Resisting 12-Step Coercion

The Efficacy of 12-Step Groups and 12-Step Treatment

This book is here courtesy of See Sharp Press logo and of its authors.
The Efficacy of 12-Step Groups and 12-Step Treatment
by Charles Bufe and Stanton Peele

The massive American treatment enterprise is based on the belief that alcohol and drug abuse can be treated away, and that we know how and are currently able to do so. Despite continuous efforts to establish its validity, this fundamental idea has not been shown to be true. Instead, research has repeatedly questioned whether the standard elements of U.S. substance abuse treatment work, and indeed, sometimes indicates that they do more harm than good. Consider that, after several expensive treatment experiences and many vows of abstinence, professional golfer John Daly resumed drinking and gambling. What had he learned in treatment? "It's sad, but I think it's great to be free," Daly told Golf World magazine. "Granted, I could go out and lose everything [by] gambling and drinking, but there's no sense in denying it. It's in my blood." Daly, like many, got the message that he had a disease for which he was not personally responsible.

This chapter will review the extensive research on substance abuse treatment. This research shows: 1) no clinical study has found the sine qua non of alcoholism treatment in the U.S., Alcoholics Anonymous, to be effective; 2) the most effective alcohol and drug treatments, as indicated by clinical trials, are rejected by U.S. treatment providers; 3) when standard substance abuse treatments have been compared to no or minimal treatment, they produce results no better than no or minimal treatment; 4) the U. S. government conducted the most expensive trial of psychotherapy ever done and found that minimal treatment, which places responsibility for allaying alcoholism on the drinker, is as good as the most elaborate 12-step treatment and cognitive/skills training; 5) a massive federal survey of American drinkers found that most alcoholics in the U.S. do not enter treatment, a higher percentage of untreated than treated alcoholics are in remission, and most untreated and treated alcoholics continue to drink, only the untreated do so with better results; 6) coerced treatment of prison inmates and probationers is actually counterproductive; 7) finally, there is no replacement for individual motivation and the provision of sufficient resources for those who want to improve their lives and cease addiction and substance abuse.

The Efficacy of AA and NA

There have been a number of controlled studies of AA and 12-step treatment, but there have been no scientifically valid, generalizable studies of NA. That is, there have been no studies of NA featuring random assignment of subjects and no-treatment control groups, or even comparison groups given other treatments, nor have there been longitudinal studies with matched comparison groups. There have been a number of uncontrolled, ungeneralizable (by their very nature) studies of 12-step groups (especially AA); and 12-step partisans love to quote such studies (and anecdotal evidence) as if they demonstrate the efficacy of 12-step programs.1 The reason that the results of such uncontrolled studies cannot be generalized was put well by Holder et al. in an important 1991 article in the Journal of Studies on Alcohol:

Prior reviews of the alcoholism treatment outcome literature have suggested that evidence from controlled clinical trials is considerably more consistent than the cumulative evidence of uncontrolled case studies and group designs. Positive uncontrolled reports can be found for virtually every treatment that has been tried for alcoholism, including psychosurgery, respiratory paralysis and the administration of LSD. . . . Positive or negative outcomes may be attributable not only to the treatment offered, but to a host of confounding factors including patient selection criteria, expectancies, additional treatment components and posttreatment factors. Uncontrolled trials also offer no basis for comparison of outcomes. Is a 40% success rate a triumph or a disgrace compared with what would be expected from no treatment or alternative treatments? . . . Controlled trials in general, and randomized clinical trials in particular, are commonly employed as the standard of evidence for specific effectiveness of medical treatments. (Holder et al., 1991, p. 52)

NA's Effectiveness

So, ignoring the ungeneralizable, uncontrolled studies cited by AA's and NA's supporters, and the self-serving, gross overestimates of treatment effectiveness by treatment providers, what are we left with? As regards NA, nothing. There is no scientifically valid evidence of NA's effectiveness. To put this another way, those being coerced into NA are being coerced into a program for which there is no scientific evidence of efficacy.

NA, however, is merely a clone of AA, as can be demonstrated by comparing the heart of both programs, the 12-steps of AA and NA, which differ by only single terms in steps 1 and 12. As well, NA borrowed its organizational structure and its meeting structure directly from AA. Further confirmation of the essentially identical nature of AA and NA can be seen in the integration of drug and alcohol 12-step treatment. The National Treatment Center Study Summary Report states that, "Only 11 of the 450 programs (2.4%) offer segregated treatment programming for patients with alcohol problems and patients with drug problems" (Roman & Blum, p. 26). Thus, NA can legitimately be regarded as a clone of AA, and one can reasonably expect that the efficacy of NA will be approximately that of AA.

AA's Effectiveness

As for AA, there are scientifically valid studies, but not many. Two, to be exact and they are of particular interest to the purposes of this book because they both involved coerced clients. Because of their importance, we'll consider them at some length.

The Ditman Study

The first of these studies was conducted in San Diego in 1964 and 1965, and its subjects were 301 "chronic drunk offenders," who were defined "as having two drunk arrests in the previous three months or three drunk arrests in the previous year" (Ditman et al., 1967, p. 64). These offenders were randomly assigned to a no-treatment control group, a group assigned to go to AA as a condition of probation, and a group assigned to clinic treatment (type not specified) as a condition of probation. Those assigned to AA needed to provide proof of attendance in the form of signed statements from AA meeting secretaries attesting to their presence. All of these individuals were followed for at least a full year after their convictions. The primary outcome measure the investigators used was the number of rearrests during the year following conviction. The results were that 69% of those assigned to Alcoholics Anonymous were rearrested; 68% of those assigned to clinic treatment were rearrested; and 56% of the notreatment control group was rearrested. Although this may sound striking, the investigators noted, "the chi-square tests . . . failed to reach statistical significance" (Ditman et al., 1967, p. 66). Their study summary reads, "No statistically significant differences between the three groups were discovered in recidivism rate, in number of subsequent rearrests, or in time elapsed prior to rearrest" (p. 64). To put this another way, this study provided evidence that AA is no more effective than no treatment at all.

The Brandsma Study

The second study, published in book form under the title Outpatient Treatment of Alcoholism (Brandsma et al., 1980), was a carefully designed, sophisticated, NIAAA-funded study of AA and three alternative therapies: lay-led Rational Behavior Therapy (similar to today's SMART Recovery program); professionally conducted one-on-one Rational Behavior Therapy (today called Rational Emotive Behavior Therapy); and professionally conducted one-on-one, traditional (Freudian-based) insight therapy. This study, like the Ditman et al. study, featured random assignment of subjects and a no-treatment control group. The researchers indicate that there were very few differences between the various groups, and that "the random assignment procedure was successful" (p. 70).

After screening, 260 clients were accepted for the study; 184 were court-coerced and were participating in the study as a condition of probation, and 76 were self-referred. The authors characterized the participants "as representative of the 'revolving door' alcoholic court cases in our cities" (p. 63).

Upon entering the study, the subjects were randomly assigned to one of the four treatment groups, or to a no-treatment control group. Treatment was scheduled to last up to a year, and the subjects were monitored during that time, with an outcome assessment after they completed treatment. They were then reinterviewed 3, 6, 9, and 12 months after the outcome assessment. The authors used several different outcome measures, including number of drinking days per 90 day period; amount consumed per drinking day; number of binges; marital status; employment status; rearrests; other social and psychological factors; and retention rate for the various treatments.

The results of this study were significant. In terms of retention rate, AA fared by far the worst of any of the treatment groups. The group assigned to AA had a 68% dropout rate; the insight group had a 42% dropout rate; lay-RBT had a 40% dropout rate; and pro-RBT had a 46% dropout rate.

As for the number of rearrests for drunkenness during treatment, all treatment groups did significantly better than the control group, with the professionally led insight group performing better than the other treatment groups.

As for economic results, in terms of employment, AA fared the worst of all groups; and in regard to employment-seeking, AA and the control group did worst, while the insight group did the best.

In terms of drinking behavior, all treatment groups reported significantly decreased drinking at the outcome assessment: pro-RBT, 80% of participants; lay-RBT, 100%; insight, 92%; AA, 67%; whereas only 50% of the control group reported decreased drinking. And, "at outcome, there were no significant differences in drinking behavior between [the lay-RBT, AA, and control] groups with regard to the number reporting abstinence" (p. 104).

In regard to binging behavior, "The mean number of reported binges was significantly greater (p = .004) for the AA group (2.37 in past 3 months) in contrast to both the control (0.56) and lay-RBT group (0.26). In this analysis, AA was [over 4] times [more] likely to binge than the control [group] and nine times more likely than the lay-RBT [group]. The AA average was 2.4 binges in the last 3 months since outcome" (p. 105).

