So Alcoholics Anonymous Is “Proven” to Work After All? Not So Fast 

    Alcoholics Anonymous and its 12 Steps, with their prescription of abstinence, have dominated America’s response to addiction for a half-century or more. Yet despite the urges of advocates and the lucrative, 12-step-based rehab industry to enshrine this status, reliable studies demonstrating AA’s effectiveness were not forthcoming.

    In 2006, a review of controlled AA research (eight studies, with 3,417 subjects) by the prestigious Cochrane group found that “No experimental studies unequivocally demonstrated the effectiveness of AA or [12-step facilitation] approaches for reducing alcohol dependence or problems” (my emphasis).

    Now, Cochrane has revisited this question with a new review (27 studies, 10,565 subjects). It announced its findings on March 11: 

    The authors found high certainty evidence that clinically delivered and manualized [12-step facilitation, or TSF] programs designed to increase AA participation can lead to higher rates of continuous abstinence over months and years, when compared to other active treatment approaches such as cognitive behavioural therapy. The evidence suggests that 42% of participants participating in AA would remain completely abstinent one year later, compared to 35% of participants receiving other treatments including CBT. (My emphasis.)

    This was news the world had been waiting to hear. “New study shows how effective AA really is,” splashed USA Today. “A new, big review of the evidence found that AA works—for some,” hedged Vox. The New York Times article on the meta-study (“An updated review shows it performs better than some other common treatments”) has recently been among its most-read pieces.

    So should those of us who have opposed the large-scale imposition of AA now throw our hands up and apologize? 


    Cochrane Switches Measures 

    The authors of the Cochrane review are John Kelly of Harvard Medical School, Keith Humphreys of Stanford University, and Marica Ferri of the European Centre for Drugs and Drug Addiction. Kelly, the most outspoken of the group, publicly advocates for the brain disease model. As a group, the authors’ work has had the effect of (a) establishing a disease basis for addiction, (b) certifying the success of AA and its 12 Steps, and (c) opposing the trend towards harm reduction, or non-abstinence outcomes, as goals of treatment or recovery.

    AA, as you may be aware, is not a harm reduction operation. So Kelly et al.’s Cochrane-reported results, unlike the prior review, laser-focused on abstinence, comparing abstinence outcomes with those of other clinical interventions over varying periods of time.

    Their touted positive finding is that “AA/TSF improves rates of continuous abstinence” at 12, 24 and 36 months,” with a seven-percentage-point advantage at one year.

    However, this announced positive result compared with the first review was due solely to switching measurements—from actual recovery rates to continuous abstinence.


    Cochrane’s Downplayed Negative Findings

    But okay! People who participate and continue in AA/TSF programs are more likely to continuously abstain in those early years than those in other programs. There are far-reaching problems with the self-selecting population studied, as we will see. But let’s turn first to the results presented by Cochrane.

    Taken overall, the multitude of measures and periods over which subjects were observed offer a lot to pick and choose from in claiming that AA/TSF is potentially beneficial. All the same, such benefits were not apparent across the board—and the reverse was often the case, with AA/TSF performing worse on critical measures.

    People in other interventions racked up, on average, just as many total abstinent days.

    AA/TSF, as Cochrane found, did not show a difference in “percentage of days abstinent” with non-12-step approaches at 12 months. This means that people in AA/TSF were more likely to have experienced unbroken abstinence over this period; but people in other interventions, which display less abhorrence of some level of drinking, racked up, on average, just as many total abstinent days.

    The review did show AA/TSF did better at 24 months and 36 months for this measure, but with only two and one studies respectively—in both cases this was rated “very low certainty evidence.”

    The review also reported that “For longest period of abstinence, AA/TSF may perform as well as comparison interventions at six months” (low-certainty evidence) but not for longer periods. This is puzzling indeed. So, unlike continuous abstinence, “longest period of abstinence” advantages did not appear in AA/TSF instantly, and AA/TSF was (maybe) only just as good as other interventions in that regard, and 12-step programs apparently did not produce equally “longest periods of abstinence” afterwards.

    This is an extremely problematic finding. Why did reports on this study not note this strange result?

    This same apparent disadvantage for AA/TSF was true in regards to drinking intensity, to wit, “For drinking intensity, AA/TSF may perform as well as other clinical interventions at 12 months, as measured by drinks per drinking day and percentage days heavy drinking” (my emphasis) but for no longer period. In other words, AA/TSF programs struggled to keep participants from not drinking more intensively than non-12-step groups.

