Leave the Past Behind by Recognizing the Effectiveness and Cost-Effectiveness of 12-Step Facilitation and Alcoholics Anonymous Free

INTRODUCTION

We thank Professor Heather (2020) for critiquing our distillation (Kelly et al., 2020a) of our recently published Cochrane Review of the effectiveness of Alcoholics Anonymous (AA) and related clinically delivered 12-step facilitation (TSF; AA/TSF). Our review subjected AA/TSF interventions to the same highly rigorous independent scientific evaluation standards used for medical and psychological treatments. AA/TSF proved at least as effective and in some cases more effective than comparison conditions on all reported outcomes and was also substantially more cost-effective (Kelly et al., 2020a,b).

For decades, AA has often been dismissed as superstitious, backwards and lacking in evidence of effectiveness. Our review indicates that it is time to turn the clock forward and leave those prejudices behind. As we acknowledge in our review, this does not mean that every person with an alcohol problem can or should recover in AA, any more than the effectiveness of cognitive behavioral treatment (CBT) carries such implications. But it does indicate that given the health and economic burden of problem drinking and the ubiquity and low cost of AA, it may be the closest thing public health has to ‘a free lunch’ (Kelly, 2017).

Heather makes some general critiques of our work that apply to all reviews of controlled research (e.g. around sampling and generalizability), and which echo points we raised ourselves in our review. We focus here, therefore, on Heather’s more specific criticisms. His raising of several of these points, however, was somewhat puzzling to us given we addressed all of these same points in our published reviews (Kelly et al., 2020a,b), as detailed more below. Nevertheless, we expand upon the three broad themes raised by Heather in the hope of clarifying them further. These were: (a) internal and external validity of our review, (b) whether our 2020 Cochrane review is an advancement over the 2006 Cochrane review on AA/TSF (i.e. Ferri et al., 2006) and (c) how applicable the results are in countries other than the USA.

INTERNAL/EXTERNAL VALIDITY

Heather (2020) argues that the finding that AA/TSF produced considerably higher rates of continuous abstinence compared with other treatments, such as CBTs, at the 12-month follow-up timepoint did not correspond with similar advantages in measures of percent days abstinent (PDA), drinking intensity (i.e. drinks per drinking day [DDD] and percentage of days heavy drinking) and alcohol-related consequences, where AA/TSF did at least as well but not better. He therefore suggests that AA/TSF may be inducing an ‘abstinence violation effect (AVE)’, because AA/TSF interventions should produce correspondingly fewer PDA, DDD and consequences. However, the hypothesis that there is an ‘AVE’ relates to the intensity or severity of use rather than the frequency or percentage of days on which someone used any alcohol. This is because someone could have only one drink and be categorized as losing a day of ‘abstinence’ (i.e. have lower PDA), though clearly such a low level of alcohol consumption is not associated with harm. Instead, what would be hypothesized under the AVE is that if people in AA/TSF did drink, they would drink more heavily on the days that they drank, because they had ‘violated’ their ‘abstinence’ rule (Marlatt and Donovan, 2005, pp. 1–44). However, this was not the case as we noted in our review. AA/TSF did just as well in producing similar rates of DDD, PDA and consequences as other treatments did (AA/TSF showed greater benefit in PDA in the years after the 12-month follow ups at 2 and 3 years after treatment (Kelly et al., 2020b; see p. 2). It should be remembered also, that DDD and alcohol-related consequences are averages based solely on those individuals who actually drank at all. Thus, as well as increasing rates of continuous and complete abstinence from alcohol, AA/TSF confers at least the same harm reduction benefits as other treatments that are more traditionally associated with reductions in use (e.g. CBTs). In other words, while more individuals in AA/TSF achieved continuous abstinence, those who were not completely abstinent did not drink more heavily, drink more frequently or experience more alcohol-related consequences so there is no evidence at all for an AVE.

Heather (2020) also states that ‘abstinence is claimed to be the crucial measure in the evaluation’. We find this puzzling, because the review made no such claim; indeed, it did just the opposite by looking at a range of alcohol outcomes beyond abstinence. He does not quote where this claim is allegedly made, but it is certainly not one that the review made. Complete abstinence was only one of several measures that were reported across studies and is no more ‘crucial’ than any other reported outcome measure. It was merely one of the most commonly reported outcome measures across studies and AA/TSF tended to consistently produce greater benefit on this measure, in particular. Of course, we do agree that there are a variety of outcomes that should be considered beyond just complete abstinence. This is why PDA is often included as an outcome in clinical trials for AUD, as was the case in these studies, as well as DDD (measure of frequency and intensity) and alcohol-related consequences. In our own recovery research (e.g. Kelly et al., 2017a), we have published many findings showing that people resolve their alcohol problem without necessarily being abstinent.