All of this led the study's authors to conclude:

Although AA plays an extremely important part [in] a community's response to alcoholism in terms of education and treatment, it seems to have a definite but delimited "place in the sun." . . . AA seems to have definite limitations of social class, ideology, flexibility of adopting new techniques, and the type of personality it appeals to. Our study suggests further confirmation of this in our severe dropout rate from this form of treatment. It is probable, as Ditman et al.'s (1967) work suggests and ours confirms, that AA is just not effective as a coerced treatment with municipal court offenders. (Brandsma et al., 1980, p. 84, emphasis added)

Finally, it's important to point out that the increase in binging behavior among those exposed to AA in this study militates against coercing DUI offenders into AA attendance. One very possible reason for the increase in binging is the emphasis in AA upon inevitable loss of control after even one drink, as codified in the AA slogan, "one drink, one drunk." (As we saw in Chapter 1, this assertion is not true, except to the extent that drinkers believe it to be true.) What likely happens is that for those exposed to AA, this inevitable-loss-of control belief becomes a self-fulfilling prophecy. So, when a true-believing AA member slips and has a drink, or even eats a rum ball or ingests a bit of mouthwash containing alcohol, he or she could be provoked to embark on a full-bore binge. Given this, one can't help but be alarmed at the common practice of coercing DUI defendants into AA attendance and 12-step treatment.

AA's Triennial Surveys

While these are the only two controlled studies with random assignment of subjects ever conducted on AA's effectiveness, there is one other source of information worthy of our attention: the AA monograph, Comments on AA's Triennial2 Surveys. AA has surveyed its members every three years since 1977 (with the exception of a four-year gap between 1992 and 1996). These are large surveys of several thousand AA members, and they measure such variables as length of membership, age distribution, male-female ratio, employment categories, the ways in which members were introduced to AA, and length of abstinence. AA normally uses the results in its promotional "Alcoholics Anonymous [insert year] Membership Survey" brochures, but following the 1989 survey, AA undertook a statistical analysis of the previous five surveys. The results were published in the Comments on A.A.'s Triennial Surveys monograph.

As regards new member dropout rate, all five surveys were in close agreement. Averaging their results, the Comments document graphs the "% of those coming to AA within the first year that have remained the indicated number of months." At one month, the "% of those . . . that have remained" is 19%; at 3 months, 10%; and at 12 months, 5% (Alcoholics Anonymous, n.d., p. 12, Figure C-1).3 This gives AA a 5% success rate at the one-year point if success is simply defined as continuing AA membership (in what is supposed to be a program for life).

But even this 5% success rate is questionable if AA's success in dealing with alcohol problems is defined as the very modest level of one-year's continuous abstinence, because, as anyone who has spent much time in AA can attest, far from all AA members are abstinent. AA's self-reported rate of recovery (as derived from membership retention) is far from impressive; in fact, it appears to be no better than the rate of spontaneous remission, which has been estimated at anywhere from 1% to 33% per year of those with alcohol problems (Prugh, 1986, p. 24). One survey of the spontaneous remission literature estimates its prevalence at 3.7% to 7.4% per year (Smart, 1975/76, p. 284). If this is true, AA's recovery rate of 5% or less could well be lower than the rate of spontaneous remission.4

This is most curious when one considers the nature of the disease of alcoholism (in reality, a behavioral problem). Given such a disease, one would expect a powerful placebo effect (from participation in AA), which in itself would substantially raise the rate of recovery in AA above that of spontaneous remission. But the placebo effect, judging from the available evidence, simply doesn't exist here. There seem two likely explanations for this: 1) There are so many coerced persons participating in AA, who very often dislike and actively resist AA's program, that AA's success (and concomitant membership retention) rate is skewed sharply downward; 2) AA is actively harmful. This is definitely possible, given AA's "powerless" and "one drink, one drunk" dogmas; and the evidence on binge drinking supplied by the Brandsma et al. study certainly seems to suggest this interpretation.

The authors of AA's Comments document, remarking on the low AA retention rate revealed by their research, note: "Individuals [in AA] may rebel against this result as contradicting our time-honored statement that 'half get sober right away, another 25% eventually make it,' etc. That statement applied to observations made at an earlier time, and there is no reason to doubt that changes in society and in A.A. since that time could create a different circumstance today. Like other findings of the survey, this may be a challenge to the membership to 'change the things we can'" (p. 13).

One change that some AA members5 could make, if they wanted to, would be to stop participating in the mass coercion of individuals into AA. While that certainly would not immediately end coercion into AA, it would likely cause a noticeable decrease in that practice and it would go a long way toward restoring AA's good name as a voluntary organization.

Ultimately, eliminating coercion into AA would almost certainly improve AA's current dismal effectiveness rate. At the very least, if AA returned to being a voluntaristic organization, it seems likely that its success rate would increase because of the placebo effect and member motivation, or readiness to change.

The Efficacy of 12-Step and Other Traditional Alcoholism Treatments

Amazingly given its predominance in the $10-billion-a-year treatment field there have been relatively few controlled studies (or long-term follow-up [longitudinal] studies of people with treated and untreated alcohol problems) involving 12-step treatment. Most studies, however, have found that 12-step treatment is ineffective as a means of combating alcohol abuse and alcohol dependence the findings have been, overall, that it is probably no better than no treatment at all.

The Vaillant Study

The two most important studies of 12-step treatment effectiveness are the massive NIAAA NLAES retrospective study (see Chapter 1) and a long-term longitudinal study of treated and untreated alcoholics by George Vaillant and other Harvard University researchers. The Vaillant study is probably the single best piece of evidence on 12-step treatment. It had several components, including a long-term longitudinal study of a group of "100 alcohol-dependent men and women followed for eight years after being admitted to a clinic for detoxification" (Vaillant, 1995, p. 2). (This clinic was the Cambridge and Somerville Program for Alcohol Rehabilitation, CASPAR, a 12-step inpatient/outpatient program, for which Dr. Vaillant worked as a psychiatric consultant.) Vaillant compared the outcomes of these 100 patients with those of alcoholic members of two comparison groups, subjects of two long-term longitudinal studies begun by other Harvard researchers: a group of 456 "Core City" Boston men, who had been followed since they were school boys; and a group of 204 former Harvard students, who had been followed since their college days. Vaillant also compared the outcomes of the 100 CASPAR patients with those of treated alcoholics in other studies, and with the outcomes shown in studies of untreated alcoholics. Vaillant's results were that the 12-step hospital treatment he helped to provide was utterly ineffective, as judged in comparison with studies of untreated alcoholics. As Vaillant remarked, "Not only had we failed to alter the natural history of alcoholism, but our death rate of three percent a year was appalling" (Vaillant, 1995, p. 352). He continued, "our results were no better than the natural history of the disorder."

Vaillant is a strong supporter of Alcoholics Anonymous. (He is currently a member of AA's General Service Board.) While reporting these "appalling" results, Dr. Vaillant commented, "if we have not cured all the alcoholics who were detoxified over 8 years ago, the likelihood of members of the Clinic sample attending AA has been significantly increased" (pp. 357 358). But did Vaillant really find that AA is that helpful? Actually, greater attendance at AA by the alcoholics he helped to treat did not enhance their treatment outcomes relative to untreated groups. Furthermore, in the Core City group which was followed for 50 years, 48 men among the alcohol abusers achieved what Vaillant termed "stable abstinence." Of these, about a quarter (27%) relied on AA (defined by having at least 30 AA visits by the age of 48 that is, well less than a year's attendance). In other words, even among those seeking abstinence, the large majority in Vaillant's untreated sample succeeded without AA.6

The National Longitudinal Alcoholism Epidemiological Survey

The results of the NIAAA's National Longitudinal Alcoholism Epidemiological Survey are as unsupportive of 12-step treatment as the Vaillant study. As mentioned in Chapter 1, the NLAES study was of 43,000 adult respondents. These included "4,585 adults with prior DSM-IV alcohol dependence," who were separately analyzed by Deborah Dawson (1996) in an article published in Alcoholism: Clinical and Experimental Research. All of these individuals were serious alcohol abusers, persons who would be termed alcoholics in common parlance. All of them had been alcohol dependent, not mere alcohol abusers. (Substance abuse and substance dependence are separate DSM-IV diagnoses.)

As to treatment, "Individuals were counted as having received alcohol treatment if they reported ever having gone to any of 24 different treatment sources . . . These sources included 12-step programs and others ranging from inpatient wards in general or psychiatric hospitals to halfway houses, employee assistance programs, and various types of doctors and health providers" (p. 773). Since over 90% of all treatment in the United States is 12-step treatment (Roman & Blum, 1997, p. 24), it's fair to regard the treated individuals reported in the NLAES study as 12-step-treated individuals, although it should also be noted that the inclusion of those who have attended "12-step programs" (read AA) muddies the waters a bit as regards the effectiveness of formal treatment.

