The Controlled Drinking
Hersey, Psy.D., April 2001 - Email:
In a 1995 editorial in the journal Addiction, Mark and Linda Sobell, two researchers closely associated with the controlled drinking controversy, wrote that "controlled drinking approaches no longer seem to arouse intense debate." (Sobell & Sobell, 1995, p. 1149). They credited this change to several developments in the field. Epidemiological research has identified a large population of problem drinkers whose drinking severity falls below the level of full-blown alcohol dependence. The preponderance of evidence suggests that while severely dependent drinkers do better with abstinence approaches, the Sobells stated, there is increasing acceptance of the proposition that those with less severe problems often improve by moderating their drinking. Sobell and Sobell (1995, p. 1150) also alluded to the potential utility of harm reduction goals for severely dependent drinkers, making the interesting point that "some of the early controlled drinking research [including their own controversial study, presumably] might be considered as demonstrations of the promise of harm reduction strategies for severely dependent drinkers."
The Sobells' implication - that the focus on non-dependent problem drinkers and on harm reduction has taken the teeth out of the controlled drinking controversy - may indeed be true within the community of those who do scientific research on the treatment of alcohol problems. But this move away from acrimony has not yet extended to the more extended treatment community and the public at large. In fact, events that took place during the summer of 2000 belie the notion that controversy around controlled drinking is dead.
Audrey Kishline, the founder of Moderation Management (MM), a non-abstinence-oriented self-help group for individuals whose alcohol problems stop short of dependence, killed two people in a head-on vehicular collision. At the time of the accident, which took place in March of 2000 but which didn't get widespread publicity until June, Kishline's blood alcohol content was measured at .26. Kishline had quit MM two months before the accident, posting a message on the organization's listserv, stating that she had decided to change her goal to abstinence. Despite the fact that Kishline was no longer following the MM program, and was in fact attending Alcoholics Anonymous (AA), some commentators used the incident to condemn non-abstinence goals. The National Council on Alcoholism and Drug Dependence (NCADD) sent out a press release stating that the incident "provides a harsh lesson for all of society, especially those individuals who collude with the media to continually question abstinence-based treatment for problems related to alcohol and other drugs." (NCADD, 2000)
In July, New York magazine published an article on the Smithers Alcoholism Treatment Center at St. Luke's/Roosevelt hospital in New York City. The piece stated that the Center, once a Minnesota Model mainstay, was offering a menu of treatment options, and wasn't turning away patients who weren't ready to sign on at the outset for an AA-style recovery program. The article quoted Alexander DeLuca, the chief and medical director, as saying: "We do find that people who go to twelve-step meetings do better. But it doesn't work for some and I'm not going to tell them, 'Come back after you've suffered some more and are ready to do it our way.'" (Szalavitz, 2000, p. 11) The article, provocatively titled "Drink Your Medicine," also mentioned that MM held a weekly meeting on the Smithers premises. Again, there was a heated response, this time from the Christopher D. Smithers Foundation. The Smithers Foundation took out a full-page ad in the July 9th issue of The New York Times admonishing Smithers' leaders for not having "learned their ABCS: A=Alcoholism is a disease; B=Booze has no place in its treatment; C=Controlled Drinking does not work." Further, the Foundation's president, Adele Smithers-Fornaci, sent out a letter saying that this move "will most certainly spawn new moderation management treatment models around the country. Alcoholics will be enticed to go into treatment without having to give up drinking! … The tragedy is that lives will be destroyed and people will die in the process." (Smithers-Fornaci, 2000).
The events described above prompted debate in the popular media. As Szalavitz (2001, p. 76) writes, "the press conflated the two stories, implying that Smithers embraces a pro-drinking program started and now renounced by a drunken killer." Shortly afterwards, St. Luke's/Roosevelt hospital, already embroiled in a costly lawsuit with Mrs. Smithers-Fornaci, fired DeLuca. A counselor, who had appeared on ABC's 20/20 discussing his doubts about 12-step models and talking on-air about his own experiences with a moderation-style recovery, lost his job at a traditional treatment center. (Snyderman, 2000) For the most part, the media coverage was quick to generalize from Kishline's tragedy to a condemnation of moderation: An edition of Larry King Live, for instance, featured relatives of the people she killed and celebrities in recovery opining on the incident: The focus was on personal stories; no historical or scientific context was provided. A casual observer might well have come away thinking that any alternatives to abstinence were new, untested and dangerous.
And yet the debate
about whether abstinence is necessary for people with
drinking problems has been going on for four decades.
