How Effective Is AA?
"Of alcoholics who came to A.A.
and really tried, 50 percent got sober at once and remained that way; 25
percent sobered up after some relapses . . ."
—Bill Wilson in Alcoholics
Anonymous, p. xx
"Everything you know is wrong."
The Problem of Definitions
AA an effective treatment for alcoholism? That seemingly simple question
is far more difficult to answer than one would expect. A major problem
is the difficulty of defining the terms "alcoholism" and "alcoholic." Since
the terms were invented over 100 years ago, a great variety of definitions
have been offered, and there is still no uniformity of opinion among the
"experts" about what constitutes alcoholism nor about what constitutes
an alcoholic. The safest thing that can be said is that definitions are
largely arbitrary and can (and do) change over time. For example, in the
"Big Book," Bill Wilson mentions "a certain type of hard drinker. He may
have the habit badly enough to gradually impair him physically and mentally.
It may cause him to die a few years before his time." Wilson goes on to
say that this person is not a real alcoholic because he can learn to "stop
Needless to say, virtually all AA members, as well as a very large majority
of alcoholism professionals, would now label such a person "alcoholic."
indication of the difficulties involved in defining the word "alcoholic"
can be seen in the wildly varying estimates of the number of alcoholics
in the United States. In the 1986 best seller, The Courage to Change,
Dennis Wholely estimated that there were 20 million American alcoholics
at that time (in other words, over 11% of the adult population). Wholely's
figure is twice as high as the estimate of 10 million which is found in
many professional journal articles and alcoholism reference texts published
in the 1970s and early 1980s, and which is still occasionally cited. A
facts sheet circulated by the NCADD, which I obtained in 1991, estimates
that there are 12.1 million heavy drinkers exhibiting one or more of the
signs of alcoholism; a 1997 NCADD facts sheet lists 13.9 million alcoholics.
And if you accept the commonly cited figure that 10% of American adults
are alcoholics, you arrive at a current figure of 19.7 million.
primary reason why these estimates vary so greatly is that "alcoholism"
is an elusive concept with several defining factors, the limits of which
are seemingly arbitrary, with the exceptions of physical damage caused
by alcohol consumption, physical dependence (habituation to the extent
that physical withdrawal symptoms appear if alcohol consumption ceases),
and tolerance (the need to drink a larger amount than the average social
drinker in order to reach a similar state of intoxication). In addition
to these physical symptoms, commonly cited defining factors include amount
consumed per day, number of drinking days per month, number of days intoxicated
(according to some number-of-drinks benchmark), legal problems (e.g., DUIs),
employment problems, and family/relationship problems. Obviously, any definition
based upon such factors must be imprecise and at least somewhat arbitrary.
For example, what is the precise amount of alcohol consumption which separates
the alcoholic from the social drinker? And what relation does alcohol consumption
have to the other defining variables? Would someone who drank 7 ounces
of alcohol per day but who had relatively minor problems in other areas
be defined as an alcoholic? Would someone who drank only half that amount
but had severe problems in other areas be defined as an alcoholic? It's
difficult to view answers to such questions as anything other than arbitrary.
this reason, researchers in recent years have begun to use the somewhat
more precise terms "alcohol dependence" and "alcohol abuse" instead of
"alcoholism." "Alcohol dependence" refers to the presence of physical dependence
and/or tolerance (as well as, almost invariably, additional problems),
while "alcohol abuse" refers to the presence (and magnitude) of at
least a certain number of the other defining factors. This distinction
represents a definite step forward, though definitions of "alcohol abuse"
are still somewhat arbitrary, and almost certainly will remain so. But
these more precise definitions do allow for more precise estimates.
How Many Alcohol Abusers?
figures commonly cited for the number of alcohol-dependent persons in the
U.S. and Canada are fairly uniform, with most estimates that I've seen
being in the 5% range. A 1995 National Institute on Alcohol Abuse and Alcoholism
(NIAAA) news release estimates the percentage of alcohol-dependent persons
in the U.S. at 4.38%,ii
while the Addiction Research Foundation (ARF), based in Toronto, estimated
alcohol dependence among the local population at 5.3%.iii
Estimates of the number of alcohol abusers vary more widely, with the NIAAA
estimate being that only 3.03% of the population are alcohol abusers, while
the ARF estimates the number of alcohol abusers at 5%.
Characteristics of AA Members
thing that is certain is that the typical AA member today is different
than the typical AA member in 1940. In the early days of AA, members were
primarily "low-bottom" alcoholics who had been hospitalized for their drinking
problems, and whose drinking had had devastating effects on their lives.
At present, at least a large minority, perhaps a majority, of AA members
are "high-bottom" problem drinkers who were never physically dependent
upon nor tolerant of alcohol and who still functioned reasonably well socially
and economically at the time they quit drinking. Thus, a welldesigned
study of the effectiveness of AA today would very probably yield a different
result than a similar study conducted 50 years ago, simply because of the
differences in the makeup of both AA's membership and the much-expanded
pool of drinkers from which it now draws.
the trend toward inclusion of those with shorter and shorter and ever-less-serious
drinking problems in AA, the composition of AA's membership will very
likely continue to change for some time to come. (According to AA's 1996
membership survey, there are now roughly 12,000 teenage AA members.