Overall, the NLAES found that a third of treated and a quarter (26%) of untreated subjects were abusing or dependent on alcohol in the past year. Of those whose alcohol dependence appeared within the last five years, 70 percent who received treatment were drinking alcoholically in the past year. At 20 or more years since the onset of dependence, 90% of once-dependent individuals who had neither been formally treated nor had participated in 12-step groups were either abstinent or drinking without abuse or dependence, compared with 80% of the treated individuals who were no longer alcoholic.

These are remarkable results. They indicate that more treated than untreated alcoholics are subjecting themselves to alcohol abuse or dependence. These results should not be over interpreted, because the treated alcoholics in the NLAES reported a greater degree of alcohol dependence on average than the untreated alcoholics and, according to the NIAAA's Bridget Grant (1996), also reported more drug problems. On the other hand, all of the untreated subjects were diagnosed with alcohol dependence. The conclusion that alcoholics (alcohol dependent persons) who enter treatment are less likely to achieve remission than untreated alcoholics in the United States, although the populations are not identical, certainly puts a crimp in the claims made by treatment advocates of the value of extending treatment into the lives of more Americans.

Meta-Analyses of Alcoholism Treatment

William Miller and his colleagues at the University of New Mexico have conducted an ongoing meta-analysis of controlled (or clinical) studies of alcoholism treatment (Miller et al., 1995). (A meta-analysis is a statistical method of combining results of disparate research studies. In order to be included in Miller at al.'s analysis, a research study had to compare among matched or randomly assigned alcoholics the results of one type of treatment for alcoholism to another type or to no treatment at all.) Miller et al. reported that 13 treatment types had "too little basis" to support conclusions about their efficacy, while 30 treatment types had sufficient basis to support such conclusions. Treatments were assigned scores based on their combined degree of superiority (or inferiority) to other therapies.

The results of the 211 studies included in the analysis support the conclusion that 12-step inpatient treatment is ineffective. Miller et al. listed 10 studies of "milieu therapy," largely residential or inpatient treatment. Of the 10 studies, seven indicate that milieu therapy is ineffective in dealing with alcohol problems. Other standard approaches fare equally badly. "Unspecified 'standard' treatment" was reported ineffective in all three studies in which it was tested, and "general alcoholism counseling" was found ineffective in 14 out of 15 studies (p. 18).

Of the 30 types of treatment on the Miller et al. list, milieu therapy came in at 20th place on their "cumulative evidence scale"; unspecified standard treatment came in at 22nd place; and general alcoholism counseling came in at 29th place putting all of these therapies below, among others, "aversion therapy, nausea," "aversion therapy, electrical," and lithium therapy (p. 18).

Two components of standard 12-step therapy, confrontational counseling and educational lectures/films, were studied as separate therapies, and both fared dismally: confrontational counseling came in at 26th place, and educational lectures/films came in dead last.

It's also worth noting that of the 30 therapies listed, only psychotherapy (which came in at 28th place on the cumulative evidence score) was more expensive than milieu therapy, and that unspecified "standard" treatment and general alcoholism counseling were tied as being the fourth most costly treatments.

Two other meta-analyses have confirmed these results (Finney & Monahan, 1996; Holder et al., 1991). They both showed that expensive, commonly used treatments, such as "milieu therapy" and "general alcoholism counseling" (in the U.S., read 12-step inpatient and outpatient treatment), were among the least effective; and they also showed that almost all of the therapies consistently supported in studies as effective, such as the community reinforcement approach, marital therapy, and social skills training, were low cost or medium low cost. This is hardly surprising given that all three of these surveys were exhaustive and that they thus evaluated essentially the same materials virtually all available generalizable studies of treatment effectiveness.

The Edwards et al. Study / Brief Interventions

One graphic demonstration of the lack of efficacy of expensive treatment appeared in a 1977 study by Edwards et al. These investigators studied "100 men who were married or in a continuing cohabitation (common law marriage), consecutive attenders at the outpatient Alcoholism Family Clinic" (p. 1005). After completing an initial assessment, the men in the "advice" group (and their wives) were given a single counseling session in which they "were told that responsibility for attainment of the stated goals [abstinence, employment, and marital improvement] lay in their own hands" (p. 1006), but that the clinic would make monthly follow-up calls to check progress. The treatment group was introduced to AA and was then given a year of outpatient treatment, including psychiatric and social worker appointments. At the end of one year, there were no significant differences between the two groups on any of the outcome measures. As measured by separate reports from both the subjects and their wives, those in the advice group fared slightly better on some outcome measures, and slightly worse on others, than the treatment group. But there were no significant differences. That is, a year's worth of treatment produced results no better than those from a single advice session.

A number of studies have found that startlingly mild therapeutic interventions comparable to Edwards et al. produce results as good as or better than standard therapies. Indeed, this is the basis for the finding in Miller et al. (1995) that so-called brief interventions rate as the most effective alcoholism treatment according to controlled research. In such interventions, a client's drinking is often assessed in an ordinary health care environment, usually by a general practitioner or other nonspecialist health care worker. The health care worker points out heavy or problematic drinking, discusses it with the person, and they then set mutually agreeable drinking reduction goals. Progress is then checked during periodic follow-ups. In this remarkably straightforward procedure, very simple and inexpensive techniques suffice to improve the drinking behavior of most alcohol abusers. Nothing could be further from the elaborate 12-step spiritual approach that is embedded in virtually every alcoholism treatment program in the United States.

12-Step Drug Treatment

Given that 12-step drug and alcohol treatment are virtually interchangeable, and in a great many cases are administered in the same facilities by the same personnel, it's reasonable to conclude that their results should be similarly dismal. (As mentioned above, the National Treatment Center Study Summary Report states that only 2.4% of the treatment facilities surveyed offered segregated alcohol and drug treatment.)

To the best of our knowledge, there have been no controlled studies of 12-step drug treatment with random assignment of subjects and no-treatment control groups. The best studies available are two studies from the early 1990s of 12-step outpatient treatment in comparison to community reinforcement approach, or CRA (Higgins et al., 1991; Higgins et al., 1993), and one study of 12-step outpatient treatment versus cognitive-behavioral relapse prevention treatment (Wells et al., 1994).

The community reinforcement approach is a behavioral program that organizes an individual's environment to reinforce sobriety rather than relying on personal commitment or any kind of spiritual or internal resolution of the problem. (For example, the Higgins et al. CRA studies tied work and financial rewards to continued abstinence from cocaine.) Such cognitive-behavioral therapy introduces and reinforces new ways of thinking and approaches to problem-solving so that addicts learn alternative ways of coping with stress and insecurity other than turning to drugs or alcohol. This type of therapy relies on practical skills, and does not teach drug users that they are addicts with a life-long disease.

The Higgins Studies

In the first study (Higgins et al., 1991), 13 consecutively admitted cocaine-dependent patients at an outpatient treatment facility were offered behavioral treatment, the main feature of which was the community reinforcement approach. All 13 accepted. An additional 15 consecutively admitted cocaine-dependent patients were offered standard 12-step outpatient treatment (described in Chapter 1); 12 of the 15 accepted. The only major differences between the two groups were that the behavioral group had been using far more cocaine than the 12-step clients (an average of 10.2 grams for the behavioral clients versus 3.7 grams for the 12-step clients during the "most recent peak use" week), and that 69% of the clients assigned to the behavioral group were intravenous users compared with 16% of those assigned to the 12-step group (p. 1221).

Despite the greater severity of the problems of the clients in the behavioral group, their results were far better than those of the 12-step clients. Eleven (85%) of the 13 clients in the behavioral group completed the full 12 weeks of treatment, whereas only five (42%) of the 12 clients in the 12-step group completed treatment. In addition, ten of the clients in the behavioral group achieved four weeks of cocaine abstinence; six achieved eight weeks; and three achieved 12 weeks. In contrast, only three of the 12-step clients achieved four weeks of cocaine abstinence, and none achieved eight weeks, let alone 12 weeks. These results were confirmed by urinalysis.