The paper that follows represents this writer's effort
to better understand the controversy's heat and
longevity, by tracing it from J.P. Davies' 1962 paper
through to the present. I will conclude my historical
review with a discussion of some important issues that
confront a contemporary observer of the controlled
The necessity of abstinence: Alcoholics Anonymous and the disease concept
To start, it is necessary to briefly consider the paradigm that is called into question when it is suggested that individuals with alcohol problems might improve by reducing rather than eliminating their alcohol use. This paradigm is the disease concept of alcoholism, as elaborated by Alcoholics Anonymous (AA), and, in slightly different form, by Jellinek. AA conceptualizes alcoholism as a progressive, irreversible disease, characterized by "loss of control" over alcohol. While abstinence is believed to be the only way to arrest the inevitable downward course of the disease, many alcoholics resist acknowledging this, and maintain that, if they could just get it right, they could drink in a less-than-catastrophic fashion. A frequently quoted passage from AA's "big book" reads:
Most of us have been unwilling to admit we were real alcoholics...Therefore, it is not surprising that our drinking careers have been characterized by countless vain attempts to prove we could drink like other people. The idea that somehow, someday, he will control and enjoy his drinking is the great obsession of every abnormal drinker. The persistence of this illusion is astonishing. Many pursue it into the gates of insanity or death. (Alcoholics Anonymous, 1976, p. 30)
In the many iterations of the controlled drinking controversy, disease-model adherents have argued that the dissemination of findings supporting the possibility of controlled drinking would feed this "illusion" in the minds of dependent drinkers. The alcoholic's predilection for "denial" might cause him/her to use reports of others' non-problematic drinking to infer that he or she could do the same. Alcoholics might put off identifying their disease, might postpone seeking abstinence or might give it up, with catastrophic consequences.
Although later experimental researchers have called many of the central premises of the disease concept into question (see Marlatt, 1983, Miller, 1986, for reviews), it has dominated (and, to a large extent, continues to dominate) American views of alcoholism and treatment.
Although AA is a
self-help group with no official ties to hospitals or
other treatment facilities, the organization's presence
in hospitals began as early as 1939, when Dr. Bob, one
of the organization's founders, introduced a number of
AA systems in St. Thomas Hospital in Akron, Ohio.
(White, 1998). Over the decades that followed, AA
members' involvement in treatment provision has grown.
The move towards the so-called Minnesota Model brought
many recovering counselors onto the payrolls of alcohol
treatment facilities, where treatment was often an
inpatient stay aimed at educating patients about the
disease model, facilitating their self-identification as
alcoholics, and socializing them to AA. Outreach
efforts, such as that of Marty Mann and the National
Committee for Education on Alcohol, advanced the notion
of alcoholism as a disease in the public arena.
Physicians looked to Jellinek's writing as a conceptual
foundation for their work with problem drinkers. In sum,
the disease model, with its absolute rejection of
recovery goals other than abstinence, has deep roots in
The Davies paper: "Alcohol addicts" who returned to controlled drinking
While scattered reports of controlled drinking outcomes had occasionally appeared in the scientific literature before 1962, most commentators (Roizen, 1987, Rosenberg, 1993) date the beginning of the controlled drinking controversy to the publication that year of a paper entitled "Normal Drinking in Recovered Alcohol Addicts." In this paper, D.L. Davies, a British psychiatrist, reports that, in the course of long-term follow-up of patients treated for "alcohol addiction" at Maudsley Hospital in London, 7 of the 93 patients investigated "have subsequently been able to drink normally for periods of 7 to 11 years after discharge from the hospital." (Davies, 1962, p. 94).
The seven male patients, whom Davies argued were addicts because they had experienced "loss of control" over their drinking, all received abstinence-oriented treatment that included individual therapy, disulfiram, and social work with significant others. All had a brief period of abstinence following discharge. In the subsequent period, according to data from both patients and collaterals, the patients drank in a so-called normal manner: "This means that their use of alcohol has never gone beyond the limits regarded as permissible in the cultural group from which they are drawn." (Davies, 1962, p. 95). They were found to have never "been drunk," and to have demonstrated improved functioning in the domains of employment and family relationships. Davies presented case summaries for all the cases, and proceeded to state, "We may conclude, on the basis of the present series and the other reports cited in the literature, that some alcohol addicts do return to normal drinking." (Davies, 1962, p. 102)
Davies went on to note that this finding challenged established notions about alcohol addiction. The established view - that addiction is inevitably progressive, that "loss of control becomes irreversible" (Davies, 1962, p. 102) and that the only route to recovery is via abstinence - becomes open to question in light of these case studies. Despite this, he also states, "It is not to be denied that the majority of alcohol addicts are incapable of achieving 'normal drinking.' All patients should be told to aim at total abstinence." (Davies, 1962, p. 103)
The publication of this paper evoked what Rosenberg (1993, p. 129) has called "a storm of comment." However, Roizen (1987, p. 249) notes that "what most characterized this episode's debate was its scholarly content and tone." Responses, published in subsequent issues of the Quarterly Journal of Studies on Alcohol, were authored by physicians and psychiatrists; Davies' article generated little to no comment in the popular press.