Heather (2020) also cites the Institute of Medicine’s quote, ‘There is no single treatment approach that is effective for all persons with alcohol problems’. This is precisely the view we have always held and which we express in our review. For example, on page 9 of Kelly et al. (2020a) we note, ‘If people with AUD are opposed to attending AA, despite the strong evidence for its potential to aid recovery, clinicians might consider linkage to alternative mutual-help organizations as they may confer benefits at similar levels of engagement’, and regarding those with AUD, ‘…those suffering can often have different preferences as to the kinds of recovery pathways they wish to follow’ (Kelly et al. 2020a, p. 9). We also echoed this sentiment in our Cochrane review (Kelly et al., 2020b, p. 36).

Similarly, we are equally puzzled by Heather (2020) criticizing the review for not considering ‘quality of life’ when it was a registered outcome in our publicly-available protocol (Kelly et al., 2017b). As we stated at multiple points in the review, none of the controlled studies included quality of life measures. That there was no such data to summarize may be a reason to critique the original studies but would not support the claim that we chose to avoid quality of life measures. Indeed, our review called for future studies to include quality of life measures, and we have previously called for such measures to be included in treatment and recovery research (Kelly et al., 2018).

Heather (2020) also alludes to ‘low quality’ ratings in some areas of the review. We should note that in conducting the sensitivity analyses (i.e. comparing consistency of results across dimensions of degree of manualization, study design and theoretical orientation of comparison treatments), Cochrane’s quality of evidence ratings are automatically lowered, because it means including fewer studies in each comparison, which the Cochrane system automatically downgrades. Consequently, overall, this resulted in more conservative ratings of evidence quality than would be the case had we combined more studies in a single analysis.

It is also noteworthy, that when evaluated along rigorous dimensions used to help to establish scientific rigor and causality (i.e. Bradford Hill, 1965), the set of findings emanating from our updated rigorous review also demonstrate what Bradford Hill (1965) describes as ‘coherence’, and ‘consistency’ with results from other less rigorous quantitative and systematic reviews of the AA/TSF literature (e.g. Emrick et al., 1993; Kelly, 2003; Kaskutas, 2009). In other words, from whatever research angle one views it, AA/TSF is shown to do as well or better for less money.

ADVANCE ON PREVIOUS COCHRANE REVIEW

Heather (2020) also questions whether the new review is in fact ‘an advance on previous findings’ over the previous 2006 review (Ferri et al., 2006), because the new review ‘rested on different outcome measures’. Again we are puzzled, because every outcome in the original review was included in the new review (Kelly et al., 2020b). Revisiting 16-year old conclusions using data from more than triple the number of research participants and trials is clearly an advance, as is expanding the range of patients (e.g. severe inpatients, less severe outpatients, psychiatrically comorbid patients, young adults, women and veterans), settings (hospitals, community clinics, inpatient, outpatient, residential and detoxification), clinicians (counselors, psychiatrists, psychologists, peers and social workers) and geographic locations (from all across the USA and internationally) from which data are drawn. Further advances were moving from a review that included many studies of poor methodological quality to one primarily composed of randomized controlled trials, and, breaking new ground by being the first to summarize cost-effectiveness studies.

A‌A/TSF IS LARGELY AN AMERICAN PHENOMENON

Heather (2020) also states that the majority of studies included in the review were conducted in the USA and that AA is largely an American phenomenon seemingly discounting the international significance of the results. However, AA is indeed a global organization and its widespread adoption and growth throughout the world (Alcoholics Anonymous World Services, 2001), Mäkela (1996) suggests that AA’s 12-step model is cross-culturally adaptable and adoptable. The review did include randomized controlled trials from Norway and the UK as well as from the USA. That said, we agree that more research is needed on the efficacy of AA in different countries, which is why we highlighted this need in our Cochrane review (Kelly et al., 2020b): ‘Most of this research was conducted in the USA; further research is needed to determine the degree to which results observed here differ in other countries.’ (Kelly et al., 2020b, p. 36).

CONCLUSION

In sum, we thank Professor Heather (2020) for his interest in this topic and his comments on our work. We agree that neither AA/TSF nor continuous abstinence should be considered the only worthy interventions and outcomes. We caution readers on this point in our review (Kelly et al., 2020b, p. 36). What is significant and remarkable, however, is that the strength of the scientific evidence is now more than sufficient to support the supposition that AA/TSF is a valid approach that is at least on par with, or superior to, other commonly used clinical interventions (e.g. CBTs, motivational interviewing [MI]-based treatments) and is more cost-effective. AA’s free and widespread availability in many middle and high-income countries globally where the burden of disease attributable to alcohol is high, suggest AA/TSF may now be considered one of many valid and helpful approaches and that the findings from the Cochrane review are good news for public health.

FUNDING

NIH National Institute on Alcohol Abuse And Alcoholism (NIAAA) 5K24AA022136-07 - Mentoring in Patient-Oriented Addiction Research.

CONFLICT OF INTEREST

The authors have no conflicts of interest to report.

References