The second study (Higgins et al., 1993), which featured a larger sample, random selection of subjects, and a longer study period (24 weeks versus 12 weeks in the original study), confirmed the first study's results. In the behavioral group, 11 of the 19 clients (58%) completed all 24 weeks of treatment, while in the 12-step group only two of the 19 clients (11%) completed 24 weeks of treatment. Only one of the behavioral clients, but eight of the 12-step clients, dropped out of the study after only one therapy session. And 74% of the behavioral clients managed to abstain from cocaine for at least four weeks; 68% abstained for eight weeks; and 42% abstained continuously for 16 weeks. The comparable figures for the 12-step group were 16% for four weeks; 11% for eight weeks; and 5% (one client) for 16 weeks.

The Wells Study

The third study of 12-step drug treatment versus cognitive-behavioral drug treatment involved 110 cocaine-abuse clients alternately assigned to 12-step outpatient therapy or cognitive-behavioral outpatient relapse prevention therapy (Wells et al., 1994). Treatment lasted for 24 weeks and consisted of 17 two-hour therapy sessions for both groups. There were no statistically significant differences between the 12-step and relapse prevention clients at intake, although the relapse prevention patients were using cocaine, alcohol, and marijuana somewhat more often than the 12-step patients at that point on average, slightly over 25% more often, as measured in days used per month.

Results were measured in the number of drug-using days per 30 days at 12 weeks and six months following the start of treatment. In regard to cocaine use, the 12-step clients had been using cocaine an average of 4.4 days per month at intake. This dropped to 1.3 days per month at the 12-week measure, but climbed back to 1.9 days per month at the six-month measure. The same pattern was seen in alcohol use by 12-step clients, who drank an average of 7.7 days per month at intake, dropped to 5.0 days per month at the 12-week measure, but climbed back to 7.2 days per month at the six-month measure. Note that such subjects succeeded neither at abstaining entirely nor at limiting their alcohol use, that they did not succeed at abstaining from cocaine, and that their use of the drug was rebounding at the six-month measure.

In contrast, the relapse prevention clients showed a steady drop in cocaine use, and only a slight rebound in alcohol use. Cocaine use fell steadily from 5.5 days per month at intake, to 2.3 days per month at the 12-week measure, to 1.6 days at the six-month measure. Drinking averaged 9.9 days per month at intake, dropped to 7.9 days per month at the 12-week measure, and then increased slightly to 8.4 days per month at the six-month measure.

Only with marijuana was there a continual decline in use by the 12-step clients: from 4.8 days per month at intake, to 3.2 at 12 weeks, to 3.0 at six months. The relapse prevention clients showed a similar, though slighter, decrease: from 5.9 days per month at intake, to 5.0 at 12 weeks, to 4.8 at six months.

Primarily because of the statistically significant (p>.04) increase in the amount of alcohol consumed by the 12-step clients between the 12-week and six-month measure, Wells et al. (p. 14) concluded, "With its emphasis on anticipating and coping with future high-risk situations, relapse prevention [treatment] would be expected to be superior [to 12-step treatment] in helping clients maintain reductions in drug usage over time."

Use by 12-step clients of alcohol and cocaine were similar they both showed an initial steep decline, followed by a steep increase between the 12-week and six-month measures. In contrast, the relapse prevention clients showed only a very slight increase in alcohol use between the 12-week and six-month measures, and a large decrease in cocaine use in the same period. Finally, the relapse prevention clients showed a continual decrease in cocaine use across all three measures; while they were using cocaine on 25% more days per month than the 12-step clients at intake, by the end of the study they were using cocaine on 13% fewer days than the 12-step clients

The Texas Prison Study

A number of research studies have claimed that exposing inmates to treatment (or, often, coercing them into treatment) reduces recidivism (returning to crime) and relapse (returning to addiction/ alcoholism) rates. Thus, the requirement of treatment in prisons and even the substitution of treatment for imprisonment (for example, through sentencing by drug courts) have become mantras for drug policy reformers and treatment providers. Yet, one of the most careful studies of the effects of treatment on inmates and probationers (those given treatment in place of imprisonment), a comparative study conducted over three years in the Texas penal system, yielded a less positive and in some respects, opposite result.

That study was released by Texas' Criminal Justice Policy Committee (Eisenberg, 1999). (The nature of the treatment used was not described in any detail, but it almost certainly was 12-step treatment.) This report covered four programs involving 1654 prisoners who had undergone treatment while in prison, and 1673 probationers who had undergone treatment while on probation.

The recidivism rate three years after treatment was exactly the same for those prisoners who had undergone treatment as for those in the no-treatment comparison groups (drawn from the same population): one program had a 42% recidivism rate for both treated and comparison groups, while the other program had a 37% recidivism rate for both the treated and comparison groups. The results with probationers were even more discouraging. Three years after treatment, treated probationers in one program had a 38% imprisonment rate, while members of the no-treatment comparison group had a 35% imprisonment rate; and in the other program, the treated probationers had a 44% imprisonment rate compared with only a 35% imprisonment rate for the comparison group.

When contemplating that other studies have yielded more positive results, the chief investigator in this study observed:

Outcome evaluations for programs in other states have focused primarily on reporting the outcomes of offenders completing the program and not on all program participants. . . . For example, an evaluation of a California therapeutic community program for prisoners reported that 25% of offenders completing the program were arrested in one year versus 67% of the comparison group arrested. When the recidivism rate of offenders not completing the program is included, the recidivism rate of all program participants is 53%. (Eisenberg, 1999, p. 12)

In many cases, optimistic claims about treatment are based only on those who complete treatment which is another way of evaluating only the best-prognosis patients. Much of the optimism about programs treating convicted felons is based on such poor science. After all, what does it mean that, in an expensive, large-scale treatment program in one of our largest states, treatment at best did no good?

Many probationers and prisoners who drop out of treatment programs must find those programs particularly objectionable, given the dire consequences they face (imprisonment or continued imprisonment) for failing to complete these programs.

Dropout Rates

The typically high dropout rate of 12-step treatment programs is a telling indication of their lack of effectiveness. Twelve-step treatment programs often make grossly exaggerated claims about their effectiveness, with some citing success rates as high as 75%, 85%, or even 95%. A brief look at dropout rates from such programs gives lie to such claims. SAMHSA's 1992 1997 TEDS report states that, overall, only 47% of patients complete treatment (and another 12% are transferred to other programs with no indication of whether they complete that treatment) (SAMHSA, 1999). Thus, the overall completion rate in American treatment is, at best, under 60%, and could be under 50%. Remembering that the 12-step approach is utilized in 93% of American treatment facilities, according to the National Treatment Center Study Summary Report, it's fair to take these percentages as the percentages for 12-step treatment.

In the group subjected to the most direct coercion, marijuana clients, 52% of whom are directly coerced by the criminal justice system and only 18% of whom are individually referred, one finds a similar pattern: only 48% complete treatment, with another 11% being transferred to other programs. That so many clients (many, undoubtedly, under threat of jail or prison time) leave these programs is a good indication of how objectionable they find 12-step treatment.

Why should dropout rates be taken into account when calculating the effectiveness of treatment programs? Consider two hypothetical alcohol-abuse treatment programs, "A" and "B." Both admit 100 comparable clients, and both show recovery rates of 80% for those who complete treatment. Based on this, are both programs equally successful? If the dropout rate for "A" was only 50% and the dropout rate for "B" was 70%, and the recovery rate in both programs was 80% for those who completed treatment, "A" would yield 40 recovered clients, or almost twice as many as "B," which would yield only 24 recovered clients. Obviously, the two programs are far from equally successful.

But 12-step advocates typically ignore the sky-high dropout rates in 12-step treatment programs, dismissing dropouts with cliches such as "the program works if you work the program" (implying that "the program" always "works," and that when it doesn't the fault is always the client's). This, of course, misses the point, as dropouts must be included when calculating effectiveness of treatment.

The "Best" Evidence of the Efficacy of 12-Step Treatment

There are no scientific studies featuring random assignment of subjects and no-treatment control groups that indicate that 12-step groups or 12-step treatment are effective means of dealing with addiction problems. In the absence of such studies, 12-step advocates have repeatedly cited uncontrolled, ungeneralizable studies as if they were generalizable, and they have latched onto three studies with comparison groups (but without no-treatment control groups) as evidence (or even "proof") of the efficacy of AA and 12-step treatment. One 12-step advocate, George Vaillant (1995, p. 360), provides a good example of the hyperbole common to such apologetics: "The recent randomized trial by Walsh and colleagues (1991) has put all negative findings about the limitations of hospital treatment for alcohol abuse in question." What exactly is Vaillant referring to?