    This is the antithesis of a benefit. AA groups, by convincing people they can’t control their drinking, make it likelier that when such subjects do drink, they do so explosively. Once again, commentaries have failed to note this drawback to the Cochrane research claim.

    Past research by Miller et al. and Heather et al. found treated subjects were more likely to relapse, and to relapse worse, when they believed in the disease theory of alcoholism (Miller et al.) and the “cultural delusion” of one-drink/one-drunk (Heather et al.). Keep in mind that per AA, no matter how long you have abstained, whether you take a single drink or have an all-out bender, you’re back at day one. Non-12-step CBT programs, on the other hand, teach relapse prevention techniques, including a person’s ability to get off the “relapse train” at any station.

    This result undercuts the entire AA good news story.

    Kelly et al. did consider overall functioning, which is a harm reduction way to think. The American Psychiatric Association, after all, diagnoses substance use disorders not according to how many days people are abstinent or not, but by the real impacts on their lives.

    Here again, Cochrane’s pro-AA/TSF stance fails to hold: “For alcohol-related consequences, AA/TSF probably performs as well as other clinical interventions at 12 months(my emphasis)—but for no longer period.

    Isn’t the sine qua non of treatment ultimately people’s functionality extending into the future beyond treatment or group participation? This result undercuts the entire AA good news story.

    For “alcoholism severity, one study found evidence in favor of AA/TSF at 12 months,” Cochrane noted. This is strongly reminiscent of the finding in NESARC (the National Institute on Alcohol Abuse and Alcoholism’s massive study conducted between 2001 and 2012) that those with chronic, severe alcohol addiction don’t respond well to available treatment. This was the case in the Cochrane review, aside from a single study, for the 12-step-type advocates claim to be most helpful for this population.

    In terms of not reducing but exacerbating relapse, not improving overall life indicators and not resolving severe alcohol addiction, the latest Cochrane results do not represent an actual change in measured outcomes over the 2006 review. The ballyhoo over the new review is strictly public relations.


    Picking Subjects Likely to Do Well

    One caveat that has been widely ignored is that Kelly et al.’s primary and best results occur with “manualized TSF programs.” That means extremely well-organized, competent treatment researchers and clinicians were involved in those programs. Detailed 12-step manuals for individual one-on-one therapy are not the standard for Delray Beach (or even Hazelden/Betty Ford) rehabs—and certainly not for any AA group.

    The template for the Cochrane-reported, manualized results is the legendary Project MATCH, the most expensive clinical trial of alcohol use disorder treatment ever conducted, which occupied eight years, including detailed statistical analyses, beginning in 1989.

    MATCH’s essential result was a failure: The null hypothesis, that there is no benefit to matching types of patients with one of three treatments (TSF, social skills training, motivational enhancement), held. That is, matching subject traits and treatments wasn’t advantageous. This entire approach has now been abandoned in the alcohol treatment field.

    The positive Cochrane results are from controlled experimental comparisons of treatments of similarly cooperative subjects.

    As with the latest Cochrane review, there were mild abstinence advantages for TSF subjects. But the dominant overall result was general improvement for subjects (measured, incidentally, in harm reduction terms of reduced days of alcoholic drinking). However, it proved impossible to generalize these results. In the first place, there wasn’t a no-treatment control group—apparently the researchers couldn’t imagine natural recovery. Kelly et al. in their review find AA/TSF cost-effective relative to other treatments—but how does that compare with zero costs?

    Most importantly, the problematic-drinking subjects in MATCH were highly selected (socially stable, no involvement in the criminal justice system) volunteers—a positive-prognosis group no matter what treatment they received. Moreover, the investment in and care level of the treatment—reflected in the exorbitant costs of the research and the creation of detailed manuals on which therapists were carefully trained and their sessions recorded and monitored—corresponded to nothing encountered in the real world.

    The positive Cochrane results are from controlled experimental comparisons of treatments of similarly cooperative subjects (Cochrane doesn’t discuss subject composition over the multitude of studies).  They thus are likely a function of the obedient, committed, socially well situated “good” subjects who both participate in such studies and enroll in high-end rehabs and remain assiduous in their AA attendance.

    These are exactly the people most likely to moderate their alcohol-dependent drinking without treatment! Put in reverse form, in the Wave 2 follow-up of the government’s NESARC research, “the Chronic Severe showed the greatest stability in the manifestations of alcohol dependence, despite having the highest rate of treatment seeking.”

    In terms of lifetime recovery from alcohol dependence (AD) measured in the 43,000+ subjects originally interviewed in NESARC, Americans who received treatment (heavily weighted towards the 12 steps and abstinence) were more likely to abstain. However, since they were also far less likely to achieve harm reduction outcomes, they were slightly more likely to maintain their AD.

