Engs, Ruth C. [Ed.] Controversies in the Addiction's Field. John Wallace. "Chapter 23 - Abstinence and non-abstinence goals in treatment: A case study in the sociology of knowledge."
Back to table of contents Home Page | Article List | Questionnaires | Books | Search my Files | Health Hints | Resume
CHAPTER 23 Abstinence and Non-abstinence Goals in Treatment: A Case Study in the Sociology of Knowledge
John Wallace Ph.D.
It is unfortunate that the issues involved in treatment goals for alcoholics often seem to have been argued in terms of simplistic assertions. The dogmatic assertion that no alcoholic can control his or her drinking precludes the possibility that spontaneous remissions may occur in alcoholism as they do in other diseases. Moreover, the assertion ignores the many formal and informal observations of periods of moderate alcohol consumption that occur routinely in the lives of persons considered to be alcoholic. An equally simplistic assertion is the statement that some (number of) alcoholics can control their drinking. Since it requires documentation of only a single successful case to provide such an assertion to be true, the assertion is trivial. Instead of attempting to debate the merits of either dogmatic or trivial statements, it would probably prove more profitable to consider several critical questions. These are as follows:
1. How many persons diagnosed as alcoholics are likely to be able to drink in a controlled, moderate, attenuated, norrnal, or nonproblem manner?;
2. Over what temporal intervals can such drinking behavior be sustained?;
3. Do reliable treatment technologies exist for achieving non-abstinence goals for diagnosed alcoholics?;
4. What risks are inherent in various definitions of "controlled", "moderate", or "nonproblem" drinking and for whom are such risks apparent?
5. Is it possible to differentiate those persons who are likely to succeed at nonabstinent treatment goals from those persons not likely to succeed at such goals?
6. Is the scientific data base on nonabstinent drinking goals reliable?
7. Do documented spontaneous remissions of alcoholism justify changing treatment goals and procedures from abstinence to nonabstinence goals and methods?
These questions from the basis for the following discussion.
The Roots of Controversy
Numerous early studies provided observations concerning non-abstinent but apparently nonproblem drinking behavior among alcoholics. Pattison, Sobell, and Sobell (1977) and Sobell and Sobell (1978) compiled a bibliography of approximately 80 articles that, in part, and in some manner, addressed the issue of controlled or moderate drinking among alcoholics. The majority of these articles did mention or report some degree of controlled, moderate, nonproblem, or "improved" drinking of varying durations among problem drinkers or alcoholics. Wallace (1983) examined this collection of articles and concluded that, "while there are studies in this collection which require serious attention, it would be misleading for anyone to advertise the entire collection as constituting 'strong scientific evidence' in favor of controlled drinking as a viable treatment goal" (p. 481). Nathan (1985) apparently reached a similar conclusion: "when you look carefully at the series of studies on which the presumed efficacy of controlled drinking treatment was based, you quickly come to realize chat only one study, the very well-known study of Mark and Linda Sobell (1973, 1976) yielded positive data. While there were other studies which, when viewed through the microscope of the statistician, yielded data encouraging to advocates of controlled drinking treatment, basically only the Sobell study yielded data that strongly encouraged the view that controlled drinking treatment could work" (p. 172).
Perhaps the most important of the early studies concerning the feasibility of nonabstinent goals for alcoholics was The classic paper by D.L. Davies (1962). In this paper, Davies described his followup of seven men discharged from London's Maudsley Hosptial before 1955. The seven men had been picked out of a large group of ninety-three men with diagnoses of "alcohol addiction" who were involved in a routine follow up system. What attracted D.L. Davies to these seven men was their apparent ability to drink normally over periods of time ranging from seven to eleven years.
One investigator, Griffith Edwards, did not ignore the important empirical questions raised by Davies' research and conducted a further follow up of tile seven patients Davies had originally followed (Edwards, 1985). Edwards' results directly contradicted the observations of Davies. Following re-investigation, Edwards reported that five of the seven putative normal drinkers "experienced significant drinking problems" both during Davies' original follow-up period and
subsequently, that three of these five at some time also used psychotropic drugs heavily, and that the two remaining subjects (one of whom was never severely dependent on alcohol) engaged in trouble-free drinking over the total period (Edwards, 1985, p. 181).
