Some other recent criticism of AA in the press

I have found other criticism of AA in the mainstream press over the last year which I should address here in this blog.

Naltrexone redux

A recent New York Times opinion piece (archive link) brought out the same old tired “AA (or abstinence) doesn’t work but Naltrexone allows alcoholics to drink like gentlemen” chestnut.

While cherry picked studies show remarkably high success rates for Naltrexone, this is offset by other studies which do not replicate those results. Naltrexone, like AA, appears to be effective for a subset of alcoholics, but, as per Cochrane, it only helps about 11% of alcoholics. 

The most disturbing thing about studies showing Naltrexone showing controlled drinking among alcoholics is that they consistently do not have long term follow-up periods. For example, that New York Times opinion piece links to two different studies which supposedly show alcoholics successfully drinking moderately again. In both cases, the final follow-up is one year after treatment.

As someone who remembers Pendery 1982 all too well, positive results for moderate drinking without long term follow-ups are not very encouraging. Sure, anything someone throws a dead cat, they will hit a study of alcoholism showing alcoholics successfully engaging in “moderate drinking” over the short term, but the few long-term follow-ups I have seen show that moderate drinking is not a long-term solution to chronic alcoholism.

As a point of comparison, the 1990s Project MATCH randomized study showed increased abstinence for people who engaged in AA-centered therapies three years after treatment (as per its data as presented in Cochrane 2020), and Moos and Moos 2006 is a longitudinal study showing that Alcoholics Anonymous keeps alcoholics sober 16 years later.

Another thing: That New York Times opinion piece claims that only 25% of alcoholics will achieve abstinence, but the study it links to supporting that outdated 25% figure doesn’t actually conclude this; it claims a 25% success rate for abstinence, and links to an old article from 2005 using results that are nearly two decades old. More recent figures from Cochrane show a 42% success rate.

Abstinence violation effect

There have been two (open access) different letters (paywall) recently published which make the same criticism of Cochrane 2020: They claim that there is an “abstinence violation effect” where the subjects undergoing AA-based treatment who are not abstinent somehow do worse when they drink again than subjects not undergoing AA-based treatment. In other words, the accusation is that AA may increase abstinence, but overall AA doesn’t improve things for alcoholics, because the alcoholics not abstinent supposedly drink more heavily.

This “abstinence violation effect” is actually a myth, as one of the authors of the 2020 Cochrane review on AA explains. Since this letter is paywalled, I will quote the most important part of that letter here:

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

Point being, AA overall improves things for alcoholics. 42% are straight up abstinent from alcohol a year later, and those who do not achieve abstinence do not, repeat not have worse consequences from their drinking.

Moderate drinking redux; also more on the Sinclair Method

Let me just post some of my recent disqus postings:

The fact is that the Sobell 1973 study [a study which claimed alcoholics could moderately drink again] was so bad, a fraud investigation was opened. As a 1984 New York Times article put it (look for “Panel finds no fraud by alcohol researchers”):

>>The panel also said it had found ''several errors'' and ''use of ambiguous terminology'' in scientific publications by the Sobells, which indicated that they were ''careless in preparing their manuscripts for publication.'' But the panel said it had seen no evidence that these errors and ambiguities ''represent scientific misconduct - defined as 'fraudulent or highly irregular research practices.' ''<<

Anyway, facts are facts. Those “moderate” drinkers from the Sobell 1973 study were either drinking alcoholically again, or were dead from drinking, or were abstaining from alcohol when a 10-year-followup was performed.

--

Sobell 1973 and the moderate drinkers: Four of those “moderate drinkers” died. Also: Six were abstinent, nine of the “moderate” drinkers were engaging in out of control drinking. There was only one person who, ten years later, could be considered a “moderate” drinker. Please read Pendery 1982 or any of the numerous books which correctly cite it.

Sobell 1973 in comparison to the non-moderate drinkers: This is cherry picking data. The Sobells only compared the fatalities of the “moderate” drinkers with the non-moderate drinkers. I have not seen the abstinent or out-of-control drinking figures for this other group; and it is telling that the Sobells never shared (or, it would seem, even bother to get) those figures.

Since there were only 20 subjects in each study, whether six or four died is an error of approximation, nothing more.

Davies 1962: These “moderate” drinkers also reverted to serious drinking again. Reference: G. Edwards. "D.L. Davies and 'Normal drinking in recovered alcohol addicts': the genesis of a paper" Drug and Alcohol Dependence 35.3 (1994)

Rand report: The 1980 four-year followup shows many of the 1976 Rand report moderate drinkers relapsing.

Moderation management: No peer-reviewed science was ever done with this group, but it’s very telling that its founder drank heavily, to the point police took her to detox in handcuffs while she was still the leader of Moderation Management. She soon afterwards killed two people while driving drunk and eventually killed herself [1].

Point being, pretty much any study showing successful “moderate” drinking, when followed up, shows those same “moderate” drinkers drinking heavily again.

