journal of addictive diseases alcoholics anonymous effectiveness

PLEASE SCROLL DOWN FOR ARTICLE

This article was downloaded by: [Brown University]On: 9 April 2010Access details: Access Details: [subscription number 784168974]Publisher RoutledgeInforma Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer House, 37-41 Mortimer Street, London W1T 3JH, UK

Journal of Addictive DiseasesPublication details, including instructions for authors and subscription information:http://www.informaworld.com/smpp/title~content=t792306884

Alcoholics Anonymous Effectiveness: Faith Meets ScienceLee Ann Kaskutas a

a School of Public Health, University of California-Berkeley, Berkeley, California

To cite this Article Kaskutas, Lee Ann(2009) 'Alcoholics Anonymous Effectiveness: Faith Meets Science', Journal ofAddictive Diseases, 28: 2, 145 — 157To link to this Article: DOI: 10.1080/10550880902772464URL: http://dx.doi.org/10.1080/10550880902772464

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

http://www.informaworld.com/smpp/title~content=t792306884

http://dx.doi.org/10.1080/10550880902772464

http://www.informaworld.com/terms-and-conditions-of-access.pdf

Journal of Addictive Diseases, 28:145–157, 2009Copyright c© Taylor & Francis Group, LLCISSN: 1055-0887 print / 1545-0848 onlineDOI: 10.1080/10550880902772464

Alcoholics Anonymous Effectiveness:Faith Meets Science

Lee Ann Kaskutas, DrPH

ABSTRACT. Research on the effectiveness of Alcoholics Anonymous (AA) is controversial andsubject to widely divergent interpretations. The goal of this article is to provide a focused review ofthe literature on AA effectiveness that will allow readers to judge the evidence effectiveness of AA forthemselves. The review organizes the research on AA effectiveness according to six criterion requiredfor establishing causation: (1) magnitude of effect; (2) dose response effect; (3) consistent effect; (4)temporally accurate effects; (5) specific effects; (6) plausibility. The evidence for criteria 1- 4 and 6 isstrong: rates of abstinence are about twice as high among those who attend AA (criteria 1, magnitude);higher levels of attendance are related to higher rates of abstinence (criteria 2, dose-response); theserelationships are found for different samples and follow-up periods (criteria 3, consistency); prior AAattendance is predictive of subsequent abstinence (criteria 4, temporal); and mechanisms of actionpredicted by theories of behavior change are present in AA (criteria 6, plausibility). However, rigorousexperimental evidence establishing the specificity of an effect for AA or Twelve Step Facilitation/TSF(criteria 5) is mixed, with 2 trials finding a positive effect for AA, 1 trial finding a negative effect forAA, and 1 trial finding a null effect. Studies addressing specificity using statistical approaches have hadtwo contradictory findings, and two that reported significant effects for AA after adjusting for potentialconfounders such as motivation to change.

KEYWORDS. Alcoholics Anonymous (AA), 12-step, self-help, mutual aid, outcomes

INTRODUCTION

Research on the effectiveness of AlcoholicsAnonymous (AA) is controversial and subjectto widely divergent interpretations. For example,the Cochrane Group published a review of theAA literature that considered outcome studies ofAA and of 12-step facilitation (TSF), a form ofspecialty treatment that introduces clients to the12-step philosophy and support system. Theirreview recommended that people considering at-tending AA or a TSF treatment program should

Lee Ann Kaskutas is affiliated with the School of Public Health, University of California–Berkeley,Berkeley, California.

Address correspondence to: Lee Ann Kaskutas, DrPH, Alcohol Research Group, 6475 Christie Avenue,Suite 400, Emeryville, CA 94608-1010 (E-mail: [email protected]).

be made aware that there is a lack of experi-mental evidence about the effectiveness of suchprograms.1 This is despite optimal outcomes forTSF at 1 and 3 years for outpatients in the ProjectMATCH trial.2,3 At the other end of the spec-trum, 12-step scholar Rudy Moos has recom-mended that referral agencies should considerreferring people to AA first rather than to treat-ment first. This is based on his own observationalstudies, which have found that longer duration ofAA attendance is associated with less drinkingat 8 and 16 years,4 and that those who attend AA

Downloaded By: [Brown University] At: 19:09 9 April 2010

146 JOURNAL OF ADDICTIVE DISEASES

before attending treatment tend to attend AAlonger than those who attend treatment first.5

The goal of this article is to provide a focusedreview of the literature on AA effectiveness thatwill allow readers to judge the evidence for AAeffectiveness themselves.

