By William R. Miller

INTRODUCTION

There are striking parallels in the nature, etiology, and course of addictive behaviors. Alcohol, tobacco, and other drug problems overlap substantially in epidemiology, are all influenced by a combination of biomedical and psychosocial factors, involve impaired personal control, and are characterized by high rates of relapse (Brownell et al. 1986; Miller 1980).

These similarities extend to a broader range of compulsive behaviors (Heather et al. 1991; Orford 1985; Peele 1985). Yet, both treatment programs and research for alcohol versus other drug problems have developed with an astonishing degree of isolation.

Although clients of the Center on Alcoholism, Substance Abuse, and Addictions are increasingly difficult to classify as alcohol versus other drug abusers and there is increasing public recognition that alcohol is a drug, there remain separate treatment systems and staffs, research institutes, and self-help organizations.

Markedly different emphases have evolved for both treatment and prevention strategies. The reasons for this partition arise from historical and political events rather than from compelling differences inherent in the problems and clients. In considering how to advance behavioral treatment for drug abuse and dependence, it may be informative to consider the state of research knowledge in treatment related to one particular drug: alcohol.

Treatment research is well developed in the alcohol field; currently, more than 250 controlled trials have been published. Behavioral treatments in particular are well studied, comprising more than half of the trials to date. For several behavioral strategies, dismantling research has been conducted to examine theoretical and practical foundations of efficacy. Alcohol treatment assessment technology has advanced substantially during the past two decades, permitting better quantification of outcomes.

Further, important gains have been made in knowledge on the interaction of client and treatment characteristics (“treatment matching”), and a multisite collaborative trial on this subject is underway at the National Institute on Alcohol Abuse and Alcoholism (NIAAA). This chapter will summarize this broad literature on alcohol treatment outcome, emphasizing behavioral treatments. It also will highlight issues that have emerged along the way, which may be informative for research and practice in the behavioral treatment of drug problems more generally.

TREATMENT OUTCOME REVIEWS

Reviews of the alcohol treatment outcome literature have progressed through several stages. The first-generation reviews were narrative summaries of findings with some commentary on methodology (e.g., Baekeland 1977; Bowman and Jellinek 1941; Voegtlin and Lemere 1941).

As studies accumulated, a second wave of reviewers attempted to derive average outcome statistics for treatment in general or for particular modalities, but these reviews remained largely narrative (Costello et al. 1977; Emrick 1974, 1975; Miller and Hester 1980).

In an attempt to discern more reliable patterns, Miller and Hester (1986a) provided a narrative review restricted to controlled trials-comparisons of two or more groups that were equated before treatment by randomization or matching.

This review concluded that only six treatment methods were “currently supported by controlled outcome research”: aversion therapies, behavioral self-control training, the community reinforcement approach, behavioral marital therapy, social skills training, and stress management.

It is noteworthy that all six of these are behavior therapies. This list was contrasted with “treatment methods currently employed as standard practice in alcoholism programs”: the 12-step method, alcoholism education, confrontation, disulfiram, group therapy, and individual counseling-none of which, it was concluded, were then adequately supported as effective treatment modalities by controlled research (Miller and Hester 1986a, p. 162).

A subsequent review by Holder and colleagues (1991) added two dimensions. First, controlled studies pertaining to each treatment modality were classified as positive or negative trials, and a weighted evidence index (WEIn) was compiled for each modality as a kind of box score of efficacy.

Second, through the polling of clinical experts, the optimal number and type of units of service required to deliver each modality were determined, and these units were costed to estimate the average cost (not price) of delivery for each modality.

This permitted the construction of a grid in which treatment modalities were located in two-dimensional space according to cost category and evidence for efficacy (WEIn).

The 10 modalities judged to have good or fair evidence of effectiveness included all six previously identified by Miller and Hester (1986a), as well as behavior contracting, brief motivational counseling, antidepressant medication, and disulfiram.

A striking finding was a significant negative correlation (r = -.385, p < .05) between the cost and effectiveness of modalities. Like the National Academy of Sciences (1990), Holder and colleagues (1991) concluded that it is wrong to ask: “Is alcoholism treatment (cost-) effective?”

Treatment modalities appear to differ dramatically in both cost and effectiveness, and no blanket endorsement can be given for “treatment.” They recommended instead focusing on the question, “Which alcoholism treatment modalities are most effective for the least cost?” (Holder et al. 1991, p. 533)

None of these reviews took into account the methodologic quality of studies in other than a narrative way. Thus, in the Holder and colleagues (1991) analysis, a poorly designed and conducted study was given the same weight in a modality’s WEIn score as a study with exemplary design and performance.

Two reviews are currently underway to take this next step: one at The University of New Mexico (UNM) and one at the National Drug and Alcohol Research Centre (NDARC) in Sydney, Australia. Using different approaches, each group is classifying studies on a variety of methodologic dimensions, with plans to use these ratings to improve the interpretation of study outcomes.

The plan at UNM is to calculate cross-products of methodologic strength scores and a treatment outcome classification. A sum of these cross-products, interpreted in relation to the total volume of research conducted, will provide a more refined index of research evidence for the efficacy of each modality.