Indeed, a review of 17 comments published in 1963 reveals a group of professionals who are struggling to integrate the discrepant data provided by Davies into their conceptual orientation towards their work treating individuals with alcohol problems. A number of authors raised questions about diagnosis: Perhaps, given the observed outcomes, the seven normal drinking patients were never addicts at all. (Williams, 1963; Block, 1963; Lemere, 1963; Smith, 1963) A number of authors noted that Davies findings were at odds with their substantial clinical experience. (Williams, 1963; Zwerling, 1963) A typical comment along these lines reads: "My own practice covers many hundreds of alcoholics, and though I have never been in the position to do a follow-up, I do not know of a single patient of mine who has been able to resume normal drinking." (Fox, 1963, p. 117)
The relationship between severity of dependence and ability to moderate, a major theme in later investigations of controlled drinking, is hypothesized. Smith (1963, p. 323) writes: "One possible explanation of Doctor Davies' findings derive from the fact that there are undoubtedly degrees of alcoholism. All patients are not equally ill."
A point raised by many commentators was the purported danger that Davies' paper posed. Alcoholics might be encouraged, some authors feared, to postpone, deny or give up abstinence. Bell (1963, p. 322) noted, "For every alcohol addict who may succeed in reestablishing a pattern of controlled drinking, perhaps a dozen will kill themselves trying." Armstrong (1963, p. 119) stated, "It seems very difficult for a person whose whole future depends on what he does about alcohol to weigh realistically the odds for him in attempting to emulate those who are the special subjects of Doctor Davies' paper." Tiebout (1963), a psychiatrist with longstanding ties to AA, noted that he had discussed the Davies article with sober alcoholics. Interestingly, these individuals were unanimous in their sense that they themselves would not be able to drink in a controlled manner, but expressed concern that others would be tempted. Selzer (1963, p. 113) noted that the possibility of controlled drinking is "anathema" to therapists working in the field, who fear that the acceptance of anything other than abstinence as a treatment goal would "wreak havoc" on existing treatment programs. When, in the course of his earlier research, he had also found evidence of alcoholics who had returned to social drinking, the agency funding his research "virtually ... order[ed]" him to "omit these 'embarrassing' findings." Given the purported dangers that these findings would lure alcoholics to forego abstinence, Bell (1963, p. 322) suggested that "clinical studies of this kind should be carried on with a minimum of publicity. Otherwise, the health and safety of a great many people could be seriously jeopardized."
Myerson (1963, p. 325) stated that he found it refreshing to witness a fixed idea challenged, citing the tendency of those who deal with alcoholics "to adhere to a rigid doctrine which they feel should be applied to an entire alcoholic population." There is also mention of the extent to which the lack of formal follow-up may serve to bias clinicians' thinking. Brunner-Orne notes that clinicians may be less likely to hear from those patients who have succeeded in controlling their drinking. In a reply to the various commentators, Davies (1963, p. 332) made much the same point, and called for more research:
Perhaps now that this matter has been brought out for frank discussion there will be further reports from those whose follow-up arrangements permit of review not just of those who, because they get worse or achieve full sobriety, may continue to remain in touch, but also of those who may have achieved completed recovery and on that account have passed beyond the ken of their therapists.
The years after Davies
In the 1960s and 1970s, psychologists began to subject the premises of the disease model to scientific scrutiny, and to use experimental methods to assess treatment outcome. As Marlatt (1983) and Miller (1986) review, a number of experimenters tested the premise that alcohol inevitably precipitates loss-of-control drinking, and found that alcoholics' beliefs about whether or not they are consuming alcohol affect consumption. Also, varying schedules of reinforcement produced different drinking patterns, arguing against the notion that all alcoholics experience total loss of control.
Marlatt (1983) recalls that this period was one of "adventurous excitement" as behaviorally oriented psychologists began to apply principles of learning theory to a wide range of severe disorders. Included in these efforts were protocols designed to train dependent drinkers to drink in a controlled fashion. Lovibond and Caddy, two Australian psychologists, published a promising report on this as early as 1970.