Prior efforts to summarize the findings on AAeffectiveness have included literature reviews6,7

and meta analyses.8−10 The most recent meta-analysis10 concluded that attending AA led toworse outcomes than no treatment at all. An ear-lier meta-analysis focusing on moderating ef-fects found that the evidence for AA effective-ness was stronger in outpatient samples, and thatpoorer quality studies (based on volunteers, self-selection rather than random assignment, and nocorroboration of self-report) somewhat inflatedthe case for AA effectiveness.9 A review sum-marizing the state of the literature 7 years later7

argued that there was a consistent, rigorous bodyof evidence supporting AA effectiveness. Again,there seems to be something for everybody andthe literature seems to be widely subject to in-terpretation. This may stem from the criterionbeing used to judge effectiveness.

At the heart of the debate is the quality ofthe evidence. AA critics have argued that AAis a cult that relies on God as the mechanismof action,11 and that rigorous experimental stud-ies are necessary to convince them of AA’s ef-fectiveness. Their concern is well-founded. Aswill be evident from this review, experimentalstudies represent the weakest of the availableevidence. However, the review also will high-light other categories of evidence that are over-whelmingly convincing with respect to AA ef-fectiveness, including the consistency with es-tablished mechanisms of behavior change. Thisreview will organize the research on AA effec-tiveness according to 6 formal criterion for estab-lishing causation,12 which should help readersto integrate the sometimes conflicting conclu-sions discussed above. These criterion were firstintroduced to assist policymakers in evaluatingthe totality of the evidence of a causal effect forsmoking on lung cancer in the absence of ex-perimental data (as randomizing individuals tosmoker and non-smoker conditions was not anoption).13,14 The criterion offer a framework forjudging the “totality” of the evidence,12 implic-

itly acknowledging that the evidence may not bestrong for all criteria, and leaving the final de-cision to the individual evaluator. These are thecriterion:

1. The relationship between an exposure (here,exposure to AA) and the outcome (here, ab-stinence because AA does not recommendany drinking for alcoholics) must be strong.According to this criteria, weak relationshipsbetween AA and abstinence would not be asconvincing of causality as strong ones norwould they be as clinically relevant.

2. There should be a dose–response relation-ship, such that more involvement in AA re-lates to higher levels of abstinence. Buildingon the first criterion, the size of the dose–response effect also is important.

3. The consistency of the association matters.If some studies find a strong relationship be-tween the number of AA meetings attendedand the rate of abstinence but many do not,this would call into question whether thedose–response relationship should be trusted,as evidence goes.

4. The timing of the purported influence must becorrect. This means that the measurement ofAA exposure must be prior to the period ofabstinence that is being studied; otherwise,it could mean that abstinent people tend togo to AA rather than AA causing people tobe abstinent. Concurrent relationships do notcount here; thus, according to this criterion,AA attendance for the past month cannot beconsidered as causal evidence for being ab-stinent during the past month.

5. The specificity of the association must bedemonstrated. One must be able to rule outother explanations than AA exposure for hav-ing led to abstinence. This addresses the con-cern that those who attend AA are a part ofa select sample who would be sober withoutever going to AA. For example, if those whoattend AA are highly motivated to do some-thing about their drinking, it could be that thismotivation is the cause of their abstinence andit would be unfair to credit AA for their suc-cessful outcome. Evidence of specificity ide-ally requires experimental manipulation ofexposure to AA. For example, individuals in


Lee Ann Kaskutas 147

a study might be randomized to attend AAor to attend psychotherapy; they do not selecttheir treatment. Because of randomization,motivated people would end up being ran-domized both to psychotherapy and to AA,so it would not be the case that the “deck wasstacked” in favor of AA. If those randomizedto attend AA were more likely than those ran-domized to psychotherapy to be abstinent 2years later, this would demonstrate an effectspecific to AA that could not be due to a se-lection bias in which only motivated peopleattend AA. Randomization would also equal-ize other pre-existing conditions (known andunknown) that might confound AA’s effect.

6. Coherence with existing knowledge is neededto establish causation. In drug trials, this isaddressed by considering biological plausi-bility. For example, the drug neurontin stopsseizures because it reduces the electrical ac-tivity in the brain. Here, in studying AA ef-fectiveness, biological plausibility is of nohelp. The notion of theoretical plausibilityis suggested as a way of addressing coher-ence with existing knowledge; that is, arethe mechanisms of action that explain be-havior change present in AA? Several the-ories and different aspects of AA exposurewill be considered in addressing this finalcriterion.