In the 1930s, AA's early members would have considered the idea of teenage
One question which arises from this is what percentage of AA's members
are now "real alcoholics"? A complicating factor is that at least
some disturbed persons whose primary problems are almost certainly not
alcohol related attend AA because it's an easy way to meet their social
needs. A further complicating factor is that a very high percentage of
AA's current members, almost certainly at least a third, and probably more
than 40%, are coerced into membership.iv
changing makeup of AA's membership is, however, a minor problem compared
with several others. The most important problem is that in attempting to
gauge the effectiveness of AA it's very difficult to tell if you're gauging
results due to the AA program or results due to the characteristics of
AA's membership. There are several factors predictive of a positive outcome
to alcoholism treatment—motivation, middle class status, marital stability,
employment, relatively mild and short-term problems with alcohol, and absence
of serious mental disturbance being probably the most important—with most
being found in higher-than-average percentages (for problem drinkers) in
AA's membership; and it should be noted that these factors are predetermining
factors which were operative in a great many AA members before they
joined AA. An indication of the importance of these predictive factors
is found in Frederick Baekeland's evaluation of different varieties of
alcoholism treatment. Baekeland compared studies of four group therapy
programs serving high socioeconomic status (SES—an important prognosticator
of treatment outcome) patients with studies of four group therapy programs
serving skid row alcoholics and other low SES patients. The improvement
rates of the programs serving skid row alcoholics were only 18%, 7.9%,
2% and 0%, while the improvement rates of the programs serving high SES
patients were 32.4%, 46.4%, 55.8%, and 68%.v
is almost universally recognized in treatment literature, the most important
favorable prognosticator is "motivation." Like most cliches, the truism
that "once you admit you have a problem, it's half-licked," seems to have
a basis in fact. Simply showing up at an AA meeting implies that an individual
recognizes that s/he has a problem, and in itself this self-selection seems
predictive of a successful outcome. Further, certain aspects of AA are
so unpleasant—especially the religiosity, anti-intellectuality, and the
gas chamber-like, tobacco smoke-filled atmosphere at many meetings—that
continued attendance in itself implies a high degree of motivation, at
least for nonreligious and critically minded (not to mention nonsmoking)
Estimates of AA's Effectiveness
factors, such as "motivation," are a serious problem, but it does seem
possible to draw at least tentative conclusions about the effectiveness
of Alcoholics Anonymous. A good starting point is AA's most recently announced
membership figures. As of January 1, 1996, AA claimed 1.251 million members
in the U.S. and Canada,vi
while there were approximately 218 million individuals 18 years of age
and over in the two countries at that time. Taking the ARF estimates of
the percentages of alcohol abusers and alcohol-dependent persons and multiplying
them by total population figures yields a total of roughly 22 million individuals
with alcohol problems in 1996; doing the same calculations using the NIAAA
percentages yields a total of roughly 16.13 million persons. Taking these
as high and low estimates of the number of alcohol abusers, as of the date
of the last avail-able AA membership figures, somewhere between 5.7% and
7.7% of U.S. and Canadian "alcoholics" belonged to AA. And the percentage
of those who will reach the AA goal of lifelong abstinence is much lower
noticeable feature of AA is that a large number of its members have been
in the organization for a relatively short time. Based on my attendance
at AA meetings in San Francisco in the late 1980s, I would estimate that
over 50% of those attending meetings in that city at that time were members
for less than one year and, in fact, that a majority were members for only
a few months. The situation appears to have change little in recent years.
(The discrepancy between my observations and AA's claim that only 27% of
its members have less than one-year's abstinence is probably accounted
for by AA's astoundingly high dropout rate; because of it, one constantly
sees new faces showing up at AA meetings, with many of them sticking around
for relatively few meetings.)
estimate, however, isn't too far out of line with the figures given by
Bill C. in a 1965 article in the Quarterly Journal of Studies on Alcohol.vii
In it, he reports that of 393 AA members surveyed, 31% had been sober for
more than one year; 12% had been sober for more than one year but had had
at least one relapse after joining AA; 9% had achieved a year's sobriety;
6% had died; 3% had gone to prison; 1% had gone to mental institutions;
and 38% had stopped attending AA. What makes these numbers even more dismal
than they appear is the fact that Bill C. defined a member as someone who
attended 10 or more AA meetings in a year's time. When you take into account
the "revolving door effect," it becomes apparent that far more persons
attended AA meetings than the 393 "members" Bill C. lists. It seems quite
probable that he picked the figure of 10 meetings in a year as a membership
criterion because AA's success rate would have been revealed as microscopic
if he had used a smaller number of attendances as his membership-defining
device. (It should also be mentioned that attendance at 10 meetings in
itself seems to imply a fairly high degree of motivation.)
success rate calculated through analysis of the 1996 AA membership survey
is hardly more impressive. The survey brochure indicates that 45% of members
have at least five years' sobriety. Using the figure of five years' sobriety
as the criterion of success, one arrives at an AA success rate of approximately
2.6% to 3.5% (in comparison with the total number of "alcoholics" in the
U.S. and Canada). And the success rate is lower than that if one defines
"success" as AA does—as lifelong abstinence.
could be argued that this is an unfair way of evaluating the effectiveness
of AA, and that only "alcoholics" who have investigated AA should be considered.
That's a reasonable argument, but there's evidence that a very high proportion
of "alcoholics" have at one time or another checked into AA. Anyone
who has attended many AA meetings can testify that droves of newcomers
show up, attend one, or a few, meeting(s), and then are never seen again—the
"revolving door effect." As well, roughly 270,000 individuals accused or
convicted of drunk driving and other alcohol-related crimes are coerced
into 12-step treatment every year in the United States.viii
Based on the sheer numbers of such persons, it seems probable that well
over 50%, perhaps as many as 90%, of American and Canadian problem drinkers
investigate AA at some time during their drinking careers.
statistical evidence to indicate that this is so. Well known researcher
Robin Room, of the Addiction Research Foundation, reports that a 1990 survey
of 2058 Americans aged 18 and over revealed that 9% of American adults
have attended an AA meeting at some time in their lives, and that an astounding
3.4% claimed to have done so in the previous year.ix
(The latter percentage is almost certainly incorrect.x)
If Room's 9% figure is even close to being correct, it's good evidence
that a very high percentage of U.S. and Canadian alcohol abusers have attended
AA at least once. In 1996, 9% of American and Canadian adults corresponded
to roughly 19.6 million individuals. This figure, when compared with the
previously mentioned estimates of alcohol abusers and alcohol-dependent
persons (16.13 to 22 million individuals), provides persuasive evidence
that the percentage of "alcoholics" who have tried AA is high indeed—and
that AA's success rate is very low.