To read Sobell and Sobell's account of their experiments at Patton State Hospital in the 1970s is also to get a strong sense of the ambition and scope of this behavioral work. The treatment unit included a simulated bar and cocktail lounge, set up so subjects could both be videotaped while drinking, and also equipped with electric shock equipment for aversive conditioning: "The simulated bar environment … reflected an attempt to structure the research environment to promote increased generalization of treatment effects to the subjects' usual drinking environment." (Sobell and Sobell, 1978, p. 50). While the Sobells' experimental work that was to generate such controversy began in 1970, I will hold off on reporting it because the uproar it evoked did not take place until the early 1980s.
The Rand Report
In the 1970s, the National Institute on Alcoholism and Alcohol Abuse (NIAAA) established a network of treatment centers around the United States, which included a monitoring system to collect data on clients served (Polich, Armor and Braiker, 1981). The Rand Corporation assumed responsibility for evaluating the efficacy of the treatment offered, and the so-called Rand Report, published in 1976, looked at 18-month follow-up data on a sample of patients treated at 44 treatment centers. In summarizing conclusions, the authors wrote:
[I]t is important to stress that the improved clients include only a relatively small number who are long-term abstainers…The majority of improved clients are either drinking moderate amounts of alcohol - but at levels far below what could be described as alcoholic drinking - or engaging in alternating periods of drinking and abstention." (Armor, Polich and Stambul, 1976, p. v)
Specifically, the authors found that 22 percent of treated individuals were "normal drinkers" at 18-month follow-up, with low-to-moderate levels of drinking and little or no symptomatology.
The publication of this report was the occasion for renewed debate and controversy. The National Council on Alcoholism denounced the report on the morning it was released, describing it as "dangerous." (Peele, 1983). According to Roizen (1987), the debate extended beyond the scientific literature to the popular press, with most accounts and editorials emphasizing the importance of abstinence. While the report was criticized by some on methodological grounds, another major focus of criticism argued that "the research was 'impersonal' or 'statistical' or that a wide gap separated the Rand authors from actual alcoholism patients or that the authors lacked personal experience and contact with the field." (Roizen, 1987, p. 262) In other words, the legitimacy of using a scientific approach was questioned, and the potential dangerousness of such questionably derived "knowledge" was argued to be a reason that this knowledge should not be disseminated.
A later report, which included data on the 4-year follow-up of treated patients and which attempted to address methodological criticisms of the earlier report, showed that a similar percentage of patients were demonstrating non-problematic drinking (Polich et al., 1981), although the authors cautioned against the conclusion that the same patients who were stable at 18 months were stable at 4 years. What became evident over the course of the longer follow-up was the extent to which individual patients' drinking statuses fluctuated: "When we examined longer time periods and multiple points in time, we found a great deal of change in individual status, with some persons continuing to improve, some persons deteriorating, and most moving back and forth between relatively improved and unimproved statuses." (Polich et al., 1981, p. 214)
As was the case with the Davies paper, these findings occurred following treatment that was focused on abstinence. However, the very fact that the Rand authors were willing to recognize success in the presence of any drinking at all was controversial. Roizen notes that while the Davies controversy focused on whether or not "normal drinking" was ever an outcome for addicted drinkers, "Rand authors argued that particularly long-term abstinence was too infrequent to make it the sole focus and measure of successful treatment." (Roizen, 1987, p.262)
So where exactly are the battle lines drawn here? The supposed proponents of controlled drinking were not saying that it should be advocated or taught, but that unproblematic drinking was observed in the aftermath of abstinence-oriented treatment, and that individuals could be judged to be improved without being abstinent. Opponents of controlled drinking appear to have focused on the potential danger of these findings for alcoholics. Their tactics included efforts to discredit the Rand authors' methodology, but also to attempt to minimize putative danger to alcoholics by arguing their pro-abstinence case in the public arena.
In the early 1970s, psychologists Mark and Linda Sobell set out to research a form of "individualized behavior therapy" for alcoholism. One treatment module tested was aimed at training alcohol-dependent subjects to drink in a "controlled" fashion (Sobell and Sobell, 1973, 1978). Subjects were 70 male patients, voluntarily admitted to Patton State Hospital in California, who were classified as meeting criteria for Jellinek's gamma-type alcoholics ("loss of control" drinkers). After subjects were accepted for participation in the study, they were assigned to either a controlled drinking (CD) or an abstinence-goal condition. This part of the study did not employ random assignment; rather the assignment was made by the research staff, based on both the patient's stated wishes and goals, and characteristics of the subject, his drinking history, and the stability of his environment. After this initial assignment, subjects were then randomly assigned to a behavioral treatment condition, or to a control condition of treatment as usual (which was, of course, abstinence-oriented). Both experimental groups (CD and abstinence-goal) received 17 sessions of behavioral treatment (including training in problem solving and aversive conditioning with electrical shocks), but the CD subjects were also trained in drinking skills oriented towards nonproblematic drinking. Follow-up was extensive, and collateral sources were used in addition to patient self-report. While the authors collected a wide range of outcome data, they used the number of "days functioning well" as a primary outcome measure. Individuals in the CD-experimental (CD-E) condition had significantly more "days functioning well" during a two year follow-up period than their counterparts in the CD-control (i.e. treatment as usual aimed at abstinence) condition. (Sobell & Sobell, 1978, 1973).
