The evidence-base for 12 step recovery

There’s a fresh round of attacks on AA as pseudo-science in need of sober debunking. All based on one book that is getting impressive publicity. The book may contain references to support its attacks, but the interviews and articles do not. The absolute language (“everyone” and “never”) hint that this may not be the objective analysis it’s reported to be. Anna David makes the case that it’s a “hit job”.

The one source he does identify is the Cochrane Review.

Problems with the Cochrane Review of AA

About 5 years ago, I saw Sarah Zemore give a presentation that very effectively rebutted the Cochrane Review of the evidence for the effectiveness of 12 step groups. It was powerful and well organized. Here are her slides and here’s video of the presentation. (It’s old school. You have to download a mega-file.)

She made the following points in her introduction:

• It was limited only to randomized trials and ignored the overwhelming observational evidence.
• It included one of Zemore’s studies which was NOT a randomized study of AA.
• She acknowledged that the randomized evidence is ambiguous.
• Randomized trials of AA are hard to do because some subjects in other groups end up participating in AA. This happened in Project MATCH.
• The Cochrane Review did not find Twelve-step Facilitation ineffective. It found it no more effective that CBT and MET.
• Finally, she cited 4 randomized studies of Twelve-step Facilitation: The outpatient arm of Project MATCH, a study by her colleague Kaskutas, and two others that I missed.

It was important for me because supporters of Twelve-step Facilitation are too often painted as the equivalent of intelligent design advocates. It’s just not so and the evidence in this presentation made this unequivocally clear. Twelve-step Facilitation is not the only approach that works, but it’s an evidence based practice.

AA and the 6 Formal Criteria for Establishing Causation

Zemore’s content was summarized in an article about the conference:

Zemore presented Kaskutas’ (2009) article, “Alcoholics Anonymous Effectiveness: Faith Meets Science.” Noting diverging conclusions about AA’s effectiveness in the literature, Zemore presented Kaskutas’ approach to evaluating the evidence about AA, highlighting many categories of evidence. She took as the framework for evaluating the research 6 formal criteria for establishing causation described in Mausner and Kramer (1985): (1) strength of the association, (2) dose-response relationship, (3) consistency of the association, (4) correct temporal ordering, (5) specificity of the association, and (6) coherence with existing information. Strong evidence for Criteria 1– 4 and 6 was presented. Evidence for Criterion 5 was reported as mixed. Emphasis was made on the totality of the evidence in favor of AA as a causal agent of abstinence. This quote from the 2009 article summarizes the findings:

… the evidence for AA effectiveness is strong: rates of abstinence are approximately 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); these relationships are found for different samples and follow-up periods (criteria 3, consistency); prior AA attendance is predictive of subsequent abstinence (criteria 4, temporal); and mechanisms of action predicted by theories of behavior change are evident at AA meetings and through the AA steps and fellowship. (criteria 6, plausibility). (Kaskutas, 2009, p. 155)

Reviewing the Evidence

The article goes on to summarize the knowledge presented as follows:

• The preponderance of evidence supports the causal pathway that AA attendance leads to abstinence (Kaskutas, Zemore).
• 12-Step affiliation significantly enhances the odds of sustaining abstinence for multiple years among polysubstance-dependent individuals (Laudet).
• 12-Step involvement yields benefits above and beyond meeting attendance (Kaskutas, Zemore, Laudet)—and this is especially important for women (Laudet).
• 12-Step attendance declines over time (Laudet, Kelly). Patterns of AA and NA attendance mirror patterns of treatment attendance with multiple stop-and-start episodes (Laudet).
• A substantial minority of recovering substance abusers in the community do not participate in 12-Step programs (Laudet).
• For adolescents, the relationship between AA meeting attendance and percent days abstinent increase in linear and positive direction at 6 months and 12 months posttreatment (Kelly).
• A combination of treatment and AA is most effective (Kaskutas, Zemore).
• Among adolescents, early posttreatment attendance, even in relatively small amounts, predicts long-term helpful outcomes. Consistent attendance over time predicts favorable outcomes (Kelly).
• Three or more AA/NA meetings per week are optimal and associated with complete abstinence. However, even one or two meetings per week are associated with sharp increases in abstinence (Kelly, White).
• Of 1.9 million people who are addicted to drugs or alcohol, only 18% are alcohol only and only 36% are drug only (White).
• Those who state AA is helpful have better drinking outcomes. Those who state AA is not helpful have poorer drinking outcomes (Robinson).
• Addiction severity predicts participation in AA and NA among adults (Robinson) and adolescents (Kelly).
• Individuals who benefit from AA identify the importance of being in a group of sober people, see AA as a source of support, benefit from others’ experiences, and search for AA meetings and members with whom they find compatibility (Robinson).

