Characterizing the Problem

If alcohol problems differ from one person to another, whether in degree or in kind, it is crucial to document the differences. Otherwise, any changes subsequent to treatment cannot be compared with the individual's pretreatment status. Some persons coming for treatment, for example, will have high alcohol consumption levels, and others will not. Some will be binge drinkers, and others will be steady drinkers. Some will have experienced many symptoms in connection with their use of alcohol, and others will have experienced few symptoms. Some will have accrued a great many adverse consequences of alcohol consumption, and others will have accrued few consequences. As with other drugs, those who have lower levels of consumption will probably exhibit more variety in the problems they manifest than will individuals with higher levels of consumption (Edwards, 1974).

But even among those with many signs and symptoms, the specific manifestations will differ from one person to the next. For example, DSM-III-R lists nine signs and symptoms of “psychoactive substance use disorders,” of which any combination of three will qualify for a diagnosis (American Psychiatric Association, 1987). Thus, among those qualifying for this diagnosis on the basis of their alcohol use, many alternative combinations of manifestations will occur. In this spirit physicians have been cautioned to “be aware that not every patient that drinks too much (for whatever reason) will be dependent on alcohol, and different patients need different help and treatment” (Edwards and Gross, 1976:1061).

As this warning suggests, the correspondence between level of consumption, pattern of consumption, signs and symptoms, and consequences is not invariably a close one. Some people with high consumption levels will drink steadily, have many signs and symptoms, and experience many consequences, but others will not. The evidence for the relative independence of these dimensions of alcohol problems will be discussed later in this chapter. That they are not necessarily highly correlated with each other, particularly in younger persons (Fillmore and Midanik, 1984; Fillmore, 1987), introduces still more variance into the clinical picture of alcohol problems.

What should emerge from a comprehensive assessment is a detailed picture of the particular kind of alcohol problem manifested by a particular individual at a particular point in time. Of major importance is to describe the person and the problem in terms that are clear and unambiguous. Not only is precision valuable in itself but, if assessment is to be maximally useful, its terms must be clearly understandable to a variety of individuals. The evolving treatment system is complex. Particularly in cases in which the problem is a chronic one (and many alcohol problems will be), a large number of different treatment personnel will encounter particular persons with alcohol problems over time.

In the absence of a clear and unambiguous picture at initial contact it may not prove possible to understand the evolution of an individual's alcohol problem over time, or to make appropriate decisions regarding care for the present and the future. Let us consider a common clinical situation: a patient reports that he “had a problem before, but it got better; now he has developed a problem again, only this time it is a little different.” What sort of problem did he have before? In what sense and to what degree did it improve? In what way is the problem he has now different from the problem he had previously? Skillful interviewing can help to clarify some of these issues, but a comprehensive, understandable, quantitative, recorded account of the patient's earlier status and of his course would be invaluable in providing solid answers.

Precise information regarding the parameters of an alcohol problem is of interest not only to therapists but also to those who manifest the problems. The feedback of assessment data in an understandable form to those from whom it has been obtained is a common and useful practice. Not only does it seem a reasonable courtesy, but there is evidence that feedback can contribute significantly to treatment-seeking behavior.

Thus, in one study, half again as many individuals seeking help for alcohol problems appeared in treatment after receiving a comprehensive assessment compared with those who were not assessed (Annis and Skinner, 1984). In another study, 95 percent of a random sample of such individuals who were given an assessment battery returned for their second appointment, compared with only 56 percent of those who were not given the assessment (Sutherland et al., 1985). General practice patients who completed a brief assessment of their use of alcohol, tobacco, caffeine, medication, and nonmedical drugs during which “feedback was given on how the patient's consumption levels compared with others of the same sex and age” were significantly more likely than those who were not so assessed to query their doctors regarding all of these substances (H. A. Skinner et al., 1985b).

“Taking these [assessment] tests,” commented one group of observers, “could have predisposed patients to attend a second session either because they may have expected to obtain some information about the test's results . . . or because they may have been impressed by the amount of care devoted to them” (Sutherland et al., 1985:212). In confirmation, one patient in an assessment program commented that “it helps you to get a hold of yourself and use your mind to sort out what makes you feel [the way you do] about life” (Segal, 1984). Another said that “it slowed down my thinking process and allowed me to have a good, long look at myself. I now know what I am and what I have to do to improve myself.”

Characterizing the Individual

Alcohol problems do not occur in a vacuum. The individuals who manifest them are at least as different from one another as are ordinary people (Chapter 2). Or perhaps, more different: Keller's law is that “the investigation of any trait in alcoholics will show that they have either more or less of it” (Keller, 1972). A precise and systematic knowledge of the differing characteristics that each individual exhibits at the time he or she is seen for an alcohol problem, as well as a characterization of the problem, is another purpose of assessment.