AA's Triennial Surveys
own statistics provide perhaps the most persuasive evidence that AA's success
rate is minuscule. Since 1977, AA has conducted an extensive survey of
its members every three years (though the survey scheduled for 1995 was
conducted in 1996). These surveys measure such things as length of membership,
age distribution, male-female ratio, employment categories, and length
of sobriety. Following the 1989 survey, AA produced a large monograph,
"Comments on A.A.'s Triennial Surveys,"xi
that analyzed the results of all five surveys done to that point. In terms
of new-member dropout rate, all five surveys were in close agreement. According
to the "Comments" document, the "% of those coming to AA within the first
year that have remained the indicated number of months" is 19% after one
month; 10% after three months; and 5% after 12 months.xii
In other words, AA has a 95% new-member dropout rate during the first year
success is defined as one-year's sobriety, on the face of it this 95% dropout
rate gives AA a maximum success rate of only 5%; and a great many
new members do not remain continuously sober during their first year in
AA, which causes the apparent AA success rate to fall even lower. Of course,
many of the 95% who drop out within the first year are probably "repeaters"
who have previously investigated AA, and this would increase the apparent
AA success rate; but at least for the present there is no way to know what
percentage of the dropouts are repeaters. Additionally, at least some of
the 95% who drop out of AA during their first year do manage to sober up;
but to date there's no way to know what their numbers are. As well, it
seems quite probable that most of those who drop out early in the program
do so because they dislike and disagree with AA, so it could be argued
that most of them who overcome their drinking problems do so in spite of,
not because of, AA. Finally, at least some curiosity seekers and relatives
of alcohol abusers show up at meetings, and this would further increase
the apparent AA success rate. But to date, there are no reliable figures
on what percentage of those who "walk through the door" fit those categories—though
my personal estimate, and that of researcher/author Vince Fox, is that
no more than 10% of new faces at AA meetings belong to relatives or curiosity
thing, however, is certain: An extremely high percentage of American drinkers
who have been hospitalized for alcoholism or who have participated in other
institutional alcoholism programs have participated in Alcoholics Anonymous.
The number of patients treated for alcoholism is now approximately 950,000
which (because 12-step treatment is used in well over 90% of institutional
programs) is a good indication that the proportion of alcohol abusers who
have been exposed to AA is very high. It should also be kept in mind that
in most parts of the country convicted drunk drivers are still routinely
forced to attend AA as a condition of probation, which pushes the percentage
of alcohol abusers exposed to AA even higher. Further, in most areas AA
is the only widely available—and widely media-promoted—alcoholism self-help
group, so AA has a very high volume of "walk in" traffic.
let's give AA the benefit of the doubt and estimate that only 50% of U.S.
and Canadian alcohol abusers have tried AA. That would double the success
rate calculated earlier (based on the total number of U.S. and Canadian
alcohol abusers), and it would increase to 5.2% to 7.0% if the criterion
of success is defined as five years' sobriety.
a worst case scenario, where 90% of U.S. and Canadian alcohol abusers have
tried AA, where success is defined as five or more years of sobriety, where
45% of AA members have been sober for five or more years (as AA indicates),
and where there are 22 million alcohol abusers in the two countries, the
AA success rate would be about 2.9% (and even lower than that if the criterion
of success is lifelong sobriety rather than five years' sobriety).xv
The true success rate of AA is very probably somewhere between these two
extremes, depending, of course, on how one defines "success"; that is,
AA's success rate is probably somewhere between 2.9% and 7% (of those who
have attended AA).
is far from impressive, especially when compared with the rate of "spontaneous
remission." Contrary to popular belief, "alcoholism" is not a progressive
and incurable "disease." Many studies have been conducted on so-called
spontaneous recovery by "alcoholics" (that is, recovery without treatment,
which can refer to achievement of either abstinence or controlled drinking),
and the consensus of these studies is that "spontaneous" recovery occurs
in a significant percentage of alcohol abusers, though the calculated rates
of recovery vary considerably.xvi
Other consistently supported conclusions are that the rate of alcohol
abuse and alcohol dependence (or, to use the older term, "alcoholism")
declines far faster than can be explained by mortality among individuals
past the age of 40,xvii
and that "spontaneous" recovery normally occurs for identifiable reasons.
In many cases, remission comes suddenly after a particularly dangerous
or humiliating incident shocks the drinker into realization of the seriousness
of his or her drinking problem. In other cases, recovery occurs as a result
of religious conversion or as the result of an "existential" decision to
quit based on a gradually increasing realization of the seriousness of
One review of available literature estimates the rate of spontaneous recovery
at 3.7% to 7.4% per year.xix
More recently, a large-scale longitudinal study of over 4,000 adults with
prior, significant, diagnosable alcohol dependence (the National Longitudinal
Alcoholism Epidemiological Survey, conducted by the Census Bureau) reported
that 20 years after the onset of alcohol dependence, 90% of those who never
received treatment were either abstinent or "drinking without abuse or
Compared with these figures, the abovecalculated rate of recovery
via AA is not impressive. In fact, it appears to be no higher—and could
actually be lower—than the rate of spontaneous re-mission.