    The studies reviewed did not include the large population coerced into formal treatment or AA groups—primarily through criminal or family courts.

    Subjects who attend and remain in AA, or who voluntarily go to rehab, are a quite small portion of the American population. NESARC found that only about a quarter of alcohol-dependent people are treated in any way, shape or form, but that even fewer, about half of these, enter formal treatment or AA. How well do these subjects compare with all of those facing addiction problems in the US?

    In addition, the studies reviewed did not include the large population coerced into formal treatment or AA groups—primarily through criminal or family courts.

    Cochrane’s singular focus on 12-step programs for alcohol only, and no other drugs, amplifies this gap. But poorer, less well-educated populations have worse drug outcomes. To take one example, the poorest US state, West Virginia, is far in the lead in drug-related deaths (as it has been since 2014). “In 2017, the states with the highest rates of death due to drug overdose were West Virginia (58 per 100,000) (followed by) Ohio (46 per 100,000).” 

    West Virginia’s commissioner of public health investigated every single death for 2017, all 887 of them, and concluded:

    If you’re a male between the ages of 35 to 54, with less than a high school education, you’re single and you’ve worked in a blue-collar industry, you pretty much are at a very, very high risk of overdosing.

    Were these deaths due to insufficient AA groups and 12-step treatment programs in West Virginia? 


    A Dead-End Review

    To summarize, we must greet these congratulatory headlines about AA and the 12 Steps critically—based on the long-held predispositions of the researchers, the choice of continuous abstinence from the abundance of measures they examined, and, finally, the restricted populations they address.

    In ideal settings and with relatively privileged populations, AA/TSF increases the chances of unbroken abstinence—but not of reductions in alcohol-related harms.

    Even taken at face-value, the review is anything but the game-changing endorsement that headlines suggest.

    The same deficiencies in AA/TSF appear in regards to severe relapses. Shockingly, a review said to indicate benefits for the 12 Steps noted only keeping longest periods of sobriety and lower intensity of relapses on a par with non-12–step treatments, but apparently could not achieve parity with other treatments in this regard for longer periods.

    This is something to hoot and holler about, as the popular press has been doing?

    We’ll shortly come to the value of abstinence in itself. But even taken at face-value, the review is anything but the game-changing endorsement that headlines suggest.

    Kelly himself acknowledged some common sense about the findings of AA’s “success” to Vox:

    I don’t think it’s something unique to AA, as if it’s got some sort of magic. It’s rather that the magic of AA is that it’s everywhere and mobilizes these therapeutic mechanisms in a very strong, socially supportive network of recovery support.

    Community and purpose are critical to overcoming addiction, as I have often written. And these can be found in many ways that don’t require the admission of powerlessness over a substance, penance for all of your past sins, or the belief that you have a disease you are ready to relapse into at any moment.


    A Fundamental Change of Direction

    We need a whole new way of thinking about our relationship to drugs and alcohol. This reconception needs to respond to problems created by substance use. But America’s drug problems are caused primarily by the lethal combination of our abstinence obsession, drug panics and inhumane capitalism.

    There is a larger background question here. How are we doing? Is this a point in history where we want to double down (or double back) on our long-tried existing approaches to addiction?

    The addiction author and journalist Maia Szalavitz, a successful graduate of a Hazelden-style 12-step rehab, wrote an important piece in 2017 cataloging the cultural weight behind the 12 Steps—“After 75 years of Alcoholics Anonymous, It’s Time to Admit We Have a Problem: Challenging the 12-step Hegemony”:

    For much of the past 50 years or so, voicing any serious skepticism toward Alcoholics Anonymous or any other 12-step program was sacrilege—the equivalent, in polite company, of questioning the virtue of American mothers or the patriotism of our troops. If your problem was drink, AA was the answer; if drugs, Narcotics Anonymous. And if those programs didn’t work, it was your fault: You weren’t “working the steps.” The only alternative, as the 12-step slogan has it, was “jails, institutions, or death.”

    I can attest to that. I have for decades been one of few addiction specialists willing to publicly question AA, and have been blacklisted as a result.

    The most prominent object of my attack has been AA’s abstinence mantra. This fixation, of course, is antithetical to harm reduction. For alcohol, at a whole-population level, it is antithetical to public health.

    The presence of 12-step programs in every US community is not succeeding by measures of life satisfaction, survival rates or mortality due to drugs and alcohol.