Eventually, such information will help to unravel which individual characteristics may predispose people to alcohol problems and which are the result of alcohol problems. Beyond these benefits for future research, however, lies the immediate therapeutic utility of such information. Individual characteristics have much to do with a person's acceptance (and, in consequence, the eventual outcome) of various forms of treatment (see the review by Ogborne, 1978). Thus, detailed knowledge of these characteristics is extremely useful in selecting an appropriate treatment.

For example, persons who are well organized and of quite decided opinions may tend to prefer relatively unstructured forms of therapy, whereas those who are disorganized and at a loss may prefer more structured approaches (McLachlan, 1972; McLachlan, 1974; Witkin and Goodenough, 1977; Hartman et al., 1988). Those who prefer structure are more likely to affiliate with programs that provide it, such as Alcoholics Anonymous (Canter, 1966; Reilly and Sugerman, 1967). Those who prefer unstructured settings, on the other hand, may prefer an approach like client-centered or insight-oriented counseling, in which the patient takes the lead and the therapist is relatively inactive. Persons with positive views of themselves may be able to tolerate and benefit from therapeutic approaches that are highly confrontational; those who view themselves negatively may be harmed by such approaches (Annis and Chan, 1983). Persons whose views of the locus of responsibility for alcohol problems (both for developing and for dealing with them) are congruent with the views of program staff may be more likely to sustain treatment (Brickman et al., 1982).

Another aspect of characterizing individuals has to do with their medical and psychiatric status. People with alcohol problems often have medical and psychiatric problems as well (Wilkinson and Carlen, 1981; Ashley, 1982; Popham et al., 1984; Mendelson et al., 1986; Ross et al., 1988). Some of these problems may be the result of alcohol consumption; some may result in drinking (for example, for symptomatic relief); still others may be independent problems. Yet all are important in themselves, requiring clarification and, often, therapeutic attention. To concentrate solely on an individual's alcohol problem and fail to recognize or to deal with a significant medical or psychiatric problem in the same individual is not only poor therapeutic practice but a potential cause for legal action.

There is also evidence that the coexistence of particular problems (e.g., depression, anxiety or panic states, schizophrenia, antisocial personality, drug dependence) may directly affect the outcome of treatment for alcohol problems (Woody et al., 1984; Strayvinski et al., 1986; Rounsaville et al., 1987; Kadden et al., 1990). The effective management of alcohol problems, in other words, may in some instances be contingent upon the effective management of intercurrent problems. Thus, an assessment of medical and psychiatric status should be a standard element of comprehensive assessment. When one considers that alcohol affects both the body and the mind directly, this is hardly a surprising conclusion.

The point that alcohol problems do not occur in a vacuum is paralleled by the point that “no man is an island.” It is important during the assessment process to characterize the person's social context as well as the person. A turbulent social context may entirely negate any attempts at individual treatment and may need to be directly addressed as the initial order of business. Individuals with problematic family or home situations or both are unlikely to sustain participation in outpatient treatment programs (H. A. Skinner, 1981c). If there is a history of marital troubles, some attention may be required in this area. If there have been job-related difficulties, vocational evaluation and training may be prudent. If there has been difficulty in allocating leisure time, or a social support network is lacking, social or recreational counseling may be in order. Thus, obtaining an adequate picture of the social context of the individual who has the alcohol problem is an important purpose of assessment.

Characterizing the manifold aspects of individuality is a highly complex matter, and an exhaustive discussion of all of the parameters that may require address during assessment is not possible here. The committee envisions such a discussion as more appropriately part of a consensus exercise that would consider both the relevance of various parameters and the means whereby they can be effectively measured (see below). What the committee hopes will arise from the foregoing discussion is an appreciation of the necessity to characterize individuals as part of a comprehensive assessment process.

Characterizing the Treatment Population

If each individual in the treatment population were characterized in a similar manner, individual data could be aggregated; with aggregation it becomes possible to characterize the treatment population as a whole. As will be discussed further below, accomplishing such a characterization does not mean that the assessment of each individual must be identical in every particular, a practice that would fail to give due recognition to the diversity of individuals and of the problems for which they are seeking treatment. It does suggest, however, that there should be common data elements in the assessment of all individuals. Common data would permit not only the characterization of the population of a given program but the comparison of one program population with another.