In the Davies and Rand reports, "controlled drinking" was used to describe a non-problematic level of drinking, but this was not an outcome that had been sought by treatment. By contrast, the Sobells' work involved the transmission of specific skills and techniques to individuals with serious and enduring alcohol problems. (This study does not get at the interesting question of whether explicitly CD focused treatment is a valuable addition to a behavioral program: The lack of random assignment to CD or abstinence goals precludes us from drawing conclusions about relative efficacy.)
In 1982, Pendery, Maltzman and West published an article in the journal Science entitled "Controlled Drinking by Alcoholics? New Findings and a Reevaluation of a Major Affirmative Study." This report, based on a ten year follow-up with subjects of the Sobell and Sobell study, states that "a review of the evidence, including official records and new interviews, reveals that most subjects trained to do controlled drinking failed from the outset to drink safely." (p. 169) Ten years out, only one subject from the CD-E condition was maintaining a pattern of controlled drinking. Eight subjects were found to be drinking excessively, six were abstinent, one was lost to followup, and four were dead.
Certainly, in reading Pendery et al.'s article, one gets the impression of subjects who are doing very poorly indeed. However, the Pendery report is severely compromised on several scores, most importantly by the fact that it provides data for the experimental group but not the control group. These authors attempt to justify this choice in a statement that seems to clearly demonstrate their bias: "we are addressing the question of whether controlled drinking is itself a desirable treatment goal, not the question of whether the patients directed towards that goal fared better or worse than a control group that all agree fared badly." (Pendery et al., 1982, 172-173)
Although the Science paper took relatively measured tones in presenting what it stated was discrepant data, outside the rarified realm of the scientific journal, the authors took a less neutral stance. They circulated a more inflammatory paper to the research community (Roizen, 1987). Maltzman was quoted in The New York Times as stating, "Beyond any reasonable doubt it's fraud." (Boffey, 1982, quoted in Marlatt, 1983, p. 1098) Marlatt (1983) also describes a 1983 edition of 60 Minutes which criticized the Sobells, and which included footage of correspondent Harry Reasoner visiting the grave of one of the patients in the controlled drinking condition. Marlatt also notes that one of the patients from the CD condition formed an organization called "the Alcoholism Truth Committee," aimed at disseminating the "truth" about the Sobells' study by attempting to have descriptions of their work omitted from textbooks and elsewhere.
Several investigations of the integrity of the Sobells' work followed. The Sobells asked their employers, the Addiction Research Foundation, to appoint a committee to investigate their research. Because some of the Sobells' research was grant-funded, a subcommittee of the Committee on Science and Technology of the House of Representatives, and a federal panel also reviewed the Sobells' data. All these investigations exonerated the Sobells (Roizen, 1987). According to Marlatt et al. (1993), while there was extensive media coverage critical of the Sobells, there was little media coverage of the exonerating verdicts, leaving the public with the impression that the Sobells' work had been not only flawed but fraudulent, and that controlled drinking was a misguided and potentially deadly treatment goal.
The Rand and Sobell and Sobell controversies had a chilling effect on psychologists and researchers. In 1984, Peele wrote (p. 1342): "Today no clinician in the United States publicly speaks about the option of controlled drinking for the alcoholic."
Two years later, Miller (1986, p. 117) wrote: "American professionals who advocate any alternative to abstinence are likely to be (and have been) attacked as naive fools, misguided intellectuals sadly misinformed about the 'reality' of alcoholism, unwitting murderers, or perhaps themselves alcoholics denying their own disease." Miller (1986, p. 118) also contends that U.S. researchers have found it hard to obtain funding for controlled drinking studies, "and the controversy regarding the Sobell and Sobell study (Pendery et al., 1982) is likely to discourage future U.S. research on this topic for some time to come."
But if researchers have moved away from talking about controlled drinking as a goal of choice for alcohol dependent clients, two semantic and conceptual shifts - alluded to in the 1995 Sobell and Sobell editorial discussed above - have permitted continued investigations of treatments that are not singularly focused on abstinence. The concept of "harm reduction" has been evoked to suggest that, given that some severely dependent individuals might be unable or unwilling to abstain, it was appropriate to try to minimize the harm caused by their continued drinking. Secondly, the increased awareness of a large population of problem drinkers whose alcohol use does not meet criteria for dependence has led to a focus on interventions aimed at reduction rather than elimination of alcohol use. With this conceptual reframing comes a terminological shift as various authors made the choice to move towards less contentious language. In 1987, Marlatt (p. 168) noted that use of the term "controlled drinking" "is a red flag that sends the bull charging in the direction of behaviorists." His suggestion for a replacement is "moderation training."