The Walsh Study

Walsh et al. (1991) was a study of 227 alcoholic industrial workers in New England, who were coerced by their employer into either 12-step inpatient treatment or AA, or were given a "choice" of treatment options although the authors note that "the staff or the employee assistance program sometimes encouraged them" to enter either 12-step inpatient treatment or AA. As a result, of the 71 workers in the choice group, 62 (87%) entered either 12-step inpatient treatment or AA. Of the remaining nine subjects, three chose psychotherapy, and six chose no treatment at all. The researchers lumped them in with the 62 workers who had chosen some form of 12-step program. Thus, this study not only didn't have a no-treatment control group, it didn't even have a non-12-step comparison group.

The subjects were followed for 24 months, and the major outcome results studied were "job outcome" (i.e., how many from each group were fired), drinking outcome, and "other drugs and group outcomes." Overall, the inpatient group did better than the AA or choice groups, though it should be noted that "better" here is in comparison with rather poor outcomes.

In terms of job outcome, there were "no significant differences among the groups." Overall, 14% of clients were fired during the study's two-year period (p. 778). As for drinking outcomes, there were no significant differences between the groups in regard to four important measures, "mean number of daily drinks, number of drinking days per month, binges, and serious symptoms," although the inpatient group did significantly better than either the AA or choice groups in regard to continuous abstention: 37% for the inpatient group; 17% for the choice group; and 16% for the AA group.

The inpatient group also did better than either the AA or choice groups in regard to (re)hospitalization: 42% of the randomly assigned subjects were hospitalized for additional treatment: 23% of the hospital group, 38% of the choice group, and 63% of the AA group over the two-year period.

As for drug outcomes, "of the cocaine-using subjects in the hospital and choice groups, less than 30 percent underwent additional inpatient treatment, as compared with 63 percent of the cocaine users randomly assigned to AA alone" (p. 779).

Overall, these outcomes can only be described as miserable, especially considering that being coerced into treatment under threat of job loss is a powerful wake-up call which in itself should have jolted some workers into making significant changes and considering the very high cost of the treatment employed. The average cost of treatment for the compulsory AA group came to $8840 over the two-year study; the average cost for the choice group was $14,400 for those who chose initial inpatient treatment, and $8800 for those who didn't; and the average cost for the compulsory inpatient group was $10,040.

Due to the nature of this study (comparing only 12-step approaches), any generalized conclusions derived from it about the effectiveness of 12-step treatment should be approached with great caution. What would be truly telling would be a similar trial, but one involving a no-treatment control group and comparisons with low-cost behavioral treatments, such as motivational enhancement, social skills training, and the community reinforcement approach. Nonetheless, what may be most overwhelming in the results of this study is the poor performance of the AA-alone group. This group had the highest (re)hospitalization rate and the lowest rate of never becoming intoxicated (although it shared the lowest abstinence rate with the few "choice" subjects, many of whom of course selected AA).

Project MATCH

The second commonly cited "proof" of the effectiveness of 12-step treatment is Project MATCH (1997). When this $35-million, NIAAAfunded study was released in 1997, treatment supporters claimed variously that it demonstrated the validity of 12-step treatment, that all of the tested forms of treatment work equally well, or even that 12-step treatment is superior to other forms of treatment. In reality, it showed none of these things.

Project MATCH 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 therapy. Post-treatment assessment was carried out over a period of 12 months following the end of formal treatment.

The significant findings of Project MATCH were that patients in all three groups experienced remarkably similar improvement, as measured by the number of drinking days per month and the incidence of binging. The only significant difference between the groups was that 12-step-treated clients with less severe psychological problems had more abstinent days than similar clients in the other two groups. But this one positive difference was not one of those predicted by the researchers. Of 21 predicted differences in outcomes, depending upon the match between a patient's characteristics and the type of treatment received, none of the hypotheses generated by this distinguished group of alcoholism researchers (comprising the leading alcoholism treatment scientists in the United States) was substantiated. What then did Project MATCH show?

On average, the alcoholics in this study reduced their drinking days from 25 to fewer than 6 per month, and from 15 to 3 drinks on each drinking occasion (Connors, 1998). Clearly, these are dramatic results, but were they the result of treatment or of other factors?

Project MATCH was so over designed that it seems likely that any form of treatment used in it would have shown similar results. 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, contrasting them with the majority of treatment enrollees in the U.S. (see Chapter 1). In addition, subjects with simultaneous drug dependencies were not accepted, even though patients with both drug and alcohol problems are the most common admittees to substance abuse treatment (SAMHSA, 1999) and show lower remission rates than alcohol-only clients (Grant, 1996). 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 problems." More than half of the remaining potential clients were eliminated from the study for reasons such as "failure to complete the assessment battery, residential instability, [and] legal or probation problems." These disqualifications ensured that only the most well-adjusted clients would participate, thus introducing another positive biasing factor: social and emotional stability. In other words, as far as alcohol-abuse clients go, Project MATCH was dealing with those most likely to succeed in any program; of 4,481 potential participants who went through initial screening, only 1,726 became study participants.

Client expectations were another positive biasing factor. The clients knew that they were taking part in a special study, and the manner in which sessions were conducted undoubtedly led to high expectations. The study entailed "compliance enhancement procedures (i.e., calling clients between sessions, sending reminder notes and having collateral contacts)" (p. 23). The most skilled professionals and researchers in the United States first wrote manuals detailing each type of treatment. Supervisors and counselors were then trained using the manuals as reference. Every one-to-one therapy session was videotaped, and supervisors monitored a quarter of the sessions.

This degree of training and supervision makes treatment in Project MATCH entirely unlike standard treatment administered in the U.S. The difference between the forms of treatment employed in Project MATCH and actual treatment in the U.S. was undoubtedly most pronounced in the study's 12-step facilitation therapy. In the real world, 12-step treatment is predominantly group treatment, and individual therapy and group sessions are normally administered by paraprofessionals with limited professional training (usually AA/NA members with the title Certified Alcoholism Counselor). Furthermore, the trained therapists administering 12-step facilitation in Project MATCH would have been unlikely to engage in the highly directive, even abusive, behavior that is commonplace in 12-step programs, where drinkers are confronted with their deficiencies and forced to confess that they are alcoholics.

The high number of follow-ups with patients (for the purpose of assessment) actually may have been the most confounding factor of all in Project MATCH. Several studies have found that follow-ups are critical determinants of treatment outcomes that is, rather than telling people what to do, checking to see how they are doing on a regular basis is the key to helping them reduce their drinking. Indeed, this is the key to brief interventions, where a minimal intervention is secondary to regular follow-ups with patients. In line with such findings, the briefest of the treatments in Project MATCH motivational enhancement (ME) produced results identical with those of the other treatments. ME had only four scheduled sessions as compared to the 12 sessions scheduled for the 12-step and social skills treatments (although patients on average attended only about two-thirds of these). The researchers thus noted that subjects in the ME groups actually spent more time in follow-up assessments (five hours) than they did in the treatment being assessed!

Given all of these biasing factors, it was hardly surprising that all forms of treatment showed remarkably similar, and positive, outcomes. Indeed, it seems possible that almost any form of treatment might have shown similar results. The study's authors recognized after the fact this possibility, when they noted: "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." They continued, "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." We'll 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" (Project MATCH, 1997, pp. 23 24).

Even accepting the findings of Project MATCH as valid, the conclusion emerges that 12-step treatment should not be the preferred form of treatment for alcohol-dependent and alcohol-abusing individuals. Because all three forms of treatment studied in Project MATCH showed very similar results, it stands to reason that the most cost-effective treatment should be the preferred treatment. And that preferred treatment would be motivational enhancement, which produced the same results as 12-step treatment with half or fewer treatment sessions.

In the end, what Project MATCH actually demonstrated was that with extremely well-designed and controlled treatment (unlike that delivered virtually anywhere in the U.S. today), even very alcoholic patients can reduce their drinking substantially. As the motivational enhancement results showed, such improvement does not require extensive time in treatment a couple of hours may be sufficient. In Project MATCH, such treatment produced excellent results at a fraction of the time and cost required by standard treatment programs.

The Ouimette Study

The third study commonly cited by 12-step advocates as evidence of efficacy was conducted by Ouimette et al. (1997) with participants in Veterans Administration substance abuse programs. This large study surveyed 3699 patients in 15 VA hospitals who had participated in either 21or 28-day inpatient treatment, and divided them by treatment type: 12-step, mixed 12-step/cognitive-behavioral, and cognitive-behavioral. The programs were classified according to descriptions provided by program directors and staff members. Results were measured using 11 variables, such as daily alcohol consumption, abstinence, remission, employment status, housing status, and arrests, and were derived primarily from self-report an average of 13.2 months after the patients left the programs.