Controlled Studies of AA's Effectiveness
haven't there been scientific investigations of the effectiveness of AA?
There have been, but there haven't been many. One reason for this could
well be that "A.A. does not like to have researchers around,"xxi
that it is highly reluctant to "open its doors to researchers."xxii
Whatever the truth of these charges, to date there have been only two well-designed
studies of the effectiveness of AA—that is, studies which have included
control groups and the random assignment of subjects. (Two recent, often-cited
studies, Walsh, et al., 1991,xxiii
and Project MATCH,xxiv
did not have control groups, and Project MATCH was not even a direct study
of AA.) Both controlled studies indicated that AA is not an effective
across-the-board treatment for alcohol abuse or dependence ("alcoholism").
The subjects in both studies were, however, court-referred alcoholic offenders
and hence different from the general alcoholic population in certain respects.
Thus one distinguishing feature of the study populations is that they did
not voluntarily seek treatment; they were forced to attend AA.
the surface, these factors—the employment of coercion and the special-population
status of alcoholic offenders—seem to lessen the credibility of the
two controlled studies of AA's effectiveness. But it could be argued that
one factor is irrelevant and the other actually enhances the studies' credibility.
If, as is commonly asserted, AA is a universally applicable treatment
for all alcoholics, the makeup of the study populations shouldn't
have mattered a whit as long as the assignment of subjects to AA and control
groups was truly random. And the fact that the studies' subjects were coerced
into participating could well
increase the validity of the studies'
findings, because a very important biasing factor, subject motivation,
was eliminated, and the remaining biasing factors were spread out fairly
evenly among the groups studied because of the random assignment procedure.
Further, since at least a third of present-day AA participants are coerced
into attendance either by alcoholism treatment programs or the courts,
through programs for DUI and other alcohol-related offenders, the populations
of these studies were perhaps not as different from the general AA population
as one might suspect.
first of these controlled studies of AA's effectiveness was conducted in
San Diego in the mid-1960s.xxv
In the study, 301 public drunkenness offenders were randomly divided into
three groups. One group was assigned to attend AA, another to attend an
alcoholism treatment clinic, and a third group, the control, was not assigned
to any treatment program. All of the study's subjects were followed for
at least one full year following conviction. Results were calculated by
counting the number and frequency of rearrests for drunkenness. Surprisingly,
the no-treatment control group was the most successful of the three, with
44% of its members having no rearrests; 32% of those assigned to the clinic
group had no rearrests; and 31% of those assigned to AA had no rearrests.
As well, 37% of the members of the control group had two or more rearrests,
while 40% of the alcoholism clinic attendees were arrested at least two
times, and 47% of the AA attendees were arrested at least twice. While
far from a definitive debunking of AA's alleged effectiveness, these results
are certainly suggestive.
other controlled study of AA's effectiveness was very carefully designed
and conducted, and was carried out in Kentucky in the mid-1970s.xxvi
A large majority of its subjects were obtained via the court system, and
seemed to be "representative of the 'revolving door' alcoholic court cases
in our cities." The investigators divided 197 subjects into five randomly
selected groups: a control group given no treatment; a group assigned to
traditional insight therapy administered by professionals; a group assigned
to nonprofessionally led Rational Behavior Therapy (lay RBT); a group assigned
to professionally led Rational Behavior Therapy; and a group assigned to
AA. Length of treatment varied from 202 to 246 days, and subjects were
evaluated at the end of treatment and also at three months and 12 months
following its termination.
general, the groups given professional treatment did better than the nonprofessionally
treated groups and the control group. A significant find-ing, however,
was that treatment of any kind was preferable to no treatment.
a great many alcohol abusers never seek professional treatment, it's particularly
important to compare the results of the AA, lay RBT, and control groups.
Lay RBT was clearly superior to AA in terms of dropout rate. During the
study, 68.4% of those assigned to AA stopped attending it, while only 40%
of those attending lay RBT sessions stopped attending them. Further, at
the termination of treatment, all of the lay RBT participants who had persisted
in treatment reported that they were drinking less than they were before
treatment, while only two-thirds of those who had continued to attend AA
reported decreased drinking. As well, during the final three months of
treatment, the mean number of arrests was 1.24 for the lay RBT group, 1.67
for the AA group, and 1.79 for the control group.
most interestingly, the number of reported binges at three months after
termination of treatment was far higher for the AA group than for the lay
RBT or control groups. The mean number of reported binges by the AA attendees
was 2.37 over the previous three months, while the mean number reported
by the controls was 0.56, and the mean for the lay RBT group was only 0.26.
This finding strongly suggests that the AA attendees had accepted AA's
"one drink, one drunk" dogma, and had then proceeded to "prove" it. It's
pertinent to note, however, that at 12 months following the termination
of treatment there were no significant differences between the AA, lay
RBT, and control groups. One possible interpretation of this finding is
that the positive effects of Rational Behavior Therapy fade with time in
the absence of continued practice, and that the harmful effects of exposure
to AA (at least in regard to bingeing) also fade with time in the absence
of further exposure to AA.
particularly intriguing aspect of this study is that the relatively successful
(compared with AA and the no-treatment controls) lay RBT group utilized
a treatment based on Rational Emotive Behavior Therapy (REBT). The reason
that this is interesting is that S.M.A.R.T. Recovery, the newest of the
national secular alternatives to AA, is based on REBT. Thus, there's at
least some slight reason to think that SMART might be more effective than
AA. Unfortunately, no controlled studies of SMART's effectiveness have
yet appeared, though one such study is now under way in Tucson, Arizona.