In effect, only one man out of ninety-two subjects with what might today be called diagnoses of "alcohol dependence" was able to sustain a pattern of moderate drinking. Working with a group of hospitalized, gamma alcoholics at Patton State Hospital in California, Sobell and Sobell (1973) reported impressive findings in favor of controlled drinking behavior therapy over abstinence treatment at followup. A three year independent followup of the Sobell patients by Caddy and colleagues (Caddy et al., 1976) continued to report superior outcomes for patients treated with controlled drinking behavioral therapy versus patients given abstinence oriented treatment.
Despite optimistic reports by the Sobells and subsequently by Caddy et al. (1978), Pendery, Maltzman, and West (1982) on further independent re-followup of these patients could not confirm the substantial amounts of successful controlled drinking that had been reported earlier. For example, Caddy et al. reported that 50 percent of the patients treated with controlled drinking therapy functioned well on lOO percent of the days during year 3 of the followup. Pendery, Maltzman, and West, however, described the six highest functioning patients differently (See Table 23.1).
These patients, described in Table 23.1, according to Caddy et al. had functioned well for 100 percent of the days the 3rd year after controlled drinking treatment! Large discrepancies such as these between the earlier characterization of these patients as "functioning well" and the subsequent characterizations in the refollowup report by Pendery, Maltzrnan, and West have raised serious doubts about the validity of the claims for sustained successful controlled drinking by the Sobell experiment's patients.
A third major study, the Rand Report, by Armor, Polich, and Stambul (1976) and Polich, Armor, and Braiker (1981) resulted in the following: media enthusiasm over the initial claims of large numbers of moderate drinking alcoholics; skeptical reaction from the alcoholism field; criticism of the field for reacting in an emotional, subjective, dogmatic, and ideological manner; subsequent extensive revision downward of the estimates of the numbers of persons capable of sustained moderate or nonproblem drinking.
Since numerous prior reports had pointed to the possibility that some ismall number of alcoholics might undergo spontaneous remissions, the Rand Report's findings of moderate drinking for brief periods by patients treated at eight
National Institute on Alcohol Abuse and Alcoholism (NIAAA) funded treatment centers might otherwise be considered trivial. What made the findings appear significant were
1) the large number of alcoholics who were reported to be drinking normally, and
2) the apparent misunderstandings by the press, lay community, and some members of the academic behavioral science community (e.g., Marlatt, 1983; Peele, 1988) that the report documented sustained long-term moderate or nonproblem drinking.
Numerous persons, lay and professional, confused the followup points used in the Rand reports with the lengths of the windows on actual drinking behavior. While the studies were described as the 6-month, 18month, and 4 year studies, the 6-month and 18-month studies involved asking the subjects about their drinking behavior for the 30 days immediately preceding the followup interview. The 4-year study attempted to measure the patient's quantity and frequency of drinking in terms of the 30 days before the subject's last drink. Hence, none of these reports by the Rand authors provided direct observations on sustained moderate or nonproblem drinking among alcoholics. In effect, the Rand study results on brief periods of nonproblem drinking among alcoholics were trivial since they essentially confirmed what clinicians and recovering alcoholics themselves have known for some time: brief periods of moderate or nonproblem drinking are common among alcoholics and not exceptional (Wallace, 1985). The second issue raised by the Rand reports concerned the large number of alcoholics that were reported to be drinking normally. This issue, however, must be viewed in terms of three considerations:
1) the length of the followup period;
2) sampling bias due to loss of subjects; and
3) reliability and validity of measurement of quantity and frequency of consumption in the Rand studies.
The Rand 18-month study reported that 22% of the subjects in these eight NIAAA funded treatment centers were normal drinkers, a rate of normal drinking that was most unexpected by prior observations. Actually, when the data are examined by individual treatment centers, they are even more deviant in terms of previous research and clinical observations. For example, of the patients in "Treatment Center B" who were drinking at all, 70 percent were described by the Rand authors as normal drinkers! While small numbers of persons admitted to an alcoholism treatment center can be expected to show subsequent nonproblem drinking, these rates were simply too deviant to be considered reliable. Is it possible that they were spuriously high due to a combination of bias on outcome, invalid measurement of normal drinking, and an extremely brief follownp window?