[1] There are claims that the Moderation Management founder was a member of AA at the time of her crash, but as per her autobiography, she did not work the 12-step program in any meaningful way. Instead, she was coming up with excuses for not going to meetings.

--

In response to a claim that the “moderate” drinkers in Sobell 1973 had better success rates than people trying abstinence:

In terms of more recent studies, people who choose to work the AA program have between a 75% (Vaillant 1995) and 67% (Moos and Moos 2006) success rate. The Sobell moderation group had, by comparison, a 30% abstinence success rate and a 5% moderation success rate.

--

In response to a claim that the Sinclair method works better when trying complete abstinence, and works better the longer it is applied:

OK, let me go back to PMC3970823. They use a ”magic number” called “g” which is summarized near the end: “the aggregated Hedges’ g for naltrexone and acamprosate compared to placebo was 0.209 (CI: 0.157 – 0.262) – indicating a small but significant effect. In comparison to other medications prescribed for mental health (e.g., for depression), the effects of naltrexone and acamprosate are somewhat smaller”

So, this information in mind, let’s look at table 2 in PMC3970823. Naltrexone has an abstinence aggregate of .116—which is not very different than a placebo. However, the numbers for “heavy drinking” (.189), craving (.144), and both heavy drinking and craving (.180) are not much better—they are “small but significant.”

It also looks like, when looking at multiple studies, the effect diminishes when doing follow-ups: “naltrexone studies tended to have effect sizes for heavy drinking outcomes at the last follow-up point (g = .135, k = 6) that were slightly smaller compared to end-of-treatment (g = .189, k = 39). Although only two studies provided data, the naltrexone-placebo effect sizes for craving at the last follow-up point were close to zero (g = .053).” In other words, this meta-study didn’t see naltrexone helping people not drink heavily over longer periods.

In light of Pendery 1982, and in light of what happened with the founder of Moderation Management (while she claimed to drink moderately without problem for years, she killed two people while driving drunk, and eventually killed herself), when people suggest moderation as an outcome for alcoholics, I will remain very very skeptical until when and if we have 10-year-followups with very high “g” values and very low “p” values.

The Sinclair method revisited

An advocate for the Sinclair method pointed out, in an online debate, the study “Targeted Naltrexone for Problem Drinkers.”  Here is my response:

That 2011 study (“Targeted Naltrexone for Problem Drinkers”) is a 12-week trial, and the study considers “moderate drinking” a successful outcome. This is a very short trial period, and short-term studies since the 1960s showing alcoholics “successfully” drinking moderately have, if followed up, shown the same drinkers having serious drinking problems five or 10 years later (the most famous example of this is Sobell 1973, showing “moderate” drinking among alcoholics, but Pendery 1982 showed that the same “moderate” drinkers were either dead, abstinent, or drinking heavily a decade later).

The meta-study (a study looking at other studies) “Meta-analysis of naltrexone and acamprosate for treating alcohol use disorders: When are these medications most helpful?” (PMC3970823) points out that “Several reviews and meta-analyses have reported small or mixed effects for each medication (both naltrexone and another medication called acamprosate), and treatment providers cite concerns over efficacy as one barrier to greater medication use.”

This isn’t the first study which showed good results for Naltrexone, but considering the number of other studies which show only a modest effect, and considering the need for a long term follow up to see if an alcoholic is in fact successfully moderately drinking, I will need to see studies with five-year or 10-year follow ups showing continued “moderate” drinking before I will believe this drug is a functional way to let alcoholics control their drinking.

Naltrexone, acamprosate, and the Sinclair Method

Since I have been in an online debate with someone who claims AA doesn't work and advocates the Sinclair Method, I should go over the newer medication assisted therapies.

The Sinclair Method is simply suggesting to use the medication Naltrexone along with engaging in moderate drinking. There have been studies showing that people can more easily moderately drink when taking Naltrexone, but the effect has also been described as being modest. 

The Sinclair Method has the same problem all other moderation methods have: Yes, it shows some, albeit modest, level of short term success, but we do not have long term studies. Short-term studies advocating moderate drinking are unreliable; long term follow ups show the "moderate" drinkers drinking heavily again. Until we have a 10-year follow up study showing people still using Naltrexone to moderate their drinking, or being able to abstain from drinking altogether after using Naltrexone for a while, I am very skeptical. We have had too many people die deceiving themselves with "moderate drinking" approaches; I have not seen evidence that Naltrexone is a silver bullet that will change things.

Naltrexone is not an effective therapy for people who want to abstain; there are two studies showing this drug has little to no effect when the patient pursues abstinence. 

Acamprosate, on the other hand, seems to work better when the goal is abstinence. The effect is "small but significant", but it may help alcoholics achieve abstinence when used with other therapies.

I think, when used with the 12 steps and regular Alcoholics Anonymous, Acamprosate may be a helpful drug for achieving long-term sobriety. Bill Wilson had no objection to the use of medications to successfully work the AA program. That said, while the effect for both drugs is there, it's small.