In many ways, these two shifts represent a tidy compromise, in that they allow for deviations from an uncompromising abstinence goal, while no longer challenging the disease model in such a fundamental way. In the case of harm reduction, abstinence is held out as the gold standard, and continued drinking for dependent drinkers is identified as a problematic (if frequent) outcome. As Marlatt et al. (1993, p. 465) wrote: "The goal of harm-reduction methods is to facilitate movement along a continuum from greater to lesser harmful effects of drug use. Although abstinence is considered an anchor point of minimal harm, any incremental movement toward reduced harm is encouraged and supported."
The second approach can be said to target individuals who are not "alcoholic." Of course, things are not really so simple, some disease-model proponents might argue. Are the subjects of these interventions really a different population than the alcoholics, or are they people with alcoholic tendencies whose "disease" has not yet progressed?
Those concerned with engaging problem drinkers in treatment also argue that offering goals other than abstinence may attract a wider audience: "Offering controlled drinking alternatives to the general public may act as a motivating push to get people 'in the door,' a low-threshold strategy that is consistent with the principles of harm reduction." (Marlatt et al., 1993, p. 483)
Although not universally successful in defusing the controversy, this re-framing might be understood as a diplomatic solution. Each side could declare itself victorious. Writing in 1995, one commentator postulated: "[A]pparently - little moved by the 'great debate' - both sides continue with their initial preferences: the 'American establishment,' AA and clinicians predominantly with the abstinence approaches, whereas psychologists, researchers and sociologists often regard controlled drinking as a feasible and often preferred alternative." (Glatt, 1995, p. 1157)
The severity of dependence issue
Sobell and Sobell (1995) noted that low severity of dependence is an important predictor of an individual's ability to moderate successfully. In their 1981 literature review, Heather and Robertson also found low severity to be correlated with controlled drinking outcomes, although they note that some of the studies they reviewed only looked at men, thus limiting the generalizability of their findings. Miller (1983, p. 77) observed that in the Rand study, "patients with high alcohol dependence were found to be less likely to relapse from abstinence than from nonproblem drinking, suggesting that for this population abstinence was the more stable outcome." Rosenberg (1993, p. 132) reviewed a number of more recent studies and reported finding general (although not universal) support for the severity hypothesis. He went on to state that:
the nature of the relationship between severity and CD has not yet been established. One possibility is that the likelihood of CD decreases monotonically as severity of dependence increases, and at some point severity is so great that the probability of CD is zero. Alternatively, although CD generally declines as severity increases, there may be plateaus in severity in which changes in level of severity do not matter. Also, even at the highest levels of severity, perhaps some alcoholics are able to control their drinking as a result of other factors. Finally, a significant association between the two variables does not necessarily mean that lower severity is the cause of CD.
Sobell and Sobell (1995, p. 1150) also urged caution in interpreting causality from these results: "[A]lthough it is tempting to view dependence severity as the critical determinant of whether a moderation recovery is attainable, it is possible that this relationship is an epiphenomenon to other life circumstances often associated with severe dependence (e.g. lack of social support, poor vocational history)."
Sobell and Sobell (1995) made the interesting point that this association between severity level and outcome seems to hold true, regardless of what is advocated in treatment. Sanchez-Craig et al. (1984) randomly assigned low-dependence drinkers to treatment aimed at either abstinence or controlled drinking. At two year follow-up, the two conditions were quite similar, and most successful outcomes involved moderate drinking. In a study in which severely dependent drinkers were assigned to treatment with either abstinence or CD goals, at 5-6 year follow-up the groups were also similar, with most successes involving abstinence. (Rychyarik et al, 1987).
The real world implications of the above findings are far from clear. While individuals with less severe alcohol problems appear to have more frequent CD outcomes, does it follow logically that CD training should not be used with dependent populations? Might CD techniques play a role in harm reduction? In 1987, Peele stated that while past research "found greater benefits for problem drinkers who were less severely dependent on alcohol," at the same time, "no comparative study had shown moderation training to be less effective than abstinence as a treatment for any group of alcoholics." (Peele, 1987, p. 175) Heather (1995) argued that some studies have shown that severely dependent individuals can sustain non-problematic drinking, and that there may be applications for CD-focused interventions with this population; much of this work is currently being done outside the United States.