With regard to bias on outcome, the loss of large numbers of subjects from the data base pointed directly to this possibility.
Description of Caddy et al.'s highest functioning controlled drinkers according to Pendery,Maltzman, and West*
CD-ET Subject and multiple collaterals state he drank heavily throughout year 3, during which he resided in three states. He used an assumed name on his driver's licence because of an outstanding alcohol-related felony bench warrant issued in year 2. In February, year 3, police were called by neighbors of subject's mother, when he threatened violence and caused a disturbance while drunk, and in April, he was too drunk to attend his brother's funeral. (This trend continued and in year 4 he was arrested for drunk drinking and rehospitalized.)
CD-ES Subject states that "the third year included some of my worst drinking experiences. In August 1972, after drinking more than a fifth of liquor per day, I went to the San Bernadino Alcoholism Services for help. I was having shakes and other withdrawal symptoms and was very sick physically. By then, a physician had told me I had alcohol cirrhosis of the liver." A record of the subject's application for treatment there, his wife's statement, documentation of subsequent hospitalization for alcoholism treatment, and continued deterioration of his health are consistent with his seff-report.
CD-E11 Subject and collateral state that year 3 was his worst year. His records show he spent time in jail, in a state hospital, and in a Veterans Administration hospital because of actions he committed while intoxicated. Toward the end of year 3, he had additional arrests, including one for drunk driving.
CD-E13 Subject and multiple collaterals state that he was abstinent throughout year 3. He states, however, that this was in spite of the controlled drinking treatment. He became abstinent only after additional alcohol-related incarcerations in hospitals, jail, and road camp. He then spent 5 months of year 2 at Twelve Step House, an AA-oriented alcoholism recovery home, to which he attributed his abstinence.
CD-E15 Subject and multiple collaterals state he was drinking excessively (sometimes as much as a fifth per day and some beer) when he was not going to be at work (His blood alcohol of 0.34 percent on a recent admission to a hospital confirms his high reported tolerance.) He had not yet experienced serious alcohol withdrawal symptoms during year 3 and did not require hospitalization. According to his family, however, his health was already beginning to deteriorate, leading to repeated alcohol-related medical problems and hospitalizations from 1976 to the present.
CD-E18 Subject and collateral state that he successfully controlled his drinking throughout year 3, although, "it would not be entirely accurate to say I never drank excessively." We found no evidence of alcohol-related problems in any major life area. In our view, this subject, who apparently had not experienced physical withdrawal symptoms, might have been appropriately designated an alpha (psychologically dependent) alcoholic.
* Pendery, Maltzman, and West 1982, p. 173.
While the Rand authors argued that their samples were not biased on outcome, further analyses by Wallace (1979; 1989a) disputed these arguments. Wallace noted that the Rand authors relied upon their composite remission rates to demonstrate that their samples were not biased on outcome due to loss of large numbers of subjects. The composite remission rate was made up of the combined abstinence and normal drinking rates for the various centers and was reported to be uncorrelated with subject drop out rates. Wallace ran separate analyses for abstinence and normal drinking. When abstinence rates for the eight treatment centers were correlated with their subject loss rates, Wallace obtained a significant correlation of +.79. This correlation of +.79 between abstinence rates and subject loss rates contrasted sharply with the Rand author's correction of -.07 (nonsignificant) for composite remission rates and subject loss rates. Moreover, Wallace's analysis of the correlation between normal drinking rates and subject loss rates was -.14 (nonsignificant). In effect, a measure of acceptable reliability and validity, abstinence rates, showed substantial bias on outcome while a measure of questionable reliability and validity, normal drinking, showed no bias on outcome. How can this internal contradiction in the Rand data be explained? The most parsimonious and likely explanation for this contradiction is simply that the Rand data were not only contaminated by substantial bias on outcome due to loss of subjects but were further contaminated by invalid measurement of normal drinking. As a result of sample bias, the Rand outcome rates for both normal drinking and abstinence in the 18-month study were spuriously high. Furthermore, the invalid measurement of normal drinking, indicates an even lower true rate for normal drinking. Moreover, the extremely short followup window of only 30 days worth of drinking behavior used in the Rand study, contributed further to the spuriously high reported normal drinking rates. In short, the total sample normal drinking rate of 22 percent that the Rand authors reported that aroused so much controversy was a product of improper data analysis, bias due to large dropout rates, invalid measurement of normal drinking, and a brief followup window.