METHODS

Articles involving Alcoholics Anonymous,Narcotics Anonymous, Cocaine Anonymous,12-step group, and 12-step facilitation in thetitle or as a keyword were considered for thisreview. Electronic searches involved all relevantdatabases (e.g., Etoh and MedLine) and wereaugmented by the author’s paper files on AA.Based on the title and in some cases the abstract,articles were considered for inclusion and werethen read and classified. Representative studieswere selected and are presented for each cri-terion. All located studies reporting a negativerole for AA in abstinence are reported, and nostudies with negative findings have intention-ally been excluded. In the interest of brevity and

clarity, many studies with positive findings forAA and several small 12-step facilitation stud-ies with mixed results among subgroups havebeen excluded. The objective was not to pro-vide another exhaustive literature review on AAeffectiveness, but rather to present representa-tive studies of AA effectiveness according to thecriterion for establishing causation.

Results are shown using figures, with the per-centage abstinent from alcohol along the y-axisand the AA exposure along the x-axis. Somestudies combined alcohol and drug abstinenceor considered 12-step group attendance, whichwould have included Narcotics Anonymous andother 12-step groups for drugs (in addition toAA). This is reflected in the figure titles andin the text. Results from studies that did not re-port rates of abstinence are not shown. The studysamples and citations are summarized at the bot-tom of each figure.

RESULTS

Criterion 1: Strength of Association

How large is the relationship between AA ex-posure and abstinence? As shown in Figure 1,which draws on a longitudinal study of male in-patients in Veterans Administration programs,rates of abstinence are approximately twice ashigh for those who attended a 12-step group suchas AA following treatment. One-year follow-upresults considered 12-step group attendance andabstinence from alcohol and drugs, whereas the18-month results reported AA attendance andalcohol abstinence. Results are remarkably sim-ilar at 1 year and 18 months for these differ-ent exposure and abstinence measures. Approx-imately 20% to 25% of those who did not at-tend AA or another 12-step group (or receiveany other form of aftercare after the inpatientstay) were abstinent from alcohol and drugs at1 year15 and from alcohol at 18 months (com-bined alcohol and drug abstinence were not re-ported at 18 months).16 The rates of abstinencewere about twice as high among those who hadattended AA or another 12-step group (but noother form of aftercare). In terms of effect sizes,this translates to a robust medium-size effect



(h = .5).17 Other studies are available that reporton other substance use measures (such as per-cent days abstinent [PDA]) and samples. Thisstudy is selected to demonstrate the strength ofthe association because it comes from a largesample (n = 3,018 at 1 year); it reported simpledichotomous measures of AA or 12-step groupexposure and abstinence; and it reported sep-arately for those who attended AA or 12-stepgroups during follow-up but were not exposedto subsequent formal treatment.

Criterion 2: Dose Response Relationship

Do higher levels of AA attendance or in-volvement relate to higher levels of abstinence?There is evidence of a dose response relation-ship for number of 12-step meetings (Figure2a), frequency of 12-step meetings (Figure 2b),and duration of AA meeting attendance (Fig-ure 2c). Again, studying male residential pa-tients in the Veterans Association system andconsidering AA meeting attendance for the 90days prior to the 1-year follow-up, the dose re-sponse curve looks almost linear (Figure 2a),with more 12-step meetings associated withhigher rates of alcohol and drug abstinence.4 Ina smaller outpatient sample, more than 70% of

those attending 12-step groups weekly for the 6months prior to the 2-year follow-up were alco-hol abstainers, whereas alcohol abstinence ratesamong those attending less than weekly werethe same as those who never attended during thatperiod18; this suggests a threshold dose-responseeffect for weekly attendance at 12-step groups(Figure 2b). In a longitudinal study of previ-ously untreated problem drinkers, 70% of thosewith 27 weeks or more of sustained AA meet-ing attendance any given year (whether at year1, years 2 to 3, or years 4 to 8) were abstinentfrom alcohol at the 16-year follow-up;4 thosewith shorter duration of attendance had lowerrates of abstinence, with the dose response mostevident for AA attendance years 1 and years 4-8(Figure 2c). This study is the reason for Moos’recommendation (see Introduction) to send peo-ple to AA first because those who went to AAfirst were more likely to be involved in AA forlonger duration.5