While it is easy to see that the population characteristics of programs especially targeted for particular population groups—women, youth, or ethnic minorities, for example—are likely to differ, it is less apparent that the populations of treatment programs with a more general orientation may differ as well (Pattison et al., 1969; Pattison et al., 1973; Bromet et al., 1976; Bromet et al., 1977; H. A. Skinner and Shoffner, 1978; Kern et al., 1978; Finney and Moos, 1979; H. A. Skinner, 1981c). Location, history, reputation, publicity, accessibility, treatment orientation, cost, staff composition, funding, and other factors undoubtedly enter into the determination of such differences. They are not stable determinants, and so the characteristics of a treatment program population may change overtime. For this reason the occasional assessment of program population characteristics is less useful than their ongoing assessment.

If the characteristics of a program population are known, and the characteristics of the general population from which it is drawn are also known, it is possible to estimate the effectiveness of the program in recruiting its target population. For example, a community assessment service in London, Ontario, saw 14.1 percent of the persons with serious alcohol problems in its catchment area over a three-year period (Malla et al., 1985). A household survey in that area of the province had found that only 3.2 percent of individuals classified as problem or dependent drinkers during the past year had ever received treatment for alcohol problems in their lifetimes (Smart et al., 1980). The authors concluded that “the assessment centre may, over a period of time, increase the penetration rate of a treatment system into the local alcoholic population” (Malla et al., 1985:41).

If comparable data exist for more than one treatment program, between-program comparisons are possible. Two programs may have similar proportions of positive outcomes, but if it is known that the two populations differ on such pretreatment characteristics as, for example, severity of alcohol problems or level of employment, a more exact understanding of the two programs and their relative efficacy is possible. The assessment center noted above (Malla et al., 1985) had a high rate of referral from physicians and employers, while other area programs had high rates of self-referrals and referrals from family and friends; this pattern speaks to differential, and possibly complementary, recruiting from the overall population. Comparable data from all treatment programs would be invaluable in revealing which segments of the community were being served and in planning further services for those who are not entering existing programs.

Planning Treatment for the Individual

Full characterization of a given individual, combined with knowledge of available treatment options, facilitates appropriate, prompt, and effective management of the individual's problem. For example, there is evidence (cf. reviews by Annis, 1986a; W. R. Miller and Hester, 1986) that the results from inpatient and outpatient treatment do not differ for heterogeneous groups of patients. Some (W. R. Miller and Hester, 1986; Saxe et al., 1983) accordingly have advocated that outpatient treatment should be tried first because it is less expensive and that inpatient treatment should be undertaken only if outpatient treatment fails.

But it is well known that individuals with low social stability (as well as other characteristics) are unlikely to sustain participation in outpatient treatment (e.g., H. A. Skinner, 1981c). Thus, rather than a wholesale embargo on inpatient programs for all persons seeking treatment, the more discriminating use of inpatient programs might be envisioned. Those with low social stability, as well as a profile of other indicative features (severe withdrawal symptoms, major medical or psychiatric complications, a markedly noxious environment, crucially aversive temporary circumstances, etc.), might be referred initially to inpatient or residential programs. Others, in more favorable circumstances and with less severe problems, might be referred to outpatient programs (cf. Hoffmann et al., 1987).

To provide another example of the potential utility of pretreatment assessment in assigning individuals to treatment, let us consider a controlled trial in which no advantage was found in the use of a particular treatment (highly confrontational group therapy, or so-called “attack” therapy) in a heterogeneous correctional population (Annis, 1979). Retrospective reanalysis of the data extended these findings. Although there had been no net benefit in the treatment group, in fact some individuals had benefitted and others (in approximately equal numbers) had not. Moreover, data were available to show that these two groups were systematically different.

Those who had benefitted were characterized on initial assessment by positive self-images (determined objectively with appropriate psychometric instruments). Those who failed to benefit—indeed, who appeared to have been harmed by the treatment—were characterized by negative self-images (Annis and Chan, 1983). In future, it is to be hoped that the self-image of persons seeking treatment could be determined in advance, and that only those with positive self-images would be assigned to “attack” therapy.

In other words, assessment prior to treatment forms the basis on which individual patients are matched to particular treatment programs. This point was stressed earlier, but is repeated here for emphasis. Matching is the subject of the next chapter in this report (Chapter 11); the implications of assessment for matching are more fully discussed there.

It is worthwhile to point out that additional information on the individual will need to be gathered by program staff following the selection of treatment in order to plan the individual's ongoing treatment course. In some respects, indeed, treatment involves a continual and ongoing gathering of information on the individual. Pretreatment assessment initiates this aspect of treatment, but information gathering continues throughout treatment.

