There are numerous studies exploring the importance of client characteristics in treatment outcome. How do the outcomes of male and female alcoholics compare? Although it is frequently asserted that female alcoholics have poorer prognoses than male alcoholics, reviews of the empirical literature reveal that alcoholic men and women do not differ in treatment outcome rates.(2) However, positive response to alcohol and drug treatment has been associated with several client characteristics other than sex: being married, employed, of a high social class, financially secure, socially active and well adjusted to work and marriage, and having little history of arrest. Unfortunately, these positive prognostic characteristics are not typically found in offender populations.

In fact, poor treatment outcome has been associated with client characteristics that are prominent in offender groups: aggressiveness, high rates of attempted suicide, organic brain syndrome and sociopathic personality. Work by McLellan and his colleagues(3) with both alcohol- and drug-dependent clients indicates that the severity of psychiatric symptomatology is an important factor in predicting response to treatment; alcohol- and drug-dependent clients with low psychiatric severity at admission have achieved good outcomes across a variety of treatments, whereas those with high psychiatric severity have shown little improvement and uniformly poor outcomes. Offenders with a dual diagnosis (a psychiatric diagnosis and a diagnosis of alcohol or drug abuse and dependence) are likely to respond poorly to substance-abuse programming.

In summary, it is important to recognize that:

1. client characteristics have played a major role in predicting response to alcohol- and drug-treatment programming;
2. some offenders can be expected to have a number of poor prognostic characteristics; and
3. any comparison of outcome rates across treatment programs must take into account differences in the characteristics of the offender populations served.

In the past few years, there has been much interest in how the intensity and duration of treatments and treatment settings affect outcomes. Spiralling health-care costs have stimulated an assessment of the effectiveness of traditional methods of service delivery compared with lower-cost alternatives. Specifically, questions have been raised about the required length of residential treatment, the cost effectiveness of residential versus day treatment, and out-patient alternatives.(4)

Findings from well-controlled clinical trials have been remarkably consistent in reporting no advantage for lengthy or intensive treatment programming. For example, residential alcoholism treatment lasting one to two weeks has been found to produce results comparable to treatment lasting several months. Furthermore, random controlled trials at the Donwood Institute in Toronto, Ontario,(5) and the Butler Hospital in Providence, Rhode Island,(6) demonstrated equally positive outcome results for clients in day-treatment programs and those in more costly residential treatment. These studies suggest that factors other than treatment length and setting should receive greater attention in the design of future substance-abuse treatment programming.

Another question that has been raised is whether treatment on an outpatient basis may be as effective as residential treatment for alcohol and drug abusers. In a large-scale evaluation of clients discharged from federally funded alcoholism-treatment facilities in the United States, the Rand Corporation found no differences in outcome for clients treated in out-patient programs and those in residential programs. Similarly, the Drug Abuse Reporting Program (DARP), involving 44,000 clients from 52 federally supported drug-abuse treatment centres throughout the United States, reported that for opioid addicts, other drug-abusing adults and youth clients (19 and under), there was an out-patient treatment option (methadone maintenance or drug-free out-patient counselling) that was equally as effective as or more effective than residential treatment. The results from these large-scale evaluations must be interpreted with caution, however, because clients who self-select to enter residential or out-patient treatment may differ in important but unrecognized prognostic characteristics.

Fortunately, a number of random controlled trials reported in the literature permit a direct comparison of the general effectiveness of outpatient and residential services. In these trials, alcoholics or other substance abusers were randomly assigned to out-patient counselling or to residential treatment. A number of such studies have been conducted in probation and parole systems. For example, 74 delinquent, drug-abusing adolescents within the San Francisco Juvenile Probation Department were randomly assigned to in-patient treatment or to the usual out-patient probationary care.(7) In-patient treatment averaged 132 days and consisted of psychodynamically oriented psychotherapy, community meetings, family therapy, recreational therapy, psychodrama, and an onward school program. Follow-up at one year after admission showed no difference on a variety of outcome measures, including alcohol and drug use and social functioning between the adolescents receiving in-patient treatment and those receiving the usual out-patient probationary services. Similar results were reported in a study of adult parolees with a history of opiate abuse.(8) Parolees randomly assigned to parole services or an experimental halfway house program showed no differences in the rate of new criminal convictions or in the number of drug-free weeks in the community during the first year following release from incarceration.

The weight of evidence from these and other well-controlled trials is clear. Out-patient treatment for substance abuse can produce outcomes essentially equivalent to those of residential treatment at substantially lower cost.

Are some treatment methods more effective than others? This question is currently the subject of some controversy in the field of substance-abuse treatment. Bill Miller, a prominent scientist in the field, argues that certain treatment methods such as aversion therapy, behavioural self-control training, social skills training, stress management, marital and family therapy, and a community reinforcement approach have demonstrated specific effectiveness, particularly in the treatment of alcoholics. (Ironically, Miller notes that these methods are not currently employed in most treatment programs.)

In contrast, other investigators conclude that differences between treatment methods have demonstrated relatively little effect on long-term outcomes. Although statistically significant outcome differences are occasionally reported for different treatment methods, the magnitude of these differences is usually small. Whereas client characteristics at treatment intake typically account for about 30% of the outcome variance, treatment variables account uniquely for only 6% to 7% of the variance, with additional variance being shared with the predictive value of client characteristics. Clearly, much remains to be learned about improving treatment programming.