Criterion 3: Consistency of Association

The similarities in abstinence rates be-tween the weekly or near-weekly AA attendees(70%) in these two latter studies with different



populations and follow-up periods is relevant tocriterion 3. Another example is shown in Fig-ure 3, which presents the rates of abstinencefor those who attended AA but no other treat-ment (third bar, labeled “AA only”) in two dif-ferent samples (Veterans Association inpatientsand previously untreated problem drinkers in thegeneral population) with different follow-up pe-riods (1, 3, and 8 years). The 1-year study con-

sidered alcohol and drug abstinence as a func-tion of 12-step group attendance, whereas the 3-and 8-year data focused specifically on AA at-tendance and alcohol abstinence. Approximately50% of those who had attended AA or 12-stepmeetings only were abstinent at 1 year15 andat 3 and 8 years;19 approximately one-fifth ofthose who did not attend AA or 12-step meet-ings or treatment were abstinent at the parallel



follow-up interviews. Another study of the gen-eral population20 found that individuals withlifetime alcohol dependence who went to 12-step meetings but did not have formal treatmentwere more likely to be abstinent than those whodid nothing (not shown).

Criterion 4: Temporally CorrectAssociation

Most of the above studies considered concur-rent AA attendance, and thus do not meet the 4thcriterion for evidence of causality. An exception



is Moos’ work, which studied 16-year alcoholabstinence in a previously untreated problemdrinking sample as a function of AA during years2 to 3 and years 4 to 8 (Figure 2c).4 ProjectMATCH also has evidence of a temporally cor-rect association, reporting that frequency of AAmeeting attendance as well as overall AA in-volvement in months 1 to 6 significantly pre-dicted the percentage of days of alcohol ab-stinence during months 7 to 12. This was thecase for Project MATCH subjects who attendedinpatient treatment prior to entering the study(“aftercare” arm) as well as those who attendedonly the Project MATCH treatment (“outpatient”arm); the beta coefficients for AA involvementpredicting abstinence were 0.34 in the aftercarearm and 0.29 in the outpatient arm (results notshown).21,22

Criterion 5: Specificity

Experimental evidence is generally consid-ered evidence of specificity. Three rigorousstudies are particularly relevant here. The first,a clinical trial of compulsory treatment thatrandomized individuals to attend AA, attendhospital inpatient treatment, or choose their own

treatment or service provider23 found signifi-cantly lower rates of alcohol abstinence for theAA and the choice conditions, with over twiceas many individuals abstinent at 2 years in thehospital inpatient condition (Figure 4a).

The second study, Project MATCH (discussedin criterion 4), randomized subjects to TSF, cog-nitive behavioral therapy, or motivational en-hancement. In the aftercare arm, there were nosignificant differences between the three treat-ments, with more than two-fifths abstinent atthe 1-year follow-up (results not shown). In theProject MATCH outpatient arm, rates of alcoholabstinence were significantly higher for thosetreated in TSF at 1 year2 [Table 4] and 3 years3

(Figure 4b). As noted above in Criterion 4,AA participation among Project MATCH clientspredicted subsequent abstinence, regardless ofstudy arm or condition.

The third trial randomized Veterans Asso-ciation outpatients to an intensive 12-step re-ferral condition or to standard AA referral,24

finding significantly higher rates of total absti-nence (from alcohol and drugs) at both the 6-and 12-month follow-ups for the intensive re-ferral condition (Figure 4c). Higher AA or Nar-cotics Anonymous involvement in the intensive



referral condition fully mediated the conditioneffect on abstinence, but AA participation pre-dicted abstinence regardless of condition.

Another relevant trial randomized individuals(mainly court-referred) to attend a weekly AAmeeting run by the investigative team but not partof mainstream AA in the community, to attend

weekly one-on-one therapy sessions led by layindividuals, or to a control condition in whichsubjects may have attended AA in the commu-nity, other available treatment, or no treatment.25

Significantly more binge drinking at the 3-monthfollow-up was found for individuals randomizedto the special AA meeting (2.37 binges in the



past 3 months) than to the other conditions (0.26in lay therapy and 0.56 for the controls), butthere was no reported difference in abstinence.However, at the 1-year follow-up, all drinkingmeasures including rates of abstinence were sim-ilar across the conditions (result not shown).A 5th experiment randomized convicted drunkdrivers to AA, outpatient treatment, or a no treat-ment condition; the study did not report drinkingoutcomes but found no differences in recidivismfor drunk driving26 (result not shown).