Methodological improvements characterized the 4-year Rand report, particularly with regard to sampling and subject followup (Polich, Armor, and Braiker, 1981) However, problems with invalid measurement of "nonproblem" drinking (labeled normal drinking in the earlier studies) and also with the brief follow window remained. The Rand authors reported that one-quarter of the sample under reported their actual consumption of alcohol. Also, quantity and frequency of consumption were still measured over too brief a period (30 days before the subject's last drink). Under these circumstances, the Rand authors' brief nonproblem drinking rate in the 4year study was 14 percent (corrected for invalid measurement of quantity and frequency of consumption).
The authors did provide an estimated rate of long-term nonproblem drinking of 7 percent. Correcting this estimated sustained rate for invalid measurement as well as other factors suggests a long term nonproblem drinking rate of approximately 3 to 4 percent. This corrected long term nonproblem drinking rate from the 4-year study is clearly markedly lower than the 22% entire sample brief normal drinking rate of the 18-month study that attracted such media attention and provoked intense controversy.
In the final analysis, the Rand reports (as well as the D.L. Davies study and the Sobell experiment on controlled drinking) are more appropriately construed as episodes in the sociology of knowledge rather than exemplars of rigorous scientific research. These studies, in my opinion, told us considerably more about the beliefs, attitudes, values, behaviors, prejudices, biases, and knowledge processes of some persons in the social and behavioral science communities than they did about the behavior of alcoholics.
Recent Negative Findings
Important studies on nonabstinence treatment outcomes for alcoholics have been provided by Foy et al. (1984), Rychtarik et al. (1987), Pettinati et al. (1982), and Helzer et al. (1985). The experiment by Foy et al. and Rychtarik et al. was significant since it provided perhaps the closest attempt to date at replication of the controversial Sobell and Sobell (1973) experiment discussed earlier. Because of numerous differences between the Foy et al. experiment and the Sobell experiment, it cannot be taken as an exact replication. The ingredients, however, were quite similar. Both studies used hospitalized, chronic alcoholics, provided patients with training in behavioral drinking management techniques, and followed patients for a long time.
Foy et al. randomly assigned patients to broad-spectrum behavioral treatment including controlled drinking skills training or to the same broad-spectrum behavioral treatment package without the controlled drinking skills component. Hence, all patients received the same treatment program with the exception that some got drinking management training as well and others did not.
The results showed that adding a controlled drinking component to a package of behavioral therapy components resulted in abusive drinking in the short run (6-month followup) but had no further effect, negative or positive, after six-month followup. Long-term followup of these patients revealed virtually no sustained controlled drinking (Rychtarik et al., 1987). Of those patients who appeared to be engaging in brief controlled drinking at the point of the 7- to 12-month followup, none were able to sustain such behavior over time.
For the entire sample, only 2 patients appeared to have achieved stable moderate drinking. While controlled drinkers did poorly in this research, so did abstainers, only 2 of whom were able to achieve stable abstinence through the behavioral methods employed. Hence, in this research as in the Sobell research, behavioral training methods were largely ineffective in producing either stable controlled drinkers or stable abstainers.
Pettinati and colleagues (1982) followed patients for four years after treatment. These investigators reported a 3 percent long term controlled drinking rate.
In discussing this study (Peele, 1988 p.378) claimed that "in populations such as these, abstinence through a lengthy followup period is exceedingly rate." Wallace (1989b, p. 264), however, pointed out that, in fact, in the Pettinati study, "abstinence throughout the four-year followup was not exceedingly rare, as claimed by Peele, but substantial." As Pettinati et al. (1982), p. 213) commented: "A substantial percentage of patients (29%) were able to maintain abstinence and show overall good life-adjustment throughout the four years." They added the statement that "abstinence with good adjustment in the first year is the only status that seems to have any stability over the four-year period."