But for now, speculation about SMART's effectiveness will remain just that—speculation.
there's a crying need for additional controlled studies of AA's effectiveness,
as well as for controlled studies of SMART's effectiveness and that of
the other secular groups. In the absence of such studies (at least as regards
the alternative groups) all that we're left with is educated guesswork.
is the situation hopeless? No. In fact, there's considerable data available
that indicates which approaches are effective and which ones aren't. William
Miller and Reid Hester, editors of the most comprehensive and most methodologically
sound evaluation of treatment methods ever published, state that, "We were
pleased to see that a number of treatment methods were consistently supported
by controlled scientific research."xxvii
But they continue, "On the other hand, we were dismayed to realize that
virtually none of these treatment methods was in common use within alcohol
treatment programs in the United States."xxviii
Worse, "A significant negative correlation (r=-.385) was found between
the strength of efficacy evidence for modalities and their cost; that is,
the more expensive the treatment method, the less the scientific evidence
documenting its efficacy."xxix
They list the treatment methods showing the most positive results, as shown
by controlled studies, as brief intervention, social skills training, motivational
enhancement, community reinforcement approach, and behavior contracting.xxx
Importantly, 12-step treatment was nowhere in evidence in the list of effective
treatments; but it was quite likely a component of four modalities
for which a number of studies show significant negative results: unspecified
"standard" treatment; confrontational counseling; milieu therapy;
and general alcoholism counseling.xxxi
As for AA, Miller and Hester list only the two controlled studies discussed
above, both of which showed negative results.
What Doesn't—12-Step Treatment
have been many studies of 12-step treatment, but the vast majority are
of little use in determining treatment effectiveness for two reasons: 1)
they lacked control groups; and 2) they were short- or medium-short-term
studies. It's impossible to draw meaningful conclusions about treatment's
effectiveness without control groups. And any apparent benefits from treat-ment
tend to disappear with time. Thus, long-term studies utilizing control
groups are necessary to determining the effectiveness of treatment. But
there have been relatively few.
important study was published in 1983. For eight years, its author, George
Vaillant, followed 100 patients who had undergone 12-step treatment;
he compared this sample to several hundred other untreated alcohol abusers.
The treated patients fared no better than the untreated group. Fully 95%
of the treated patients relapsed at some time during the eight years Vaillant
followed them, and he concluded that "there is compelling evidence that
the results of our treatment were no better than the natural history of
He added, "Not only had we failed to alter the natural history of alcoholism,
but our death rate of three percent a year was appalling."xxxiii
very important, very scientifically sound study appeared in 1996. The National
Longitudinal Alcoholism Epidemiological Survey was designed and sponsored
by the NIAAA and was conducted by the U.S. Bureau of the Census. It was
notable both for its size (4,585 subjects) and its study period (20 years).
Its subjects were divided into a treated group and an untreated group.
All of the study's subjects "had to have satisfied the criteria for prior-to-past
year DSM-IV alcohol dependence by meeting at least 3 of the 7 DSM-IV criteria
for dependence: tolerance; withdrawal (including relief or avoidance of
withdrawal); persistent desire or unsuccessful attempts to cut down on
or stop drinking; much time spent drinking, obtaining alcohol or recovering
from its effects; reduction or cessation of important activities in favor
of drinking; impaired control over drinking; and continued use despite
physical or psychological problems caused by drinking."xxxiv
study's findings were surprising: At 20 years after onset of symptoms,
80% of those who had undergone treatment were either abstinent or "drinking
without abuse or dependence." But those who had never undergone treatment
were doing even better: 90% of them were either abstinent or drinking nonproblematically.
That is, 10% of those who had never been treated were still drinking abusively
20 years after the onset of symptoms, as were 20% of those who had been
treated. In other words, twice as many of those who had undergone
treatment were drinking abusively as those who had never been treated.xxxv
important findings included the following: of those who had never been
treated, fully 60% were drinking nonproblematically 20 years after the
onset of symptoms; and even of those who had undergone treatment and had
received dire warnings of loss of control, incurability, and progressivity,
28% were drinking nonproblematically 20 years after the onset of symptoms.xxxvi
As well, in both groups, the percentage of those recovered (abstinent or
drinking nonproblematically) steadily increased with time.xxxvii
who had been treated reported more initial problems than the untreated
group, but as anyone who has ever gone to a few AA meetings can attest,
there's status in reporting horrendous drinking behavior: the worse the
problems overcome, the more impressive the apparent recovery. And in treatment
centers, patients are routinely encouraged to exaggerate their problems.
Stanton Peele quotes major league pitcher Dwight Gooden on his experiences
being browbeaten in 12-step rehab by fellow patients: "My stories weren't
as good [as theirs] . . . They said, 'C'mon, man you're lying.' They didn't
believe me . . ."xxxviii
There are many similar anecdotal reports. Given such rewards and pressures,
it would be surprising indeed if many persons who have undergone treatment
didn't "come to believe" their own stories and would thus over-report previous
symptoms. Indeed, there's evidence that only treated individuals display
all of the classic symptoms of alcoholism.xxxix
As Stanton Peele puts it, "Treatment here seems to be necessary for the
development of the classical alcoholism syndrome."xl
way of gauging treatment effectiveness is through recidivism rates. By
all indications, they're sky high for 12-step treatment. The NCADD reports
that in 1992, "nearly 13.8 million Americans had problems with drinking";
and it claims that in that same year 1.9 million Americans underwent treatment.xli
These are astounding figures. If these figures are accurate and are typical
of preceding and subsequent years, and assuming that "treatment works"
and has a 0% recidivism rate, every single alcohol abuser in the United
States should have been treated in an eight-year period (accounting for
mortality, population increase, and the development of new problems), and
well over 90% of the nation's alcoholism treatment centers should have
shut down by now for lack of clients. Obviously, this hasn't happened.