Criterion 6: Coherence with ExistingKnowledge

To evaluate the literature on AA effective-ness according to this criterion (which usually isstudied by considering biological plausibility),theoretical plausibility will be discussed; that is,does AA work in a way that is consistent withmajor theoretical perspectives on health behav-ior and behavior change? For example, a recentinterpretation of contemporary psychodynamictheory has characterized alcoholism as an inter-action between one’s abilities to express feel-ings and self-regulate one’s behavior.27 The the-ory argues that despite low self-esteem, manyalcoholics have a narcissistic personality28 anda sense of omnipotence. They drink to self-medicate as a way of addressing unmet needsand uncomfortable psychological states. AA so-lutions consistent with this characterization ofthe problem are evident at meetings, in the AAsteps, and through people in the AA fellowship.Meetings provide an opportunity to share one’sown struggles, to learn how to talk about one’sfeelings, to increase one’s motivation to abstain,and to get outside of one’s self and change one’smood by hearing others talk about their problemsand how AA helped them. The steps help withself-governance, narcissism, and omnipotence:accepting powerlessness over alcohol (step 1);recognizing that one cannot do it alone but thata higher power, which can be operationalizedas the AA group, is there to help (steps 2-3);realizing how one’s behavior affected and af-fects others (steps 4–9); treating other peoplebetter (step 10); finding meaning in life (step11); and relinquishing one’s negative self-focusby helping others (step 12). Through the peo-

ple in AA, one learns how to live a sober lifeand how to regulate one’s behavior one day at atime.

Bandura’s social learning theory29 adds to thepsychodynamic perspective, saying that a largepart of the problem arises from social influencesand from self-efficacy: if everyone around youdrinks and if you don’t think it is within yourability to not drink, you will be unable to abstain.The antidote includes changing environmentalcues (such as staying away from bars), role mod-eling (seeing others succeed at not drinking),and self-efficacy (believing you can abstain). AAmeetings and spending time with people in AArepresent changes in environmental cues (i.e.,you’re not at a bar seeing alcohol and watchingpeople drink alcohol when you’re at a meetingor out with AA friends). At an AA meeting, youare exposed to successful role models, insteadof current drinkers, who suggest a new approachto abstinence: not drinking 1 day at a time (in-stead of saying you are “quitting forever”). See-ing yourself able to abstain for one day beginsto build self-efficacy, which accumulates withthe passage of every sober day. Spending timeat AA meetings and with people in AA alsoleads to relapse prevention mechanisms put for-ward by standard behavioral modification tech-niques. These include learning how to say no to adrink when offered, having a plan of action whenconfronted with likely drinking conditions, andchoosing alternative behaviors to take the placeof drinking.

Several studies offer empirical support forthese mechanisms. The positive relationship be-tween AA involvement and abstinence has beenshown to be partially mediated (explained) by (1)psychological and spiritual mechanisms includ-ing finding meaning in life,30 greater motivationfor abstinence,31 and changes in religious be-liefs and spiritual experiences;32 (2) social influ-ences such as fewer pro-drinking influences,33

more friends in general,34 having AA friendssupportive of abstinence,35 and enhanced friend-ship networks;36 and (3) social learning and be-havioral mechanisms including improved self-efficacy,31,37 and effective coping and relapseprevention skills34,36 to abstain. These mecha-nisms (and theories) are inter-related. For exam-ple, AA friends represent a particularly effective



source of social support because they provideexpertise in preventing relapse.

DISCUSSION

Limitations

This is not a thorough review of the litera-ture on AA effectiveness. For example, we didnot keep track of the number of relevant stud-ies located or the relative numbers of studieswith positive versus negative findings for AA orTSF effectiveness. However, we did take careto present any study where the effect of AA wasnegative. The goal was not to provide an exhaus-tive review of the evidence, but rather to presentrepresentative studies that address AA effective-ness according to six accepted criterion for es-tablishing scientific causation. This frameworkmay be especially appropriate for consideringAA effectiveness because it acknowledges thevalue and limitations of experimental evidencein the context of other criterion for determiningtreatment effectiveness.

Another limitation is the choice of theoreticalframeworks for consideration. Biological theo-ries were not considered here because their so-lutions are not behavioral but rather pharma-cological: genetic theory (one is predisposedto develop alcoholism) and neurobiologicaltheories (the brain becomes addicted to alcohol).For ideas about other behavioral theories thatmight be at work in AA, readers are referred toMoos’ recent article on the active ingredients ofsubstance use-focused self-help groups, whichconsiders social control theory, behavioral eco-nomics, and stress and coping theory in additionto social learning theory.38 The breadth of the-oretical frameworks through which AA mecha-nisms can be understood is encouraging.