The NDARC group, under the direction of Dr. Richard Mattick, is adding another important dimension to its review: the calculation of effect sizes. This will be the first comprehensive review of outcome literature in this area to include effect sizes. Furthermore, the reviews encompass treatments for alcohol, tobacco, and other drugs.

PROMISING BEHAVIORAL STRATEGIES

From these reviews, what behavioral approaches appear to be most promising? Supported modalities can be grouped subjectively into two general strategies:

1. Treatments designed to suppress use: Antidipsotropic medication with compliance contracting, Aversion therapies (covert sensitization), Behavior contracting, Behavioral self-control training, and Brief motivational counseling.

2. Methods to teach skills for successful sober living: Behavioral marital therapy, Community reinforcement approach, Social skills training, and Stress management training.

All five modalities designed to suppress use are readily applicable to other drug problems. Pharmacologic parallels to disulfiram are drug antagonists or agonist/antagonist combinations.

The principal problem in the use of medications such as disulfiram and naltrexone is compliance. Various procedures have been used to increase disulfiram compliance, including behavior contracting, spouse monitoring and encouragement, more frequent clinic visits with monitoring, and implantation (Fuller 1989).

Behavior contracting more generally has been found helpful in maintaining sobriety, and it is a familiar approach in treating drug abuse. Aversion therapies have a long and controversial history.

Holder and colleagues (1991) found adequate experimental support for covert sensitization, the only form of aversion therapy that involves no physical unconditioned stimulus but is conducted in imagination.

It clearly is possible to induce a conditioned aversion reaction through covert sensitization, and the establishment of conditioning has been found to be predictive of favorable outcome (Elkins 1980; Miller and Dougher 1989).

More broadly, conditioning-based procedures (e.g., cue exposure) have been explored as relevant processes in treatment for alcohol and other drug problems (Greeley and Westbrook 1991).

Behavioral self-control training (BSCT) involves teaching learningbased procedures to clients to assist them in altering their own behavior. In the Holder and colleagues (1991) review, BSCT had the largest number of trials (17) and the second highest WEIn score (after social skills training) of any modality.

BSCT has most often been applied with a goal of moderation of alcohol use (Hester and Miller 1989) or other drug use (Wilkinson and LeBreton 1986), but it also has been applied with a goal of abstinence (e.g., Sanchez-Craig et al. 1984).

Typically, BSCT includes a combination of strategies designed to modulate use (e.g., specific goal-setting, self-monitoring, alteration of the topography of use, stimulus control, self-reinforcement, and refusal training) and methods for altering the probability of future use (e.g., functional analysis and coping skill training).

Applications to other drug use are straightforward. The success of brief counseling in altering problem drinking has been documented consistently. In a recently completed review, Bien and colleagues (in press) identified 32 controlled trials with strikingly consistent results.

The methodology of these studies compares favorably with that for the alcohol field in general (mean of 13 on a composite methodologic quality scale of 0-17). Brief counseling (typically one to three sessions) has been found consistently superior to untreated controls and in 11 of 13 randomized trials has had comparable impact to that of more extensive treatments.

These brief interventions have included virtually no behavioral skill training, medication, or contracting. Their content is more cognitivemotivational (Miller and Rollnick 1991).

Six elements, summarized in the acronym FRAMES, have been described as common components of effective brief interventions (Miller and Sanchez, in press): FEEDBACK of personal assessment results, Emphasis on personal RESPONSIBILITY for change, ADVICE to change use, Description of a MENU of options for change, Therapeutic EMPATHY as a predominant counseling style, and Support for client SELF-EFFICACY and optimism.

In the Holder and colleagues (1991) grid, brief motivational counseling was the only entry in the box with highest evidence of efficacy and lowest cost. Researchers at UNM have developed and tested a “Drinker’s Check-up” to manifest the FRAMES elements (Miller and Sovereign 1989; Miller et al., in press) and are exploring the extension of this model to a brief intervention for other drug use.

All four effective behavioral methods that support a sober lifestyle are also readily applicable in the treatment of drug problems in general. Behavioral marital therapy teaches skills for more effective communication and positive reinforcement within intimate relationships (O’Farrell and Cowles 1989).

Social skills training focuses more generally on communication skills such as assertiveness for effective social relations (Monti et al. 1986).

Studies of stress management procedures in alcohol treatment have focused primarily on relaxation training and systematic desensitization, although a broader array of technologies can be applied (Stockwell and Town 1989).

Azrin’s community reinforcement approach, which has yielded some of the largest treatment effects in the literature, combines many of the wellsupported components described above, including monitored disulfiram, behavior contracting, motivational counseling, behavioral marital therapy, social skills training, and mood management (Sisson and Azrin 1989).

As with other behavioral strategies, the community reinforcement approach can be applied directly in the treatment of other drug problems (e.g., Higgins et al. 1991).

Investigators at UNM currently are conducting three clinical trials of this approach with 172 alcoholics (R0l-AA07564), heroin addicts (R18-DA06953), and dually diagnosed homeless individuals (R0l-AA08331).