Guiding Treatment for the Population

Assessment provides information that can be used to develop a clinical data base. “A clinical data base is created when well-defined, discrete, and continuous data elements concerning patients are routinely recorded and coupled with outcome descriptors” (Pryor et al., 1985:623). Given knowledge of pretreatment characteristics and knowledge of the outcome of treatment, a comprehensive picture of individual responses to treatment can be elaborated. This information can then be used to estimate the probable responses of future patients to particular treatments. Their characteristics can be documented during the assessment process, and treatment can be selected on the basis of information about how individuals with similar characteristics have previously responded to the available alternatives.

Such a system has been recommended as the basis for medical care generally (Ellwood, 1988). To manage the large amount of information involved and to provide rapid access to that information, computerization of the clinical data base is logical. Yet it is worth noting that the fundamental model is the human clinician. “The ability of a practitioner to couple the process of patient care to the outcome of a disease is the underlying principle enabling physicians to learn from their previous experience” (Pryor et al., 1985:623). Computerized data bases seem foreign or even outlandish to many. Yet they simply imitate and extend a familiar model, formalizing what is done by good clinicians in the management of patients but doing it with greater scope, capacity, accuracy, and speed.

Such data bases are already in existence for many particular kinds of problems. Tumor registries are perhaps the most familiar example (Laszlo, 1985), but clinical data bases exist for such prevalent problems as cardiovascular disease (Hlatky et al., 1984) and such uncommon problems as systemic lupus erythematosus (Fries, 1976), a severe disease that involves the destruction of connective tissue throughout the body. There is at least one extensive clinical data base for alcohol problems that includes outcome information, that of the Chemical Abuse/Addiction Treatment Outcome Registry (CATOR) (Belille, n.d. [ca.1987]; Harrison and Belille, 1987; Harrison and Hoffmann, 1987).

At present, existing data bases are for the most part not used to guide treatment for populations (but see Fries, 1976). Given the increasing availability of computers there is every prospect that they could be so used. In fact, treatment programs offering different treatments could assemble around a shared clinical data base and use the information contained, in it to guide the selection of treatment for all individuals presenting to the programs collectively (see Chapter 13). For this proposal to be feasible, however, a comprehensive pretreatment assessment must be an integral part of the clinical process.

To summarize: assessment is a comprehensive gathering of information about each individual who is being considered for specialized treatment for alcohol problems. Its purposes include the characterization of the presenting alcohol problem, the individual who has the problem, and the population seeking treatment, and the facilitation of appropriate treatment for all. Although widely advocated, comprehensive assessment prior to treatment is the exception rather than the rule. To facilitate its more general use, the committee in the next three sections discusses its structure, its content, and its administration.

Sequential Assessment

Gathering information, and the attendant processes of recording, storing, and retrieving it for various uses, should not be lightly undertaken. Such activities are costly in terms of time, money, and effort. One wants to be certain, therefore, that all of the information gathered is necessary and that no more information is gathered than is required for the purposes at hand. Accordingly, it is advisable to divide the process of assessment into a series of stages, each of which may or may not lead into the next stage (H. A. Skinner, 1981a; 1981b). This approach, which is called sequential assessment, is graphically portrayed in Figure 10-1.

FIGURE 10-1

Sequential Assessment. As one moves from screening to problem assessment to personal assessment, the extent of information developed is greater but the costs of assessment are also greater. Performing an assessment sequentially ensures that further information (more...)

The initial stage in the assessment sequence for those seeking specialized treatment for alcohol problems is screening. In common with the process of identification in the community sector of treatment (see Chapter 9), the basic questions asked here are (1) whether an alcohol problem is present and (2) whether it requires specialized treatment. This duplication of what may occur in the community is necessary in a specialized assessment setting for alcohol problems because some individuals—those who did not first attend a primary care physician, social agency, or another community setting in which the identification process is available—will seek specialized treatment directly. Of those who do present for treatment, many will prove to have alcohol problems, but some will not. Hence, screening as the first order of business makes practical sense and, in at least some instances, will suggest that the remainder of the comprehensive assessment process is not necessary.

Even if a problem is present, it may prove to be one that can readily be dealt with through brief intervention. Referral to a community setting rather than to specialized treatment can in such instances be made on the basis of screening alone. Although the yield again will be small, the saving of time and effort devoted to subsequent assessment stages even in a small number of cases will be worthwhile.

If screening suggests that the individual probably does have a problem that is likely to require specialized treatment, the next step in the sequence may be thought of as the problem assessment. This stage of assessment represents a major increment over screening in the extent and variety of the information it yields (as well as in the effort and time required to implement it). Because screening has indicated the likelihood that an alcohol problem is present, this next stage of assessment both tests and extends that observation.