CONCLUSIONS

As stated at the outset, the experimental ev-idence for AA effectiveness (addressing speci-ficity) is the weakest among the six criteria con-sidered crucial for establishing causation. Onlytwo studies provided strong proof of a specific

AA or TSF effect: the outpatient arm of ProjectMATCH (with effects at 1 and 3 years)2,3 andthe intensive referral condition in Timko’s trial(with effects for abstinence at 6 months and 1year).24 The effect sizes were similar, with theTSF/intensive referral conditions having a 5%to 10% advantage in abstinence rates. It is note-worthy that neither of these studies attempted torandomize patients to AA per se; instead, theyfocused on interventions intended to facilitateAA involvement.

One reason that several of the other trialsmay not have found positive effects for AA/TSFis because many individuals randomized to thenon-AA/non-TSF conditions also attended AA;thus, the AA or TSF condition ended up be-ing compared to a condition consisting of analternative treatment plus AA. This was the casein Walsh’s hospital inpatient treatment versusAA study23 and in the aftercare arm of ProjectMATCH,22 and arose because the patients in thenon-AA/non-TSF conditions also had attended12-step-based inpatient treatment, which in turnengendered strong participation in AA. Thus,AA attendance levels were high in the inpa-tient hospital condition in the former study andin the cognitive behavioral therapy and METconditions among the Project MATCH aftercaresubjects. In fact, cognitive behavioral therapyand MET aftercare patients attended more meet-ings than the TSF outpatients, and the aftercarepatients overall attended twice the number ofmeetings at every follow-up compared to theoutpatients.22

There are other concerns with the Brandsmatrial,25 which call its experimental results intoquestion. The control condition allowed for par-ticipation in actual AA meetings, whereas thosein the AA condition attended a weekly AA-likemeeting administered by the study that was notan actual AA meeting. The description of theAA condition states that the steps were used fordiscussion content, the group focused on new-comers, and they told patients about sponsors,25

but it is not clear whether the meetings were ledby AA members, whether crosstalk was allowed,whether the meeting leader shared their story aspart of the meeting, or whether the meeting for-mat was what one would encounter at an actualAA meeting. The meetings may not have been



open to other AA members in the community andmay not been listed in the AA meeting directory,which would mean that a potentially importanttherapeutic ingredient of AA—the experienceof longer-term members—would not have beenpresent in the AA condition. This is of specialconcern because the control condition did allowfor attendance at such meetings.

Given these challenges in conducting rigor-ous randomized trials of AA effectiveness, re-searchers have turned to statistical methods toaddress the selection bias associated with AAattendance in observational studies. These ef-forts are intended to address criteria 5, specificityof the AA effect. The goal with these methodsis to statistically adjust for study participants’likelihood or propensity to attend AA prior toevaluating AA’s impact on subsequent drink-ing. One approach, used in two studies of AAeffectiveness, is an econometric method usingso-called “instrumental variables” to parse-outAA attendance. The instrumental variables inone study were the availability of AA meet-ings in one’s community and being able todrive to meetings;39 after adjusting for these po-tential confounders, AA’s effect on abstinencewas reduced from OR = 3.70 (P < .05) toOR = 1.69 (not significant). Using different in-strumental variables (perceived seriousness ofdrinking, and having a coping style tending to-wards information-seeking solutions), anotherstudy40 found that AA’s impact on heavy drink-ing was significant and doubled in magnitudeafter correcting for the instrumental variables.A third study30 adjusted for baseline motivationand psychopathology as potential confoundersand found that those with more AA involve-ment at 1 year had fewer alcohol problems at the2-year follow-up interview. Another statisticalstudy of selection bias used Propensity Scoresto adjust for study participants’ propensity toattend AA42 and found that the odds of absti-nence associated with AA attendance were re-duced but remained significant after adjustingfor individuals’ propensity to attend AA. Themethod allowed investigators to study whetherthe selection bias operationalized by the Propen-sity Scores varied based on whether an individ-ual had a low versus a high propensity to attendAA. AA’s effect was minimal (e.g., OR = 1.3)

among those with a high propensity to attendAA; however, the odds of abstinence associatedwith AA attendance were significant and of con-siderable magnitude, ranging from 2.7 to 6.9,among those with a lower propensity to attendAA.