Even if treatment were only 50% effective and there were a 50% recidivism
rate, the number of alcohol abusers in this country should have dropped
by at least a third over the last decade. Again, obviously, this hasn't
happened. Instead, the literally billions of dollars spent on treating
the nearly two million people per year reported by the NCADD have had seemingly
no effect whatsoever on the number of alcohol abusers in the United
States. This is good evidence that the treatment industry's drumbeat chant,
"Treatment Works!," is an outright lie, and it's also good evidence that
the recidivism rate in 12-step treatment is astronomical. Reports from
those who have undergone treatment indicate that this is so. One report
from a pro-AA participant/observer in a 28-day program states that of the
42 patients who made it through treatment, "Twenty were in treatment for
the first time, 15 were in treatment for the second time, and seven patients
had had at least two previous admissions."xlii
As well, the "National Treatment Center Study Summary Report" states that
the recidivism rate at the overwhelmingly privately owned inpatient facilities
it surveyed is fully 40%.xliii
One suspects that the rate at publicly funded facilities is even higher.
Anecdotal evidence indicates that this is so: a report written by a client
who had undergone treatment in a VA hospital states that 11 of 12 patients
in his 28-day program had previously undergone inpatient treatment, and
that one of the recidivists had been in treatment 19 times.xliv
Misinterpretation of Data
often hears claims that one or another new study has "proven" the efficacy
of AA or of 12-step treatment. Invariably, these claims are made about
studies that didn't include control groups and that often had other methodological
problems as well. The most obvious recent example is Project MATCH.xlv
Treatment industry spokesmen claimed variously that it demonstrated the
validity of 12-step treatment; that it demonstrated that all of the tested
forms of treatment work equally well; and some even claimed that it demonstrated
that 12-step treatment was superior to other forms of treatment. In reality,
it did none of these things.
MATCH was an incredibly expensive study ($27 million—some reports have
placed the total at $35 million) funded by the NIAAA that compared three
forms of outpatient and aftercare treatment: motivational enhancement;
cognitive behavioral coping skills therapy; and 12-step facilitation
therapy. All three forms of treatment were delivered in one-on-one counseling
sessions, though the number of scheduled sessions was only four for motivational
enhancement, compared with 12 for the other two forms of treatment. The
significant findings of Project MATCH were that patients in all three groups
experienced very similar improvement, as measured by the number of drinking
days per month and the incidence of bingeing, and that 12-step-treated
clients with less severe psychological problems had more abstinent days
than similar clients in the other two groups. The number of days that outpatient
clients drank fell from 22 to 6, while the number of days aftercare clients
drank fell from 25 to 4; and the number of drinks per drinking day fell
from 12 to 3 among outpatient clients, and from 18 to 2 among aftercare
Clearly, these are dramatic results, but were they the result of treatment,
or of other factors?
MATCH was so over-designed that it seems likely that any form of
treatment used in it would have shown similar results. First, all clients
in Project MATCH were volunteers, and their volunteer status in itself
shows a fairly high degree of motivation, a very important biasing factor.
This is in stark contrast to the coerced status and presumably low motivation
of the average alcoholism treatment client. Second, during the screening
process approximately 10% of potential clients opted out of the study for
reasons such as "the inconvenient location of the study or transportation
This helped to narrow the study to only the most motivated clients. Third,
more than half of the remaining potential clients were eliminated from
the study for reasons such as "failure to complete the assessment battery"
and "residential instability."xlviii
This not only helped to ensure that only the most motivated clients would
participate in the study, but also that only the most well-adjusted clients
would participate, thus introducing another positive biasing factor: social
and emotional stability.
expectations were still another positive biasing factor. The clients knew
that they were taking part in an expensive study, and the manner in which
sessions were conducted undoubtedly led to high expectations: every session
consisted of one-on-one therapy with a competent professional; and every
session was videotaped. As well, the study's conductors engaged in "compliance
enhancement procedures (i.e., calling clients between sessions, sending
reminder notes and having collateral contacts),"xlix
which certainly must have helped drive home the study's apparent importance,
thus reinforcing clients' positive expectations.
another problem with Project MATCH is that the treatments employed
were of universally high quality—not only were all sessions videotaped,
but supervisors monitored fully a quarter of them; they bore little resemblance
to commonly employed treatments. The difference between Project MATCH and
real-world treatments was probably most pronounced in the study's 12-step
facilitation therapy. It seems highly unlikely that the therapists employing
12-step facilitation in Project MATCH would have engaged in the abusive
behavior that is routine in 12-step treatment. It seems very unlikely that
they would have bullied clients in order to coerce false confessions from
them; it seems equally unlikely that they would have ridiculed clients
who questioned the therapy or made comments critical of it or of AA; and
one doubts that they would have lied to clients about "inevitable" loss
of control after one drink, or about the "inevitable" progressivity
of their incurable "disease."
all of these biasing factors, it was hardly surprising that all forms of
treatment showed remarkably similar positive outcomes; almost any
form of treatment would probably have shown similar results. The study's
conductors even recognized—at least after the fact—this possibility.
They note: "Compliance enhancement procedures . . . and the greater attention
of individual treatment may have produced a level of overall compliance
that made it difficult for differences between treatments to emerge."l
They continue, "The overall effect of being a part of Project MATCH, with
its extensive assessment, attractive treatments and aggressive follow-up,
may have minimized naturally occurring variability among treatment modalities
and may, in part, account for the favorable treatment outcomes."li
never know for sure, though, because, as the researchers put it, "the efficacy
of the three treatments cannot be demonstrated directly since the trial
did not include a no-treatment control group."lii
But there does seem to be one clear lesson in Project MATCH: if you introduce
enough positive biasing factors, almost any form of treatment will "produce"
a positive outcome.