The results were surprising:

Although 12-step patients were somewhat more likely to be abstinent at the 1-year follow-up, 12-step, C-B, and combined 12-step C-B treatment programs were equally effective in reducing substance use and improving most other areas of functioning. The finding of equal effectiveness was consistent over several treatment subgroups: Patients attending the "purest" 12-step and C-B programs, and patients who received the 'full dose' of treatment. Also, patients with only substance abuse diagnoses, those with concomitant psychiatric diagnoses, and patients who were mandated to treatment showed similar improvement at the 1-year follow-up, regardless of type of treatment received. (Ouimette et al., 1997, p. 230)

A few pages later the authors note:

[A]nalyses examining the main effects of treatment type were performed on the subset of patients who received the 'full dose' of 12-step, C-B, and mixed treatment. Results were the same as in the complete sample. (Ouimette et al., 1997, p. 236)

This statement is highly provocative. It makes one wonder about the results for those who dropped out were these very different from the results of those who went all the way through treatment? Unfortunately, there is no way to examine this. The authors who analyzed the results for several different subgroups did not analyze the results for treatment dropouts (P.C. Ouimette, personal communication to C.B., September 22, 1999).

There were other loose ends in this study, especially the presence of factors that tended to bias its results in favor of 12-step treatment. The most important of these biasing factors is that in calculating their results, the authors disregarded the difference in follow-up rates between the 12-step and C-B groups. Fully 85% of the C-B patients participated in the follow-up self-report, while only 78% of the 12-step patients did so, a statistically significant difference (p>.001). Given that failure to participate in follow-up studies is often an indication of a bad outcome, and that 12-step devotees would want to report favorable outcomes (probably more so than C-B patients, who wouldn't have religious zeal as a motivator), this omission could well have biased the results in favor of 12-step treatment.

Other aspects of the study also seemingly biased the results in favor of 12-step treatment. One is that, in calculating their results, the researchers refused to classify any patient as remitted who had smoked as little as a single joint in the more than a year between treatment and follow-up. Given the extremely heavy emphasis on abstinence from alcohol and illicit drugs (but not tobacco) in 12-step treatment, and this study's reported higher abstinence (but not overall improvement) rate among those who underwent 12-step treatment, it seems very likely that more otherwise-remitted C-B patients than 12-step patients were classified as continuing substance abusers in this study's conclusions because of marijuana use, no matter how infrequent.

In addition to these factors potentially biasing results in favor of the 12-step treatment, there were additional problems with this study's methodology that impede drawing any firm conclusions from it. The programs studied were not "pure" 12-step or C-B programs; all contained significant elements of both approaches. It is also relevant that "[s]taff in all three program types did not differ on psychosocial model beliefs, probably reflecting the emphasis on social and environmental change in all programs" (p. 232). This is an essentially cognitive-behavioral emphasis (in contrast to the 12-step emphasis, which reduces alcoholism to an individual disease, the cure for which is individual turning one's will and life "over to the care of God"), and at least suggests that any success that the 12-step VA treatment showed might have been the result of its inclusion of cognitive-behavioral components.

As with other research, "all programs expected aftercare participation and provided referrals to both outpatient treatment and community-based self-help organizations" (p. 232). But the authors give no details on how many patients participated in aftercare, nor how long, nor what type. Since it seems likely that a fair number did participate in such care, this makes it extremely difficult to determine whether any effects shown were the result of the 21 or 28-day treatment (or patient effects) or the possible year-plus of aftercare.

Finally, the Ouimette et al. study's population was very distinctive. It was composed entirely of military veterans; it was entirely male; blacks were greatly over represented (49%); hispanics (3%) and anglos (46%) were greatly under represented; only 24% of its subjects were employed; only 19% were married; and subjects were older than average (43). It seems very likely that militarily trained subjects would respond better to the regimented, religious approach embodied in the 12 steps than average. This may have been a population best suited to the 12 steps (in which case, the absence of superior results from this type of therapy suggests that even for this group alternative approaches are at least equally effective). Would a youthful, middle-class, college population respond similarly well to the 12 steps?

We should also note that the Ouimette et al. study did not have a no-treatment control group, which means as the authors conclude "the possibility exists that these three treatment programs produce patient outcomes similar to those that naturally occur with no treatment" (p. 239).

Some Thoughts On 12-Step Treatment Effectiveness

Normally, in scientific discourse on the treatment of diseases, the burden of proof falls on those proposing a treatment. That is, those who assert that a treatment is effective are obliged to provide convincing evidence of its efficacy before it passes beyond clinical studies, let alone is administered on a mass scale. Yet those who promote the disease concept of alcoholism and 12-step treatment have never provided such proof, and they ignore the preponderant conclusion of the available generalizable scientific studies that 12-step groups and 12-step treatment are ineffective (or are neither the most effective nor the most cost-effective) means of dealing with addiction problems. That they ignore the available scientific studies and offer virtually no evidence beyond anecdotal claims and uncontrolled studies, yet continue to trumpet 12-step groups and 12-step treatment as the best, if not the only, means of dealing with addiction problems, is a telling sign that their advocacy of the $10 billion dollar-a-year, frequently coercive 12-step industry is not a matter of standard medical evaluation. Rather, their support for this approach is most likely a matter of religious belief and/or of financial interest.

Is Abstinence Necessary?

In addition to the questions of whether standard substance abuse treatment works, of whether standard treatment is superior to minimal or no treatment, and of whether forcing people into treatment against their will is justified (if it can be) by successful outcomes, a related question is whether substance abusers must be required to abstain from all recreational substance use. Over 1,000,000 Americans are coerced into treatment annually, and probably nearly all are required to abstain from alcohol entirely during and often after treatment under threat of probation or parole violation, prosecution, loss of employment, loss of professional certification, loss of child custody, or loss of organ transplant candidacy, among other penalties. This adherence to abstinence is routinely enforced by mandatory urinalysis (or, at least, the threat of it).

Why should anyone be forced to submit to such indignities and to suffer such drastic penalties for taking so much as a single drink? The answer lies in 12-step alcohol treatment ideology and its influence on public policy. It has become an article of faith in the American alcoholism treatment industry that abstinence is the only acceptable treatment goal. Indeed, in a survey of treatment providers, 75% of those who responded stated that "nonabstinence was not an acceptable goal for patients in their program," and of the remaining 25%, "70% reported moderate drinking as appropriate for only 1 25% of their clients." In addition, of those who accepted moderate drinking (in any form) as a goal, only one-fifth (that is, 5% of the total) worked in inpatient facilities (Rosenberg & Davis, 1994). That the requirement of abstinence is almost total in such treatment programs is clear in the National Treatment Center Study Summary Report. In addition to reporting that 93% of treatment facilities (and over 95% of inpatient facilities) utilize the 12-step approach, this report found that "nearly all the programs (98.6%) said they advocated abstinence from alcohol for all their alcohol and/or drug dependent patients" (Roman & Blum, 1997, p. 24). This rejection of moderate drinking as a goal is attributable to two closely interrelated belief systems: AA ideology and its offspring, the disease concept of alcoholism, both of which posit that alcoholics can never learn to drink safely, and that if they attempt it they will only get worse.

But what is the evidence regarding moderate (or "controlled") drinking by alcoholics? Do they really suffer inevitable disaster if they take so much as a single drink?

One widely publicized study provides evidence that a significant number of problem drinkers can and do return to nonproblem drinking rather than turn to abstinence. That study is Project MATCH. Even though virtually all subjects were diagnosed alcohol dependent, and all three forms of treatment in Project MATCH had abstinence goals, significant numbers of individuals in both the outpatient and aftercare groups (of all three forms of treatment) resumed "regular drinking" (as opposed to "heavy drinking") during the 12-month follow-up period: this included 35% of outpatient clients and 25% of aftercare clients (Project MATCH, 1997, p. 14). This is perhaps the most significant finding of Project MATCH.

Other studies have confirmed that a great many alcoholics (and not only problem drinkers or alcohol abusers, but those classified as alcohol dependent) resolve their alcohol problems through moderation rather than abstinence. The NLAES (Dawson, 1996) provides a particularly important demonstration of this fact, because this extremely large (n=4,585) national survey of ever-alcohol-dependent individuals found that reduced drinking is the most common way in which people overcome alcoholism. NLAES showed that at 10 to 20 years since onset of dependence, fully 30% of those who had been treated and 64% of those who had never been treated were drinking without abuse or dependence, and that at 20 or more years since the onset of dependence, 24% of the treated individuals and 60% of the untreated individuals were drinking without abuse or dependence. Of all NLAES subjects, more than half (58%) of untreated and fully half of all alcoholics, whether treated or untreated, were able to drink without achieving a DSM-IV diagnosis. This means that a high percentage of individuals who undergo 12-step abstinence treatment end up drinking moderately despite the disease-concept indoctrination they receive during treatment, and that moderate drinking is the standard outcome for formerly alcohol-dependent individuals who never undergo treatment.