Mutual Aid Mechanisms for Change

Bill White has summarized research on AA’s various mechanisms of change (Look at page 128. It includes citations.):

• problem recognition and commitment to change;
• regular re-motivation to continue change efforts;
• counter-norms that buffer the effects of heavy drinking social networks and alcohol and other drug use promotion in the wider culture;
• sustained self-monitoring;
• increased spiritual orientation;
• enhanced coping skills, particularly the recognition of high-risk situations and stressors;
• increased self-efficacy;
• social support that offsets the influence of pro-drinking social networks;
• helping others with alcohol and other drug problems;
• exposure to sober role models and experience-based advice on how to stay sober;
• participation in rewarding sober activities;
• 24-hour accessibility of assistance; and
• potentially lifelong supports that do not require financial resources.

The Bottom Line

We’ve still got a lot to learn, but here’s some of what we know:

• Is 12-step effective at initiating recovery? – YES
• Are other approaches effective at initiating recovery? – YES
• Is 12-step involvement associated with maintaining abstinence? – YES
• Are other approaches associated with maintaining abstinence? – I haven’t seen the evidence.
• Do 12-step programs work for everyone? – NO
• Does anything work for everyone? – NO
• Are there other paths to recovery? – YES
• Do some people initiate recovery with one approach and maintain recovery by other means? – YES

We should continue to research 12-step recovery and other approaches. Learning more about the factors that contribute to the benefits of 12-step involvement might help in developing recovery maintenance strategies to help people who won’t attend 12-step groups or don’t benefit from 12-step groups.

UPDATE: DJMacUK’s comment is so good, I wanted to add it to the post to be sure you don’t miss it.

As far as randomised controlled trials go, it’s not just contamination of the control group that makes it hard to study AA. It’s a bit of a catch 22 with complex interventions like mutual aid. Keith Humphries makes good points on this: Some of this is quote and some paraphrase.

It is difficult to generalise because, most notably, of their extensive exclusion criteria ending up with a small and unrepresentative subset of patients. E.g. Exclude those with mental health disorders, physical health problems: exactly the sort of patients seen every day in treatment.

The common conviction that rcts always generate more accurate estimates of treatment effects is simply incorrect. The NEJM, perhaps the most respected source of controlled clinical trials in the world recently published literature reviews comparing the observed outcomes of medical treatments that had been studied both by randomised trials and by other evaluation approaches. Across methodologies, outcome results were almost always similar (Benson & Hartz, 2000; Concato, Shah and Horwitz, 2000)

Shifting sands: The idea that treatments are applied by outside forces before change begins and are then not affected by any subsequent changes in the patient is poorly matched to chronic dynamic disorders like addiction in which patient factors (e.g motivation, progress or regress) and treatment factors are in constant interplay (Moos 1997) Such processes are easier to understand when patients have the option of choosing which treatments they want, how they want them, when they want them and so forth, all of which is impossible in the context of a typical RCT.

RCTs depend on professional control of who receives the intervention and when and by definition, mutual aid is not professionally controllable. Participation in self help cannot specifically be denied to ‘controls’ in the way that a medication or procedure can be. Patients in the non mutual aid group arm have often gone to mutual aid anyway (this contaminated some of the project match data)