What We Know
now, the best evidence available suggests that AA is ineffective as a means
of overcoming alcohol problems, and there's some evidence that exposure
to AA worsens at least one significant abusive behavior—binge drinking.
But the evidence is not conclusive, and until additional controlled studies
are conducted, it will remain impossible to draw firm conclusions about
AA's (in)effectiveness. One thing, however, bears repeating: there's
no good evidence to indicate that AA is any more effective than "spontaneous
recovery." Assertions that AA is an effective means of overcoming alcohol
problems, let alone assertions that AA is the most or the only
effective means of doing so, are just that—assertions, and groundless ones
for 12-step treatment programs, we'll deal with them further in the next
, by Bill Wilson. New York: Alcoholics Anonymous
World Services, 1976, pp. 20-21.
"NIAAA Releases New Estimates of Alcohol Abuse and Dependence," March 17,
"How Many People Are Alcoholic?" page on the Addiction Research Foundation's
web site: http://www.arf.org/isd/stats/alcohol.html
I base this estimate on AA's 1996 membership brochure. Because of the limitations
of the data supplied by AA, my conclusions here must be somewhat tentative.
I arrived at my figures as follows: 16% of those attending were openly
coerced by the courts or penal system. I started with this as a baseline
figure, because it involves undisguised coercion. Adding all of the percentages
listed of other "important factors," one arrives at a total of 241%. To
arrive at the coercive total percentage, I added the full 40% listed for
treatment facilities (clients are almost invariably coerced into AA attendance
by treatment facilities), three quarters of 16% listed for counseling agencies
(counseling agencies often make counseling contingent on AA attendance),
the full 9% listed for "employer or fellow worker" (undoubtedly, almost
all of them were coerced into treatment by EAPs or professional diversion
programs), 7% out of the 39% listed for family (the "National Treatment
Center Study Summary Report" indicates that 17.5% of inpatient clients
are adolescents, who would not enter treatment voluntarily), and half of
the 8% listed for health care providers (who sometimes make treatment contingent
on AA attendance). This yields a total of 65%, which I divided by 2.41,
which yields a figure of roughly 27%. Adding that 27% to the 16% who were
outright coerced by the legal and penal systems yields a total of 43% of
current AA members who belong to the organization primarily as a result
of some type of coercion. Of course, this method is inexact, but it does
yield a reasonable ballpark figure.
"Evaluation of Treatment Methods in Chronic Alcoholism," by Frederick Baekeland,
in Treatment and Rehabilitation of the Chronic Alcoholic
Kissin and Henri Begleiter, eds. New York: Plenum Press, 1977, p. 392.
"Membership" page on AA's web site:
"The growth and effectiveness of Alcoholics Anonymous in a Southwestern
City," by Bill. C. Quarterly Journal of Studies on Alcohol
National Admissions to Substance Abuse Treatment Services: The Treatment
Episode Data Set (TEDS) 1992-1995.
Rockville, Maryland: U.S. Department
of Health and Human Services, 1997. Table 10, p. 46.
"Alcoholics Anonymous as a Social Movement," by Robin Room, in Research
on Alcoholics Anonymous
, Barbara McCrady and William Miller, eds. New
Brunswick, New Jersey: Rutgers Center of Alcohol Studies, 1993, p. 169.
If it were correct, and if it were even roughly comparable to rates for
previous years, the full 9% of the population who claim to have attended
at least one AA meeting would likely have done so during the previous three
years, given minimal attendance overlap. And even if there were a 50% overlap,
the full 9% figure would be reached in just over five years. These time
periods are simply too short for both the 9% figure and the 3.4% figure
to be correct. Further, given that AA grew in the U.S. and Canada by approximately
50,000 members in the year cited (1990), if 3.4% of the adult population
had attended an AA meeting in that year, that would have come to 7,000,000
people, giving AA a new-member dropout rate well in excess of 99% for that
year. AA has a very busy "revolving door," but it doesn't revolve quite
that only 42% of the 3.4% who claim to have attended an AA meeting in 1990
admitted to doing so because of an alcohol problem. This simply doesn't
wash with experience. From my own observations and those of other sharp-eyed
former members (including Vince Fox and Ken Ragge), curiosity seekers are
not terribly common at AA meetings, and neither are family members of alcohol
abusers (though coerced persons, many of whom shouldn't be there, make
up a significant percentage of attendees). My best estimate is that well
over 90% of newcomers show up at AA either because of their own alcohol
problems or because of coercion. It should also be remembered that even
in anonymous surveys many individuals lie about things that they consider
embarrassing, which would help to explain the low percentage admitting
to attendance because of their own problems.
"Comments on A.A.'s Triennial Surveys," no author listed. New York: Alcoholics
Anonymous World Services, n.d. (probably 1990).
Ibid., p. 12, Figure C-1.
Telephone conversation with Fox in August 1997.
Drug and Alcoholism Treatment Unit Survey (NDATUS): Data for 1994 and 1980-1994."
Rockville, Maryland: SAMHSA, 1996, Table 10.
The minimum success rate I calculated in the original edition of this book
was 1.3%. Most of the difference between the two figures is due to AA's
reported increase in five or more years' sobriety among its members from
29% in 1989 to 45% in 1996. Most of the rest is a result of AA's having
grown faster than the rate of population growth.