Peele (1992) argued that the consensual move away from CD treatments for more seriously dependent drinkers resulted from political pressures rather than from the weight of unequivocal empirical evidence. In responding to Peele, Miller (1992, p. 80) argued that data linking severity to treatment outcome do "provide for clinicians the basis for a probabilistic argument in favor of abstinence, as severity increases." That said, he affirmed that he favors a de-escalation of the CD controversy: "There is little to be gained by continually exacerbating points of disagreement. The effect is only to deepen already wide chasms among significant factions, all of whom are trying to alleviate alcohol problems." (Miller, 1992, p. 81)
The importance of what the patient believes
In his 1993 review of the literature on predictors of controlled drinking, Rosenberg observed that individuals' beliefs about the feasibility of CD is a potentially useful predictor of their ability to moderate. While the nature of individual beliefs were operationalized in different ways by different investigators, the majority of studies that Rosenberg reviewed supported the so-called persuasion hypothesis. Rosenberg noted a number of questions that grew out of these findings: What is the source of drinkers' beliefs? To what extent are beliefs shaped by pre-existing notions, what the drinkers are told in treatment, and/or experiences after treatment as they attempt to achieve their goals? And how stable are these beliefs? Typically, these beliefs are measured once in the course of most studies, and then used as a predictor of behavior months after the measurement (Rosenberg, 1993). In fact, these measures may change frequently. For instance, Ojehagen and Berglund (1989) reported on a Swedish treatment program in which participants chose their own treatment goals (abstinence or CD) and were allowed to change these goals every three months. Forty-four percent of patients changed goals at least once during the treatment program, a finding suggesting that beliefs about what is both possible and helpful changed over time, presumably as a result of experience.
Interestingly, the implications of this association between belief and behavior can be spun in different ways. A belief in the necessity of abstinence may help an individual stay sober, which is obviously a desirable outcome. However, a belief in the efficacy of CD may serve harm-reduction ends, if a dependent drinker believes that he or she can cut down intake,. As Peele (2000, p. 43) writes, "[T]he very subjective elements that American alcoholism treatment derides as 'denial' can improve the chances of recovery: It is easier to achieve what you believe."
Despite the aforementioned decades of research and debate, when Audrey Kishline sought help for her problematic drinking in the late 1980s, it took her years to learn that there were any professionally sanctioned alternatives to abstinence. As she describes it in her 1994 book, she consulted 30 to 40 professionals, many of whom steered her towards AA, and emphasized that she would have to attend meetings for the rest of her life. When she began to explore moderation options, she states she was "amaze[d]" to find the extent to which these approaches had been explored by addiction professionals and put into practice in other countries. She writes:
The first major revelation that I came across was that many experts in the alcohol studies field do not believe that alcohol abuse is a disease. From my previous experience with traditional treatment, I had been under the impression that the disease model of alcohol abuse represented a biological and medical fact, proven beyond a shadow of a doubt. I was amazed to find out that the disease theory was just that: a theory - one that has been highly criticized, and discarded, by many researchers in the field. (Kishline, 1994, p. 12)
Kishline's experience may be representative: Despite many encounters with the treatment community, she did not learn that there are multiple ways to conceptualize substance abuse problems, and she did not learn that there are ways to recover that do not necessitate AA.
In founding MM, Kishline integrated many behavioral techniques into a self-help format: "The purpose of Moderation Management is to provide a supportive environment in which people who have made the healthy decision to reduce their drinking can come together to help each other change. That's it. It is very simple and straightforward, and I admit that MM stole it from the forerunner of the mutual help movement, AA." (Kishline, 1994, p. 25)
The program explicitly states that it is not for dependent drinkers. It advocates a month-long abstinence period before the institution of a program of moderate drinking. It offers a mechanism by which problem drinkers can try to cut down; in theory, failure at this effort suggests the advisability of abstinence. The MM movement has garnered attention as a grassroots movement reaching out to and providing free support and technical assistance to the large population of non-dependent drinkers. Articles on the organization have appeared in Time in 1995, and in U.S. News and World Report in 1997; the organization has also been featured on television shows with large audiences, such as Good Morning America and the Oprah Winfrey Show.