What, then, is the scorecard for AA’s effec-tiveness in terms of specificity? Among the rig-orous experimental studies, there were two posi-tive findings for AA effectiveness, one null find-ing and one negative finding. Among those thatstatistically addressed selection bias, there weretwo contradictory findings and two studies thatreported significant effects for AA after adjust-ing for potential confounders such as motivationto change. Readers must judge for themselveswhether their interpretation of these results, onbalance, supports a recommendation that thereis no experimental evidence of AA effective-ness (as put forward by the Cochrane review).As for the scorecard for the other criteria, theevidence for AA effectiveness is strong: ratesof abstinence are approximately twice as highamong those who attend AA (criteria 1, mag-nitude); higher levels of attendance are relatedto higher rates of abstinence (criteria 2, dose-response); these relationships are found for dif-ferent samples and follow-up periods (criteria 3,consistency); prior AA attendance is predictiveof subsequent abstinence (criteria 4, temporal);and mechanisms of action predicted by theoriesof behavior change are evident at AA meetingsand through the AA steps and fellowship (crite-ria 6, plausibility).

REFERENCES

1. Ferri M, Amato L, Davoli M. Alcoholics Anony-mous and other 12-step programmes for alcohol depen-dence. Cochrane Database of Systematic Reviews. 2006;3: Art. No. CD005032.

2. Project MATCH Research Group. Matching al-coholism treatment to client heterogeneity: ProjectMATCH posttreatment drinking outcomes. J Stud Alcohol.1997;58:7–29.

3. Project MATCH Research Group. Matching alco-holism treatments to client heterogeneity: Project MATCHthree-year drinking outcomes. Alcohol Clin Exp Res.1998;22:1300–11.



4. Moos RH, Moos BS. Participation in treatmentand Alcoholics Anonymous: a 16-year follow-up of ini-tially untreated individuals. J Clin Psychol. 2006;62:735–50.

5. Moos RH, Moos BS. Help-seeking careers: con-nections between participation in professional treat-ment and Alcoholics Anonymous. J Subst Abuse Treat.2004;26:167–73.

6. Emrick CD. Alcoholics Anonymous: affiliation pro-cesses and effectiveness as treatment. Alcohol Clin ExpRes 1987;11:416–23.

7. Kelly JF. Self-help for substance-use disorders: his-tory, effectiveness, knowledge gaps, and research opportu-nities. Clin Psychol Rev. 2003;23:639–63.

8. Emrick CD, Tonigan JS, Montgomery HA, Little L.Alcoholics Anonymous: what is currently known? In: Mc-Crady BS, Miller WR, eds. Research on Alcoholics Anony-mous: opportunities and alternatives. New Brunswick, NJ:Rutgers Center of Alcohol Studies, 1993:41–78.

9. Tonigan JS, Toscova R, Miller WR. Meta-analysisof the literature on Alcoholics Anonymous: sample andstudy characteristics moderate findings. J Stud Alcohol.1996;57:65–72.

10. Kownacki RJ, Shadish WR. Does AlcoholicsAnonymous work? The results from a meta-analysis ofcontrolled experiments. Subst Use Misuse. 1999;34:1897–916.

11. Bufe C. Alcoholics Anonymous: cult or cure? SanFrancisco, CA: Sharp Press, 1991.

12. Mausner JS, Kramer S. Epidemiology: an introduc-tory text. 2nd ed. Philadelphia: W. B. Saunders Company,1985.

13. Kleinbaum DG, Sullivan KM, Barker ND. ActivEpicompanion textbook: a supplement for use with the Ac-tivEpi CD-ROM. New York: Springer Science+BusinessMedia, Inc., 2005.

14. Hill AB. Principles of medical statistics. 9th ed. NewYork: Oxford University Press, 1971.

15. Ouimette PC, Moos RH, Finney JW. Influence ofoutpatient treatment and 12-step group involvement on one-year substance abuse treatment outcomes. J Stud Alcohol.1998;59:513–22.

16. Thurstin AH, Alfano AM, Nerviano VJ. The efficacyof AA attendance for aftercare of inpatient alcoholics: somefollow-up data. Int J Addict. 1987;22:1083–90.

17. Cohen J. Statistical power analysis for the behav-ioral sciences. 2nd ed. Hillsdale, NJ: Lawrence ErlbaumAssociates. 1988.

18. Fiorentine R. After drug treatment: are 12-step pro-grams effective in maintaining abstinence? Am J Drug Al-cohol Abuse. 1999; 25:93–116.

19. Timko C, Moos RH, Finney JW, Lesar MD. Long-term outcomes of alcohol use disorders: comparing un-treated individuals with those in Alcoholics Anonymousand formal treatment. J Stud Alcohol. 2000;61:529–40.

20. Dawson DA, Grant BF, Stinson FS, Chou PS. Es-timating the effect of help-seeking on achieving recoveryfrom alcohol dependence. Addiction. 2006; 101:824–34.