    Yet, as Szalavitz noted, by 2000, “90 percent of American addiction treatment programs employed the 12-step approach.”

    In this context, does anyone really feel that things are going well addiction-wise in the US today? Consider:


    * The Commonwealth Fund Health Care Report compared the US with 10 other economically advanced countries. Its chief findings: the US has the highest suicide rate and the lowest life expectancy, along with the highest (by far) per capita health care costs. (Feeling good about America’s health care system, including addiction treatment, were you?)

    * The 2019 World Happiness Report chapter on addiction notes that suicides most often involve drugs and alcohol. More startlingly, the report cites a Global Burden of Disease study finding that, out of 196 countries worldwide, the US population ranked second worst globally in life years lost to death and disability traceable to drugs (cocaine, opioids and amphetamines). This stunning result includes impoverished nations, nations at war, and those that produce opiates and coca, as well as countries comparable in wealth and lifestyle to the US.

    * Finally, the book Deaths of Despair and the Future of Capitalism, by economists Anne Case and Angus Deaton (Princeton University Press, 2020), reviews how and why, for three years in a row, for the first time in modern history, the life expectancy of white working-class Americans declined (African Americans already had a significantly lower life expectancy). This decline was the consequence mainly of drug deaths (including alcohol) and suicide, which have been increasing precipitously since 1999. 


    So it might seem poor timing just now to celebrate the great success of our most popular form of addiction therapy. The presence of 12-step programs in every US community is not succeeding by measures of life satisfaction, survival rates or mortality due to drugs and alcohol. Far from it.

    TV chef Andrew Zimmern’s series on MSNBC, What’s Eating America, recently devoted a segment to America’s addictive proclivities. Zimmern is another Hazelden graduate, and it was instructive watching him try to pull off the standard pro-12-step sleight-of-hand. While Zimmern lauds AA and Hazelden for his 27 years of recovery, along with other success stories, he simultaneously sounds the alarm bells for America’s submergence in a sea of addiction. But he never addresses why the ubiquity of the former did not prevent the latter.

    Like everything we do in life, we learn how to use drugs.

    A new vision will need to take into account the following factors:


    *Problems with drugs and alcohol follow, rather than create, overall life outcomes.

    When large groups of people are suffering, it manifests in deadly interactions with drugs. Drugs in themselves are not driving people to their deaths.

    * The disease approach—which underlies AA/TSF—is itself a source of our current difficulties.

    It depreciates our lived experience in America by targeting resources on drugs and alcohol rather than on the conditions of people’s lives. It disempowers people. And it provides a convenient excuse for those in power to avoid the deep changes required to reduce inequality.

    * Since we can’t eliminate drug and alcohol use—and it is pathological rather than normal use that should concern us—abstinence cannot be our sole, or main, goal. 

    Abstinence uber alles is particularly wrong-headed in regards to helping people with the most dangerous patterns of using mood-modifying substances. It pours our energy and resources into preventing any and all use while failing to address the nature of the worst-case problems.

    In contrast, harm reduction, which focuses on empowering people with substance use issues to improve their lives and on reducing associated deaths and disease, mitigates the problems that matter. Safer use practices, and limiting the often deadly adulteration of drugs that is inherent to illicit markets, are two important facets of harm reduction.

    * A new approach might focus on modeling successful alcohol and drug use, or at least on identifying their existence, rather than focusing on abstinence.

    Like everything we do in life, we learn how to use drugs. Southern Europeans, for example, typically learn to drink from a young age at multi-generational family meals; their societies experience far fewer alcohol problems than Northern European cultures. 

    At the opposite end of the socialization spectrum, whose use of drugs are we typically presented with? In what settings do young people in the US first navigate drug use? Our current system, born of our abstinence fixation and drug demonizing, ensures that we magnify untrammeled, chaotic drug use.


    In these and other ways, amid our abject failure to manage drug use successfully, we should be searching for new solutions.

    By elevating continuous abstinence to the pinnacle of policy and treatment concentration, the recent Cochrane review badly misses the mark and shores up the failed status quo


    Update, March 19: This article has been edited to include the review’s findings for percentage of abstinent days at 24 and 36 months.

    Photo of AA chips by Chris Yarzab via Flickr/Creative Commons 2.0

    • Show Comments

    You May Also Like

    The Invisible Majority: People Whose Drug Use Is Not Problematic

    For years, Mark* woke up each morning, made breakfast for his two young children, ...

    In 2018, the Temperance Movement Still Grips America

    Our society—even some of its most progressive elements—vilifies alcohol. This stands in opposition to ...