Other studies confirm this finding. Reports that at least some formerly alcohol-dependent individuals resolve their problems through moderation have appeared in scholarly journals since at least 1962 (Davies, 1962). For example, in 1976 the first Rand Report (Armor et al., 1976) reported that of treated alcoholics, 12% were drinking normally 6 months after treatment and 22% were drinking normally 18 months after treatment. A follow-up study (Polich et al., 1981) reported that a similar percentage were drinking moderately four years after treatment. A 1985 study of treated alcoholics reported that 20 years after treatment 15% of those responding were drinking moderately (O'Connor & Daly, 1985). And a 1986 study of treated, formerly alcohol-dependent persons indicated that 16 years after treatment, 20% of the surviving subjects were drinking moderately whereas only 15% were abstaining (McCabe, 1986).

More recent studies have confirmed that a high number of formerly alcohol-dependent persons resolve their problems through moderate drinking. A 10-year study of treated alcoholics reported that in the year prior to the 10-year follow-up, 24% of the subjects were drinking moderately (Finney & Moos, 1991, pp. 48 49), and that over the years there was even "a small increase in the number of persons reporting social or moderate drinking," rather than an increased danger of moderate drinkers falling off the wagon. Two sizable 1996 surveys of Canadians (of, respectively, 11,634 and 1,034 respondents) reported that "most individuals (77.5% and 77.7% respectively) who had recovered from an alcohol problem for 1 year or more did so without help or treatment. Sizable percentages (38% and 63%) also reported drinking moderately after resolving their problems" (Sobell et al., 1996, p. 966).

Over the years, these and other studies have confirmed the general finding that moderate drinking is a common means of resolving drinking problems, including alcohol dependence. Those who resolve their problems in this manner include a surprisingly large number of treated alcoholics who have been indoctrinated into the belief that they cannot under pain of an eventual, horrible death ever drink again. At the same time, the resumption of nonproblematic drinking is far more common among untreated alcoholic and problem drinkers; in NLAES, the much greater likelihood of continued drinking that does not meet DSM-IV criteria for alcohol abuse accounts for the superiority in outcomes favoring those alcoholics who do not enter treatment.

All of these studies but especially the long-term studies, such as NLAES, McCabe, and Finney and Moos provide powerful evidence that abstinence is not the only way to beat an alcohol problem, and that the disease concept of alcoholism, with its loss-of-control beliefs and claims of inevitable progression of alcohol problems, is simply unfounded assertion. Thus, it seems grossly unfair in fact, a crime that individuals are routinely coerced into total abstinence and suffer drastic consequences if they take so much as one drink, in the complete absence of scientific evidence that total abstinence is necessary.

A number of studies have compared the results of controlled drinking and abstinence goals in treatment. These studies have generally found results to be comparable (and particularly when alcoholics are allowed to select their treatment). One of the first studies of traditional abstinence treatment versus controlled drinking treatment (Pomerleau et al., 1978) reported a 72% improvement rate in the controlled drinking group at the nine-month follow-up, versus only a 50% improvement rate in the abstinence group. A longer-term study of individuals assigned to either behavioral abstinence treatment or behavioral controlled-drinking treatment (SanchezCraig et al., 1984) showed no significant differences in rates of improvement throughout the two years of the study. An even longerterm study (Rychtarik et al, 1987) showed no differences between those given abstinence treatment and those given controlled-drinking treatment at follow-ups five to six years after treatment, with about equal numbers in both treatment groups abstinent or drinking moderately.

Furthermore, the opportunity to select and pursue a treatment goal whether abstinence or controlled drinking enhances the likelihood of success. This predictable outcome would be anticipated in any other field than alcoholism and substance abuse, where the abstinence fixation has been in place for decades. Again, the seemingly remarkable results of NLAES, that more untreated alcohol-dependent individuals overcome alcoholism than treated individuals, could be largely the result of the opportunity untreated individuals have, once they become serious about resolving an alcohol problem, of selecting a realistic goal which they believe they can attain.

Studies specifically measuring the effects of goal-choice upon outcome provide no support to those who endorse coerced abstinence. One of the first such studies (Booth et al., 1984) reported on a behavioral residential program in the United Kingdom in which patients were given the choice of an abstinence goal, a moderation goal, or, because of medical conditions, were assigned an abstinence goal. There were no significant differences between the groups voluntarily choosing either abstinence or moderation at the 12-month follow-up. The only significant difference was that the assigned-abstinence group did considerably worse than either of the voluntary goal groups.

A slightly longer-term (two-year) study (Elal-Lawrence et al., 1986) showed that abstinence, successful moderation, or relapse were unconnected to patients' initial choice of treatment goals. Nonetheless, alcoholics' success at either outcome was related to their belief systems and previous experiences, rather than to the severity of their alcohol dependence (whereas treatment specialists and researchers have generally claimed that more severely alcoholic individuals have no choice but to abstain). This result was clarified in Orford and Keddie's (1986) finding that alcoholics' persuasion that controlled drinking or abstinence is possible is more critical for achieving either of these goals than is their level of alcohol dependence.

Graber and Miller (1988), on the other hand, showed that there were differences but not statistically significant ones between groups given the choice of abstinence or controlled-drinking treatment three-and-a-half years after treatment. In this study, the number of those achieving abstinence was equal in both groups, while the number of "asymptomatic" or "improved" drinkers was higher in the abstinence-goal group. Indeed, in the abstinence group, 75% of those who had improved were drinking, not abstinent; all of the apparently greater improvement in the abstinence-goal group was attributable to the number of abstinence subjects who were still drinking either nondestructively, or less destructively than before treatment and not from the small number who had achieved abstinence.

A 1992 study by Booth et al. confirmed the results of Booth et al.'s first study. In the second study, there was again no difference in outcome between those choosing abstinence treatment and those choosing controlled-drinking treatment. There was a slight shift from moderation to abstinence goals initially, 60% chose an abstinence goal, rising to 64% by the end of the study but the investigators reported no difference in overall outcomes for the two groups.

A 12-month study of "chronic alcoholics" given the choice between behavioral abstinence or controlled-drinking treatment (Hodgins et al., 1997) again showed the typical pattern of clients moving from moderation to abstinence goals during treatment. In this study, 44% of clients chose an abstinence goal at intake, but by the fourth treatment session almost half of the moderation-goal clients had switched their goal to abstinence. The authors reported, "Goal at initial assessment was not significantly related to consumption at follow-up generally, except for the number of intoxicated days. Those with an initial goal of abstinence reported fewer intoxicated days at follow-up" (p. 251). They also reported that their data "did not reveal a relationship between goal choice and outcome in terms of employment (days worked), social stability, and participation in leisure and social activities at the 3, 6, and 12-month follow-ups." While 88% of those who chose an abstinence goal showed at least improved status at the 12-month follow-up, a significant percentage of those who chose a moderation goal also improved 69% at the 12-month follow-up.

During the course of many of these studies, patients shifted their choice of goals often from moderation to abstinence; those who chose abstinence after an initial moderation goal often had good outcomes. Miller (1991) labeled this approach to abstinence through a process of personal searching and the development of a realization that abstinence is best for the individual "warm turkey" (as opposed to "cold turkey").

In summary, studies of assigned-abstinence versus moderation-goal treatment, and of client choice in treatment, indicate that there is often little connection between initial goal and eventual outcome. Other studies show advantages in motivation and outcome from allowing subjects to choose their own treatment goals. In any case, the research consistently indicates that a substantial number of former alcohol abusers and alcohol-dependent persons resolve their problems through moderation rather than abstinence. It also indicates that there is simply no justification in a large majority of cases for coercing anyone into total abstinence; and the evidence also indicates that the often-barbaric penalties inflicted upon persons who have as little as one drink are unjustified.

DUI Studies

DUI studies are not a good measure of alcohol-abuse treatment effectiveness for several reasons. In the first place, at least half of those coerced into participating in treatment as a result of DUI charges do not meet "diagnostic criteria for alcohol abuse or alcoholism" (NHTSA, 1996, p. 1). Most drunk drivers are, in fact, better characterized as antisocial or criminally oriented individuals who have extensive arrest records for offenses other than DUI, both before and after their DUI convictions (Argeriou et al., 1985).

Given the non-alcoholic nature of most of the individuals subjected to DUI-induced treatment, it is reasonable to expect that any observed changes in their behavior are the result of factors other than efficacy of treatment for a problem alcohol abuse or dependence from which they do not suffer. In addition, almost all DUI studies measure outcome in rearrest or accident rates that is, outcomes in DUI studies are at best an indirect measure of treatment efficacy.