"Recovery Without Treatment," by Thomas Prugh. Alcohol Health and Research
, Fall 1986, pp. 24, 71 and 72.
"Alcoholism as a Self-Limiting Disease," by Leslie R.H. Drew. Quarterly
Journal of Studies on Alcohol,
Vol. 29, 1968, pp. 956-967.
"Spontaneous Remission in Alcoholics: Empirical Observations and Theoretical
Implications," by Barry S. Tuchfeld. Journal of Studies on Alcohol
Vol. 42, No. 7, 1981, pp. 626-641.
"Spontaneous Recovery in Alcoholics: A Review and Analysis of the Available
Research," by R.G. Smart. Drug and Alcohol Dependence
, Vol. 1, 1975-1976,
"Correlates of Past-Year Status Among Treated and Untreated Persons with
Former Alcohol Dependence: United States, 1992," by Deborah A. Dawson.
Alcoholism: Clinical and Experimental Research
, Vol. 20, No. 4,
June 1996, p. 773.
"Is Alcoholism Treatment Effective?," by Helen Annis. Science
236, April 3, 1987, p. 21.
Baekeland, op. cit., p. 407.
"A Randomized Trial of Treatment Options for Alcohol-Abusing Workers,"
by Diana C. Walsh, et al. New England Journal of Medicine
325, No. 11, Sept. 12, 1991, pp. 775-781. This study is sometimes cited
as having a control group, when in fact it did not. A true control group
would have received no treatment, whereas over 85% of the "choice" group
in this study chose either hospitalization or AA attendance, thus rendering
the results of the study ungeneralizable. Walsh et al. measured AA attendance
only versus coerced hospitalization with AA attendance, and versus
a mongrel "choice" group that voluntarily chose in large part either AA
alone or hospitalization featuring AA. Thus the data in this study is of
very limited use and cannot be cited (at least honestly cited) as evidence
that AA or 12-step hospitalization "works" or doesn't "work," except in
relation to each other. And even that comparison is of limited value given
that this was a short-term study with a relatively small sample.
"Matching Alcoholism Treatments to Client Heterogeneity: Project MATCH
Posttreatment Drinking Outcomes," by Project MATCH Research Group. Journal
of Studies on Alcohol
, January 1997, pp. 7-29. This amazingly expensive
($27 million) study not only did not have a control group, but did not
directly measure AA's effectiveness vis a vis other treatments, and its
methods of client selection and client handling probably served to distort
the outcome. As the Project MATCH researchers themselves pointed out, "The
overall effect of being a part of Project MATCH, with its extensive assessment,
attractive treatments and aggressive follow-up [clients were paid, among
other things] may have minimized naturally occurring variability among
treatment modalities and may, in part, account for the favorable treatment
outcomes." (p. 24)
"A Controlled Experiment on the Use of Court Probation for Drunk Arrests,"
by Keith S. Ditman, George G. Crawford, Edward W. Forby, Herbert Moskowitz,
and Craig MacAndrew. American Journal of Psychiatry
, 124:2, August
1967, pp. 160-163.
Outpatient Treatment of Alcoholism
, by Jeffrey Brandsma, Maxie Maultsby,
and Richard J. Walsh. Baltimore: University Park Press, 1980.
Handbook of Alcoholism Treatment Approaches: Effective Alternatives
William Miller and Reid Hester, editors. Boston: Allyn and Bacon, 1995,
Ibid., Chapter 2, "What Works?," by Miller, Hester, et al., p. 13.
Ibid., Table 2.4, p. 18.
The Natural History of Alcoholism: Causes, Patterns, and Paths to Recovery
by George Vaillant. Cambridge, MA: Harvard University Press, 1983, p. 284,
"Correlates of Past-Year Status Among Treated and Untreated Persons with
Former Alcohol Dependence: United States, 1992," by Deborah A. Dawson.
Alcoholism: Clinical and Experimental Research
Vol. 20, No. 4, June 1996, p. 772.
Ibid., Table 1, p. 773.
"AA Abuse," by Stanton Peele. Reason
, November 1991, pp. 34-39.
Reproduced at http://www.frw.uva.nl/cedro/peele/lib/aaabuse.html (p. 3
of html document).
See "Treatment Seeking Populations and Larger Realities," by Robin Room,
in Alcoholism Treatment in Transition,
G. Edwards and M. Grant, eds. London: Croom Helm, 1980, pp. 205-224.
"Denial—of Reality and of Freedom—in Addiction Research and Treatment,"
by Stanton Peele. Bulletin of the Society of Psychologists in Addictive
5 No. 4, 1986, pp. 149-166. The document is reproduced at http://www.frw.uva.nl/cedro/peele/library/denial.html
and the quoted
text is taken from page 9 of the html document.
"Alcoholism and Alcohol-Related Problems: A Sobering Look."
(pp. 1 & 2).
"Goal Setting and Recovery from Alcoholism," by Donna Marie Wing. Archives
of Psychiatric Nursing
4, No. 3, June 1991, p. 179.
"National Treatment Center Study Summary Report," Paul Roman and Terry
Blum, principal investigators. Athens, Georgia: Institute for Behavioral
Research, 1997, p. 17.
"Twenty-Eight Days in Wilson's Inferno," by Brian Barton. Journal of
, Vol. 9, No. 5, May-June 1997, p. 8.
"Matching Alcoholism Treatments to Client Heterogeneity; Project MATCH
Posttreatment Drinking Outcomes," by Project MATCH Research Group. Journal
of Studies on Alcohol
1997, pp. 7-29.
Ibid., Figure 1, p. 15. These figures aren't exact, as there were minor
variations in the outcomes for the three forms of treatment, and I derived
these figures by interpreting graphs of results.