Despite the controversy about the impact of treatment variables on outcome, there is a growing consensus in the field that the search for a single, universally effective treatment approach is misguided. It is now widely acknowledged that there is broad heterogeneity among alcoholics, cocaine abusers and other substance abusers, and that a client with one set of characteristics may respond favourably to one type of treatment or treatment setting, whereas a client with another set of characteristics may respond more favourably to another treatment approach. The attempt to match clients to treatments in order to improve outcome results is referred to as client-treatment matching or the matching hypothesis. Although the development of empirical evidence of matching effects is in its infancy, there is general agreement that the differential assignment of clients with drug and alcohol problems to different treatments has the potential to substantially improve outcome results.

Over the past few years, the promise shown by a variety of pharmacological agents in the treatment of cocaine abuse has generated much excitement. The agents studied have included desipramine, lithium, bromocriptine, and other psychotropic drugs. Recently, prominent researchers in the area have cautioned that the development of a single, definitive treatment for all cocaine abusers now appears no more likely than it has for opiate abusers or alcoholics. Progress is being made, however, in defining appropriate matches of pharmacological agents to types of cocaine abusers. Studies to date have used DSM-III Axis 1 symptomatology and have tended to be non-blind, non-placebo preliminary trials. Early data suggest that some pharmacological agents may be effective for specific diagnostic subpopulations of cocaine abusers. For example, methylphenidate, a substitute medication that shares cross-tolerance with cocaine, has been found effective only in the treatment of the 5% of cocaine abusers who have a clearly established attention-deficit disorder. Similarly, non-cyclothymic cocaine abusers have shown no benefit from lithium, whereas lithium has been associated with cessation of cocaine abuse and reduced cocaine craving in cyclothymic patients. Larger samples and double-blind controlled trials are needed to substantiate these findings.

In order to demonstrate a differential treatment response or matching effect, it is necessary to study clients who vary on a particular characteristic under two or more treatment conditions. A simple case is illustrated in the figure: client type "A" has a positive outcome under treatment "X" but a poor outcome under treatment "Y," whereas the reverse is true of client type "B." In this example, a matching effect would still be demonstrated if client type "B" showed a similar outcome under treatments "X" and "Y," and client type "A" continued to show a better outcome under treatment "X."

The search for patient-treatment matching effects requires the reliable assessment of patient variables on the one hand, and treatment variables on the other. More progress has been achieved in the conceptualization and assessment of salient characteristics of patients than in the measurement of treatment variables; nevertheless, advances are being made in the evaluation of certain aspects of treatment environments. The table lists some patient and treatment variables that have received attention in the alcohol- and drug-abuse treatment literature in relation to patienttreatment matching.

The prevention of relapse is increasingly being recognized as a central problem in the treatment of alcoholism and other substance abuse. One influential theoretical framework that has been applied to the problem of relapse is Albert Bandura's cognitive-social learning approach. In Bandura's theory of self-efficacy, the critical distinction between initiation and maintenance strategies heralded a significant conceptual development for the addiction treatment field. The maintenance of behavioural change had been largely neglected in alcoholism and other substance-abuse programming. However, attention has recently focused on the development of relapse-prevention treatment strategies explicitly designed to foster the maintenance of behavioural change.

The Addiction Research Foundation in Toronto has been evaluating a cognitive-social learning approach to relapse prevention. The relapse-prevention model essentially involves a highly individualized analysis of the client's drinking or drug-use behaviour over the previous year to determine the high-risk situations experienced by that client. A 100-item self-reported questionnaire called the Inventory of Drinking Situations(10) has been developed to assess drinking within the eight categories of relapse situations identified in the work of Allan Marlatt(11): unpleasant emotions, physical discomfort, pleasant emotions, testing personal control, urges and temptations, conflict with others, social pressures to drink, and pleasant times with others. The IDS subscales have received positive reports on reliability, content and external validity, and a client classification system based on the profile of IDS subscores has been shown to be associated with age, sex and consumption-related variables. A parallel questionnaire for drugs other than alcohol, the Inventory of Drug-Taking Situations (IDTS), is currently undergoing psychometric evaluation.

In the cognitive-social learning approach to relapse prevention, the first step in the development of an individually tailored treatment plan is the assessment of a client's high-risk situations. Treatment focuses on encouraging the client to engage in homework assignments designed to develop alternative coping responses in high-risk situations for relapse. Mastery experiences in successfully implementing alternative behaviours to drinking or drug use in these situations have a powerful impact on the client's cognitive appraisal of personal coping abilities, resulting in an improved perception of self-efficacy and a change in future drinking or drug-use behaviour.

Our increased understanding of effective treatment for drug and alcohol problems is reflected in the evolution of question-guided clinical investigation in the field. More simplistic questions about the effects of patient characteristics on outcome and the comparative effectiveness of treatments varying in duration, intensity, setting and method are leading to a greater focus on more complex questions about client-treatment interaction effects and the development of client-specific relapse-prevention strategies. Fundamental to this evolution has been an acknowledgment of the tremendous heterogeneity among alcoholics and other substance abusers, and of the great diversity of possible treatment approaches.

Evidence to date suggests that under certain conditions, matching can yield a major improvement in client outcome. Studies with a strong theoretical orientation in the choice of client-treatment variables have tended to produce the most dramatic effects of matching, accounting for 16% to 30% of the treatment outcome variance. These results are extremely encouraging. Clearly, much remains to be learned about the most salient characteristics of clients with alcohol and drug problems and how they can be matched to theoretically relevant treatment structures. Nevertheless, current evidence strongly suggests that treatment outcome results for substance abusers will improve substantially with our growing knowledge of the optimal matching of clients to treatment alternatives and of the application of relapse-prevention strategies that are designed to promote improved maintenance of treatment gains in the community.