstinence from alcohol may trump the use of tobacco, with some AA members arguing that efforts at smoking cessation would jeopardize abstinence from alcohol.

What does the research have to say? As background, Figure 3 shows the average number of cigarettes used per day over a 6-month period by two AA-exposed groups (Tonigan, R01AA014197, NIAAA). To magnify the potential benefits of AA exposure, this sample was divided into high and low AA exposure groups, with high AA exposure defined as attending AA at least once every 5 days for 6 months. All study participants had minimal prior AA and treatment exposure. On average, the high AA group reported lower levels of tobacco use at baseline and the two follow-ups, but the rate of change in tobacco use over the 6 months of assessment did not differ between the two AA-exposed groups. In a naturalistic setting, then, it appears that the use of tobacco may not change during early AA affiliation.

Figure 3.

Cigarette use over time by two self-selected groups of AA attendees.

While tobacco cessation may not be a normative behavior among AA members, there is some evidence that the prescribed program and practices of AA can be useful in reducing tobacco use. Bobo and Davis (1993), for example, reported that 31% of the recovering alcoholics working in treatment facilities in the Midwest applied the principles of AA in their successful quit efforts. Likewise, Patten, Martin, Calfas, Lento, & Wolter (2001) reported that having an active relationship with an AA sponsor was predictive of increased tobacco abstinence, but not under all kinds of smoking inventions. In particular, Patten et al. found that the sponsorship relationship was beneficial when combined with a standard smoking cessation program that did not include nicotine gum replacement (40% not smoking). Of the three interventions under consideration, the standard 20-day quit program was least beneficial for those participants without a sponsor (10% not smoking). Finally, in a study of recovering alcoholics (96% AA members), Hughes et al. (2003) reported that use of the nicotine patch was efficacious relative to a placebo patch, and importantly this positive effect occurred regardless of length of sobriety.

AA exposure appears beneficial regardless of treatment orientation when abstinence is the measure of outcome. Two independent studies, however, have reported that the intensity of a relapse was stronger for AA-exposed people after they have received cognitive behavioral therapy (relative to twelve-step or motivational enhancement therapies). In particular, in the Project MATCH outpatient sample, Tonigan et al. (2003) reported that there was a strong negative association between AA attendance and number of drinks consumed when drinking occurred among twelve-step clients at both the proximal (months 1–6) and distal (months 7–12) follow-ups. To a lesser degree, this relationship was also found among clients assigned to the motivational enhancement condition. In contrast, clients assigned to the cognitive behavioral group reported a positive relationship between drinking intensity and AA attendance during proximal follow-up (i.e., those attending more AA meetings were actually drinking more). This relationship did not persist into distal follow-up.

Humphreys et al. (1999) reported similar findings based on a large VA alcohol-dependent sample, with clients attending cognitive behavior-based inpatient treatment programs drinking more intensely when attending AA in early follow-up. Interpretation of this finding is difficult because the data in these studies does not permit ascertainment of the temporal order between drinking and AA attendance. Although conflicting cognitive behavioral and AA ideologies may have resulted in more drinking, it is equally plausible that clients who were faring poorly (drinking heavily) in cognitive behavioral therapy elected to attend AA as another attempt to alter their drinking—that is, they sought out AA because they continued to drink more heavily.

Conclusions

AA is beneficial for many, but not all, problematic drinkers. Described in this chapter, the findings of early correlation-based AA studies have been largely replicated using more sophisticated, prospective longitudinal studies. It is important to stress that the magnitude of benefit is modest and that such benefit is most clearly evidenced in measures of alcohol abstinence. Arguably, our estimates of the magnitude of AA benefit may be conservative. Several moderators of AA-related benefit have been identified (e.g., Tonigan et al., 2002), for instance, and our pooling of optimal and suboptimal responders to AA in order to derive global estimates of benefit provides biased estimates of drinking reduction.

The weight of evidence indicates that facilitating AA attendance is important for sustaining abstinence but that such attendance ought not be considered an end in itself. Rather, AA attendance ought to be viewed as a bridge to the practice and internalization of prescribed AA-related behaviors and beliefs. With a few exceptions, AA-focused studies show that commitment to, and practice of, prescribed AA behaviors is a stronger predictor of later abstinence than sheer frequency of AA meeting attendance. Little is known, however, about the factors that may influence readiness for engagement into AA-related practices and beliefs. Investigation of these factors offers an important step to understanding how and why AA is beneficial.

Core AA literature suggests that deep psychological transformations will occur as problem drinkers progress through the prescribed program of AA. Investigations of secondary AA-related benefit only partially support this claim. Clearly, drinking reduction (with or without AA exposure) is associated with improvement in a number of psychological measures, e.g., reduction in depression. Furthermore, members of AA do report positive albeit very modest improvements in quality of life and purpose in life. Whether these changes are the result of AA participation and/or simply increased abstinence is unclear and certainly warrants investigation. Currently, there is little research on AA participation and tobacco use. The few studies that have been conducted suggest that AA-related practices may support quit efforts and, equally important, that formal smoking interventions can be successfully integrated into the larger gestalt of the AA experience.