21. Connors GJ, Tonigan JS, Miller WR. A longitudi-nal model of intake symptomatology, AA participation, andoutcome: retrospective study of the Project MATCH outpa-tient and aftercare samples. J Stud Alcohol. 2001;62:817–25.

22. Tonigan JS, Connors GJ, Miller WR. Participationand involvement in Alcoholics Anonymous. In: Babor TF,Del Boca FK, eds. Treatment matching in alcoholism. Cam-bridge UK: Cambridge University Press. 2003:184–204.

23. Walsh DC, Hingson RW, Merrigan DM, LevensonSM, Cupples LA, Heeren T, Coffman GA, Becker CA,Barker TA, Hamilton SK, McGuire TG, Kelly CA. Arandomized trial of treatment options for alcohol-abusingworkers. N Engl J Med. 1991;325:775–82.

24. Timko C, Debenedetti A. A randomized controlledtrial of intensive referral to 12-step self-help groups: one-year outcomes. Drug Alcohol Depend. 2007;90:270–9.

25. Brandsma JM, Maultsby MC Jr, Welsh RJ. Outpa-tient treatment of alcoholism: a review and comparativestudy. Baltimore, MD: University Park Press. 1980.

26. Ditman KS, Crawford GG, Forgy EW, MoskowitzH, MacAndrew C. A controlled experiment on the useof court probation for drunk arrests. Am J Psychiatry.1967;124:160–3.

27. Khantzian EJ, Mack JE. Alcoholics Anonymous andcontemporary psychodynamic theory. In: Galanter M, ed.Recent developments in alcoholism. New York: PlenumPress, 1989:67–89.

28. Tiebout HM. Therapeutic mechanisms of AlcoholicsAnonymous. Am J Psychiatry. 1944;100:468–73.

29. Bandura A. Social learning theory. Morristown, NJ:General Learning Press, 1971.

30. White W, Laudet A. Life meaning as potential me-diator of 12-step participation benefits on stable recoveryfrom polysubstance use. In: The college on problems ofdrug dependence. Conference 2006, June 17–22: Scotts-dale, AZ.

31. Kelly JF, Myers MG, Brown SA. Do adolescentsaffiliate with 12-step groups? A mulitvariate process modelof effects. J Stud Alcohol. 2002;63:293–304.

32. Zemore SE. A role for spiritual change in thebenefits of 12-step involvement. Alcohol Clin Exp Res.2007;31:76S–9S.

33. Kaskutas LA, Bond J, Humphreys K. Social net-works as mediators of the effect of Alcoholics Anonymous.Addiction. 2002;97:891–900.

34. Timko C, Finney JW, Moos RH. The 8-year courseof alcohol abuse: gender differences in social context andcoping. Alcohol Clin Exp Res. 2005;29:612–21.

35. Bond J, Kaskutas LA, Weisner C. The persistentinfluence of social networks and Alcoholics Anonymouson abstinence. J Stud Alcohol. 2003;64:579–88.



36. Humphreys K, Mankowski ES, Moos RH, FinneyJW. Do enhanced friendship networks and active copingmediate the effect of self-help groups on substance abuse?Ann Behav Med. 1999;21:54–60.

37. Morgenstern J, Labouvie E, McCrady BS, KahlerCW, Frey RM. Affiliation with Alcoholics Anonymousfollowing treatment: a study of its therapeutic effectsand mechanisms of action. J Consult Clin Psychol.1997;65:768–77.

38. Moos RH. Active ingredients of substance use-focused self-help groups. Addiction. 2008;103:387–96.

39. Fortney J, Booth B, Zhang M, Humphrey J, Wise-man E. Controlling for selection bias in the evaluation ofAlcoholics Anonymous as aftercare treatment. J Stud Al-cohol. 1998;59:690–7.

40. Humphreys K, Phibbs CS, Moos RH. Addressingself-selection effects in evaluations of mutual help groupsand professional mental health services: an introduction totwo-stage sample selection models. Eval Program Plann.1996;19:301–8.

41. McKellar J, Stewart E, Humphreys K. AlcoholicsAnonymous involvement and positive alcohol-related out-comes: cause, consequence, or just a correlate? A prospec-tive 2-year study of 2,319 alcohol-dependent men. J ConsultClin Psychol. 2003;71:302–8.

42. Ye Y, Kaskutas LA, Bond J. Using propensity scoresto adjust for selection bias when assessing the effective-ness of Alcoholics Anonymous in observational studies.2008 (under review). Alcohol Research Group, Emeryville,CA.


journal of addictive diseases alcoholics anonymous effectiveness

Documents