It would appear then that part of the strategy for advancing acceptance of nonabstinent treatment goals for alcoholics is to denigrate the achievements of abstinence-oriented treatment programs even when the actual data do not justify such denigration. Hence, positive outcomes of abstinence-oriented programs are often dismissed, ignored, or distorted while studies purporting to show the case that studies of abstinence-oriented treatment effectiveness have been carried out on very poor prognosis patients who were unlikely to recover without major social rehabilitative efforts far in excess of what one normally expects of alcoholism treatment as such. Wallace et al. (1988), Patton (1979), and Hoffman and Harrison (1986) have all shown that abstinence-oriented programs can be effective with treatment populations whose social circumstances do not preclude an adequate response to brief alcoholism treatment.
An important study by Helzer et al. (1985) found very little evidence for stable moderate drinking among 1,289 alcoholics discharged from medical and psychiatric treatment facilities. Only 1.6 percent of the subjects met Helzer et al. very liberal definition of stable moderate drinking for the three year period preceding the followup interview. Nonmoderate drinking in this research was defined as 7 or more drinks a day on four or more days in any one month. As Wallace pointed out (1989b, p. 265), "In effect, provided that medical, legal, or social problems were denied and not detected, an alcoholic could drink as much as six drinks a day for twenty-seven days on each and every month and
be categorized as a moderate drinker. Moreover, this same alcoholic could also get drunk three times each month, or thirty-six times per year for a total of 108 episodes of intoxication over the 3-year followup period and still be considered a moderate drinker." Despite such very liberal criteria, Peele (1988, p.378) was distressed that Helzer and colleagues would "disqualify from remission" any alcoholic who "got drunk four times in any one month in a three year period."
In general, then, consideration of the major studies that have claimed successful nonabstinent outcomes of treatment of alcoholism and of studies that have failed to support such claims indicates the following:
1) nonabstinent outcomes such as reduced drinking and nonproblem drinking appear possible for only very small numbers of persons who have been diagnosed as alcoholics;
2)such outcomes are far more likely to occur over brief temporal spans than over lengthy periods;
3) such outcomes are more likely to be reported from methodologically inadequate studies than as a result of scientifically rigorous investigations;
4) a reliable treatment technology for producing normal drinkers out of alcoholics does not exist;
5) while some observations suggest that level of alcohol dependence may be related to brief remission of alcoholism (e.g., Foy et al., 1984; Orford et al., 1976) it is not possible to predict with an acceptable degree of accuracy, the small numbers of alcoholics who may succeed at nonabstinent treatment goals from the vast majority of alcoholics who cannot;
6) nonabstinent treatment goals may be appropriate for some persons with drinking problems other than alcoholism (e.g., Sanchez-Craig and Wilkinson, 1987);
7) while certain types of problem drinkers may be able to moderate their intake of alcohol, alcoholics are at considerable risk for grave physical, psychological, and social consequences if they continue to drink alcohol and they should be so informed. In the face of such severe consequences, small numbers of spontaneous remissions do not justify routine attempts to train alcoholics to drink in a controlled or nonproblem manner;
8) claims in support of nonabstinent goals for alcoholics will continue to be made and will probably be extended to include drug addicts as well. Such claims must not be accepted uncritically by the behavioral science community but must be subjected to rigorous evaluation and criticism.
As we have seen, a close examination of the details of the major pieces of empirical research upon which prior enthusiastic claims have been based suggests that such claims have been greatly exaggerated. Nonabstinence goals for alcoholics, for the most part, have been achieved in very small numbers of alcoholics for only very brief periods of time.
This fact should be home in mind as new claims in favor of nonabstinence goals are made and revisions in the traditional goal of abstinence for alcoholics are recommended.
Armor, D.J., Polich, J.M., and Stambul, H.B. (1976). Alcoholism and treatment. Rand Corporation: Santa Monica, CA.
Caddy, G.R., Addington, H.J., and Perkins, D. (1978) Individualized behavior therapy for alcoholics: a third year independent double-blind follow-up. Behavior Research and Therapy, 16 345-62.
Davies, D. (1962). Normal drinking in recovered addicts. Quarterly Journal of Studies on Alcohol, 23 94-104.
Edwards, G. (1985). A later follow-up of a classic case series: D.L. Davies's 1962 report and its significance for the present. Journal of Studies on Alcohol, 46 (3) 181-190.
Foy, D.W., Nunn, L.B., and Rychtarik, F.G. (1984) Broad-spectrum behavioral treatment for chronic alcoholics: Effects of training controlled drinking skills. Journal of Consulting and Clinical Psychology, 52 (2) 218-230.