Even where efficacy in reducing DUI recidivism is shown, this does not demonstrate that the individuals in question are drinking less, merely that they are driving less often under the influence of alcohol or are simply being more careful and are being caught less often. In fact, rather than indicating treatment efficacy, these results speak to the "harm reduction" approach. That is, rather than helping people overcome a disease or disorder, punishment (sometimes in the form of treatment) generally works by prodding people to resolve to avoid harmful behaviors, while they continue to drink.

There have been a number of studies on the effects of treatment and alcohol education in reducing DUI recidivism and accident rates as compared with normal judicial sanctions. The bulk of this research shows that suspension or revocation of a driver's license is more effective than treatment in dealing with driving-under-the-influence offenders (see, for example, Hagen, 1980; Salzberg & Klingberg, 1983). In a major quasi-controlled comparison, four California counties in which an alcohol rehabilitation treatment program was implemented were compared to four counties where simple licensure actions were imposed. The major difference was that DUI convicts had fewer crashes in the licensure action counties. However, although license revocations or suspensions were more effective than treatment, this advantage arose primarily because drivers were more careful, and not because they were rehabilitated, or even drove less while intoxicated (Perrine & Sadler, 1987). "Neither rehabilitation nor sanctions had much effect on rearrests for DWI" (USDHHS, 1990, p. 248). Here, as elsewhere, the emperor (AA/12-step treatment) wears no clothes. For information regarding DUI studies useful to individuals facing 12-step coercion for both DUI and non-DUI charges, see Nichols (1990), Ross (1992), and USDHHS (1990).

One important piece of information for those charged with or convicted of driving under the influence is that most such individuals are not alcohol dependent or even alcohol abusers a number are social drinkers who made a mistake and were caught at it and others are dangerous drinkers rather than alcohol-dependent drinkers. A deputy director in the National Highway Traffic Safety Administration states that, "During any given 1-year period, approximately 20 percent of licensed drivers drive while intoxicated" (Nichols, 1990, p. 45). Compare this with the 7.4% of the population who are either alcohol abusers or alcohol dependent, as estimated by the Department of Health and Human Services (USDHHS, 1997, p. 21), and it immediately becomes apparent that a majority of those who drive under the influence are not alcoholics.

Yet, in most areas of the U.S., a large majority, if not all, of those convicted of DUI or placed in DUI pretrial diversion programs are forced as a condition of probation (or deferred prosecution) to participate in AA or 12-step treatment (Brodsky & Peele, 1991; Weisner, 1990). Clearly, there is no legitimate rationale for coercing non-alcoholics into alcoholism treatment especially given the demonstrated lack of efficacy of such treatment in reducing DUI recidivism.

What Works?

America's present, religiously based alcohol-abuse treatment system is a dismal failure. It is extremely expensive, massively coercive, apparently no more effective than the rate of spontaneous remission, and has blocked the implementation of more effective alternatives (see Brodsky & Peele, 1991; Bufe, 1998; Peele et al., 1991). But no matter how depressing the current situation may seem, there is at least some good news: effective, inexpensive treatments exist although they are not widely used.

The three extant meta-analyses of treatment effectiveness (Finney & Monahan, 1996; Holder et al., 1991; Miller et al., 1995), all of which reviewed the available controlled studies of treatment effectiveness (roughly 200 in all three analyses), reached similar conclusions about which treatments work and which ones don't. Interestingly, all three treatments that showed good evidence of efficacy in all three meta-analyses were cognitive/behavioral approaches. Here are brief descriptions of these treatments, as well as descriptions of two additional well-supported therapies.

Community Reinforcement Approach (CRA). This is the single best-supported therapy. It's a moderately low-cost form of outpatient treatment; it was devised and first tested over a quarter-century ago; the first controlled study of its efficacy appeared in 1973; every study of its efficacy six, so far has shown extremely encouraging results; and it is not in use as a regular form of treatment at a single one of the 15,000 alcoholism treatment centers in the United States (though, at this writing, there is an ongoing efficacy study in Albuquerque).

The basic premise of the community reinforcement approach most often a one-on-one therapy, although it can be used in group settings is that alcohol abuse does not occur in a vacuum, that it is highly influenced by marital, family, social, and economic factors (the exact opposite of the AA/12-step premise that alcoholism [or drug addiction] is a purely individual disease that exists independently of social conditions). CRA attempts to help the client improve his or her life in all of these areas, in addition to giving up drinking (or using drugs). Thus, a CRA program will typically include the following components: 1) communications skills training; 2) problem-solving training; 3) help finding employment; 4) social counseling (that is, encouraging the client to develop nondrinking relationships); 5) recreational counseling (that is, encouraging or helping the client to find rewarding nondrinking activities); and 6) marital therapy. Other treatment components are sometimes used for example, disulfiram (Antabuse), drink-refusal training, or rewarding the client materially for abstinence but these six components form the core of the CRA approach.

Social Skills Training. This form of group therapy is another very well supported approach. The basic premise of social skills training is that alcohol/drug-abuse clients lack basic skills in dealing with work, family, other interpersonal relationships, and their own emotions. Thus, they benefit from training in areas such as communications skills (including giving and receiving criticism, listening and conversational skills), conflict resolution, drink-refusal, assertiveness, and expressing feelings.

Behavioral Marital/Family Therapy. This is essentially standard marital counseling, but with an emphasis on altering behaviors related to drinking, such as helping the non-alcohol-abusing spouse abandon futile nagging about drinking and instead begin to reward sober behavior. The remainder of the therapy involves couples counseling, the goal being to repair alcohol abuse-caused damage to the relationship, as well as dealing with non-alcohol-caused problems.

There are a number of other forms of therapy with reasonably good indications of efficacy, but the above are highly supported across all three meta-analyses cited. Other therapies with indications of efficacy include behavior contracting, brief intervention, disulfiram, and motivational enhancement.

Brief Intervention / Motivational Enhancement. Brief intervention was the highest scoring (i.e., most effective-rated) treatment in the Miller et al. (1995) meta-analysis, while motivational enhancement was ranked as the third most effective form of treatment. At the same time, they were also among the most inexpensive therapies, with only "self-help manual" being lower in cost.

Brief intervention shares elements with motivational enhancement (one of the treatments tested by Project MATCH) in that the patient and the therapist create a mutually agreed-upon goal. The first step in creating this goal is often an objective assessment of the person's drinking habits, or a comparison of his or her drinking levels with community standards, or else a comparison with optimum levels of drinking for health purposes. In brief intervention, the goal is usually reduced drinking; in motivational enhancement, it is either reduced drinking or total abstinence. The key is to allow patients to select a goal that is consistent with their own values and that they thus "own" as an expression of their genuine desires.

In a brief-intervention session, the health-care worker simply sums up the goal: "So, we agree you will reduce your drinking from 42 drinks a week to 20, no more than four on a given night." Motivational enhancement is a bit more subtle: the therapist nudges without directing, by responding to the patient's own values and desire for change. The dialogue in a motivational-enhancement session might go like this:

THERAPIST: What is most important to you?
PATIENT: Getting ahead in life. Getting a mate.
T. What kind of job would you like? What training would that take?
P. [Describes.]
T. Describe the kind of mate you want. How would you have to act, where would you have to go, to meet and deal with a person like that? P. [Describes.]
T. How are you doing at achieving this?
P. Not very well.
T. What leads to these problems?
P. When I drink, I can't concentrate on work. Drinking turns off the kind of person I want to go out with.
T. Can you think of any way to improve your chances of succeeding at work or with that kind of mate?

Here we see that the goal of therapy is to draw the connection between what people genuinely want their own goals and the institution of helpful behaviors, or the elimination of behaviors that interfere with achieving their goals. In brief intervention, in addition, drinkers know that they and their helper will be regularly assessing progress toward the agreed-upon goals in systematic but nonjudgmental meetings a central element, as we saw in the highly praised results of Project MATCH. (To examine these nondirective approaches, see Horvath, Sex, Drugs, Gambling and Chocolate: A Workbook for Overcoming Addictions; Miller & Rollnick, Motivational Interviewing: Preparing People to Change Addictive Behavior; and Peele et al., The Truth About Addiction and Recovery.)


One finding shines through all of these complicated results: the most beneficial outcomes to drug and alcohol problems occur with minimal treatments, or else depend primarily on the characteristics and motivation of the patient, rather than on any specific treatment. Not only is 12-step treatment unjustified, but hardly any formal treatment seems to be necessary if drug and alcohol abusers become properly involved in defining and directing their own efforts toward change. That is, they'll likely succeed if they want to change, and if they have support in their efforts to change.