To its supporters,
MM represent a self-help-style embodiment of a promising
approach to drinking problems. To its detractors, it
represents a mechanism by which alcoholics can
perpetuate their denial.. Although a number of academic
researchers have provided advice and support to MM, the
movement can be understood as a form of CD that has
moved out of the research domain into the general public
Is public opinion becoming more open to alternatives?
absence of survey research, we don't actually know the
current state of public opinion on moderation approaches
to problem drinking. We can speculate that Kishline's
accident and the press coverage it received has
convinced some that moderation is dangerous. On the
other hand, increased publicity may spawn increased
interest and debate about alternative conceptualizations
of alcohol problems and routes to recovery. An edition
of 20/20 broadcast in June of 2000 featured interviews
with a number of proponents of moderation approaches,
and highlighted the fact that, in many other countries,
treatment often involves the teaching of moderation
skills. Szalavitz, the author of the New York article
about Smithers, wrote a follow-up piece for Brill's
Content, a magazine widely read by journalists,
detailing the conflation of the Kishline and the
Smithers stories; her piece may have educated writers
and editors about the scientific and political backdrop
to the news events. And the recent publication of the
book Sober for Good (Fletcher, 2001) communicates in
hopeful, accessible language that there are many ways
for former problem drinkers to deal with their drinking.
A significant proportion of those she interviewed got
sober without AA, and a smaller proportion have made
major improvements in their drinking and their lives
without giving up alcohol altogether. It seems quite
possible that this book will reach and educate a broad
Summary and open questions
The various iterations of the controlled drinking controversy can be summarized briefly. Davies observed data that called certain premises of the disease model into question, leading some in the field to resort to various semantic twists to minimize or deny the import of his findings. When the Rand authors made the case that nonproblematic drinking was widespread in its outcome studies, disease model adherents attempted to minimize the impact of these findings on public opinion. Rand opponents echoed Davies commentators in voicing the fear that the acknowledgment that some alcoholics can learn to drink in a nonproblematic fashion might lure other alcoholics to postpone self-identifying or to reject their commitment to abstinence. When the Sobells published results showing that a behavioral treatment which included CD techniques produced better outcomes than treatment as usual in a severely dependent population, another team of researchers attacked their reputations in the course of arguing that CD was no panacea for these very sick folk. Prodded in part by political realities and in part by evidence that CD training doesn't seem to significantly boost the efficacy of treatment, researchers have backed off from advocating controlled drinking treatments for alcoholics. However, strategies like harm reduction and moderation training for non-dependent drinkers have kept residual CD strategies alive. It is now common practice in treatment studies to acknowledge that abstinence is not the only successful treatment outcome, and that reduced drinking in fact constitutes improvement (Peele, 2000). Despite these compromise positions, the issue remains a hot-button topic in public discourse, with Audrey Kishline's recent tragedy being touted by opponents as a (supposed) reminder that the mere existence of moderation approaches can support and prolong alcoholics' denial.
A review of the history of this debate provides few incontrovertible answers. Thus, it seems appropriate to conclude this review by highlighting some of the important questions that remain salient 40 years into the debate:
(1) What does the research really say? Several enduring themes, which came up again and again in the body of literature considered, seem fairly well supported: Pure, uninterrupted abstinence is rarer than we would like in treated dependent drinkers, and some problem drinkers do seem able to reduce their drinking, with an accompanying reduction in the severity of life problems. However, other questions, such as the potential role of CD as a harm reduction strategy and the role of severity of in determining the ability to moderate, remain open to debate. In part, this is a result of embarrassment of riches: Some studies examine multiple outcome measures at multiple points in time, with the result that their findings are open to multiple interpretations. Different investigators define "controlled drinking" in different ways (Heather & Tebbutt, 1989). As Cook (1985) demonstrated in his re-analysis of the Sobells-Pendery controversy, readers from either side of the divide can find support for their position in the same data.
(2) What role does money play in all this? To what extent do politics determine which studies get funded? What kinds of public educational campaigns are funded and by whom? To what extent has the clout of anti-CD organizations like the NCADD and the Smithers Foundation affected the willingness of those who benefit from their largess to acknowledge evidence in favor of CD? To what extent does the changing economics of treatment (i.e. the advent of managed care, with its preference for brief, effective treatments) inform both sides of the debate?
How can we step back from the divisiveness that an issue
like this engenders? In the course of reviewing this
literature, I recognized the extent to which reactions
to the debate are based on values as well as facts. My
personal values are such that I am offended by attempts
to prevent information from being disseminated, and by
those who have tried to cut off or to silence debate.
What's tricky, I recognize, is that those gut emotional
reactions lead me to cast heroes and villains in my
head, and to lose sight of nuance. When I "side" with
the psychologists and scientific researchers, I have to
remind myself to step back and focus on the fact that I
also believe that abstinence is a highly desirable goal
for those who accept it, and that I have deep and
enduring respect for AA. Perhaps a prerequisite for
synthesizing reactions to this 40-year-old debate is for
the observer to know his or her biases, and to identify
those issues that trigger affective responses. Perhaps
only when we lay claim to our own values can we
adequately reflect on this emotion-driven debate.