Helzer, J.E., Robins, L.N., Taylor, J.R., Carey, K., Miller, R.H., Combs-Orme, T., and Farmer, A. (1985). The extent of long-term moderate drinking among alcoholics discharged from medical and psychiatric treatment facilities. New England Journal of Medicine, 312 (26) 1678-1682.
Hoffman, N.G. and Harrison, P.A. (1986). The CATOR 1986 Report. Ramsey Clinic Publications: St. Paul.
Marlatt, G.A. (1983). The controlled drinking controversy: A commentary. The American Psychologist, 38 1097-1110.
Nathan, P.E. (1985). Alcoholism: A cognitive social learning approach. Journal of Substance Abuse Treatment, 169-173.
Orford, J., Openheimer, E., and Edwards, G. (1976). Abstinence or control: The outcome for excessivedrinkers two years after consultation. Behavior Research and Therapy, 14 409 418.
Pattison, E.M., Sobell, M.B., end Sobell, L.C. (1977). Emerging Concepts of Alcohol Dependence. Springer: New York.
Patton, M. (1979). Validity and Reliability of Hazelden Treatment Followup Data. Hazelden Education Services: Center City, MN.
Peele, S. (1988). Can alcoholism and other drug addiction problems be treated away or is the current treatment binge doing more harm than good? Journal of Psychoative Drugs, 20 (4) 375-383.
Pendery, M.L., Maltzman, I.M., and West, L.J. (1982). Controlled drinking by alcoholics? New findings and a reevaluation of a major affirmative study. Science, 217 169-175.
Petinatti, H.M., Sugerman, A.A., DiDonato, N., and Maurer, H.S. (1982). The natural hisory of alcoholism over four years after treatment. Journal of Studies on Alcohol, 43 (3) 201-215
Polich, J.M., Amour, D.J., and Braiker, H.B. (1981). The Course of Alcoholism: Four years After Treatment. Report prepared for the National Institute on Alcohol Abuse and Alcoholism, U.S. Department of Health, Education, and Welfare. Rand Corporation: Santa Monica, CA.
Rychtarik, R.G., Foy, D.W., Scott, T., Lokey, L., and Prue, D.M. (1987). Five to six-year follow-up of broad-spectrum behavioral treatment for alcoholism: Effects of training controlled drinking skills. Journal of Consulting and Clinical Psychology, 55 (1) 106-108.
Sanchez-Craig, M. and Wilkinson, D.A. (1987) Treating problem drinkers who are not severly dependent on alcohol. In M.B. Sobell and L.C. Sobell (eds.) Moderation as a Goal of Treatment for Alcohol Problems: A Dialogue. Haworth Press: New York.
Sobell, M.B. and Sobell, L.C. (1978). Behavioral Treatment of Alcohol Problems. Plenum: New York.
Sobell, M.B. and Sobell L.C. (1976). Second-year treatment outcome of alcoholics treated by individualized behavior therapy: Results. Behavior Research and Therapy, 14 195-215.
Sobell, M.B. and Sobell, L.C. (1973). Individualized behavior therapy for alcoholics. Behavior Therapy, 4 49-72
Wallace, J. (1979) Alcoholism Treatment Revisited. Originally appeared in World Alcohol Project, 2 (1) 1979. Reprinted in Wallace, J. (1989a) Writings. Edgehill Publications: Newport, RI.
Wallace J. (1985). predicting the onset of compulsive drinking in alcoholics: A biopsychosocial model. Alcohol, 2 589-595.
Wallace J. (1989a). Alcoholism treatment revisited. In Wallace, J. Writing. Edgehill Publications: Newport, RI.
Wallace J., McNeill, D., Gilfillian, D., MacLean, K. and Fanella, F., (1988). I. Six-month treatment outcomes in socially stable alcoholics: Abstinence rates. Journal of Substance Abuse Treatment, 5 247-252.
Wallace, J. (1983). Alcoholism: Is a shift in paradigm necessary? Journal of Psychiatric Treatment and Evaluation, 5 (6) 479-485.
Wallace, J. (1989b). Can Stanton Peele's opinions be taken seriously? Journal of Psychoactive Drugs, 21 (2) 259-271.
Back to table of contents/