Many instruments have been developed which may be utilized for problem assessment (cf. Lettieri et al., 1985b). As discussed in the previous chapter, a single scale instrument is often used for screening purposes. It may be appropriate in the next stage of assessment to utilize a multiscale instrument, such as the Alcohol Use Inventory (AUI) (Wanberg et al., 1977; H. A. Skinner and Allen, 1983a; Horn, Wanberg & Foster, 1987). With its extensive item pool and multiple scales, the AUI, together with other elements of the problem assessment, can provide confirmation or disconfirmation of the screening finding that an alcohol problem exists; moreover, it can help to determine what kind of alcohol problem it might be. Additional effort is expended, but additional information is gained. As is discussed later in the chapter, other measures at this stage of assessment can also be used to provide similarly extensive data on other aspects of the presenting alcohol problem.

Ideally, both the screening stage and the problem assessment stage are uniform in their content for all persons seeking treatment. Such uniformity is desirable because all such persons may or may not have alcohol problems. If no alcohol problem is present, or the problem that is present is appropriate for brief intervention rather than specialized treatment, the assessment process can end.

Alternatively, once the presence of a problem appropriate for specialized treatment has been confirmed, and the nature of that problem has been fully characterized during the problem assessment stage, it is appropriate to move on to the next stage of assessment. As discussed earlier, to determine the most appropriate treatment one must take into consideration not only the characteristics of the problem but those of the individual manifesting the problem. Thus, the third stage of a comprehensive assessment, following screening and the problem assessment, is the personal assessment.

Before beginning this stage of the assessment, however, it is advisable to undertake a specific screening process as the first order of business. Some of the procedures that must be implemented to gather a full complement of data during the course of a personal assessment are among the most extensive and time-consuming in the assessment repertoire. They therefore should not be deployed unless there is preliminary evidence that it is necessary to do so.

For example, confirmation of the presence of a psychiatric disorder may involve the administration of a structured instrument such as the Diagnostic Interview Schedule (DIS) (Robins et al., 1981), or a psychiatric consultation, or both. Before engaging in these complex procedures, it would be appropriate first to screen as quickly and as accurately as possible for the presence or absence of psychiatric problems. The screening could be accomplished by the use of a brief instrument such as the General Health Questionnaire (GHQ) (Goldberg, 1972, 1978; Ross and Glaser, 1989) or the psychiatric scale of the Addiction Severity Index (ASI) (McLellan et al., 1980; McLellan et al., 1985).

Screening for this and the many other substantive areas one might wish to explore during the personal assessment is essential to ensure that the assessment process is parsimonious; that is, that only those dimensions of the individual that require an extensive assessment receive it. There should be variability in the procedures of the personal assessments of specific individuals because there will be variability in the personal areas in which they have problems. With the exception of certain, individual attributes that are sufficiently relevant in all cases to merit routine assessment (e.g., personality), the highly specialized measures would only be utilized if screening indicated a reasonable probability that treatment-relevant information would be gained.

To summarize, the committee views comprehensive assessment as a sequential process that proceeds from one stage to the next if such a progression is indicated. Three stages are proposed. The first is a screening stage , in which the presence or absence of a problem and the likelihood that specialized treatment may be required are determined; this stage is similar to the identification process in the community setting discussed in Chapter 9. The second stage comprises the problem assessment , that is, the characterization of the alcohol problem that screening has indicated is present. The third stage is the personal assessment stage , in which the nature of the individual who is experiencing the problem is fully and uniquely characterized; the emphasis in this stage is on areas in which personal problems are being experienced. The overall goal of the assessment is to produce sufficient information to make treatment-relevant decisions.

Multidimensional Assessment

In the previous section of this chapter, it was suggested that assessment be divided into stages. Each of these stages, however, ideally involves the eliciting of information along several important dimensions rather than along a single dimension. Alcohol problems are complex; the people who manifest them are complex; and these complexities defy simple characterization. Thus, the assessment of alcohol problems should be multidimensional .

To illustrate the principle of multidimensional assessment, let us concentrate for the moment on the problem assessment. The task of problem assessment is to describe as fully (and yet as parsimoniously) as possible the problem or problems with alcohol that an individual may have. From the standpoint of multidimensionality, the relevant question is the following: how many different dimensions are required to provide a reasonable description of a given alcohol problem?

There has been a tendency to rely on only a single dimension, a measure of the individual's use of alcohol, to characterize his or her alcohol problem. This measure can be taken, for example, by a tally of the average number of standard drinks the individual consumes per day. Certainly, this is important information, but such a measure of a person's level of use does not even fully characterize alcohol use. Of additional importance is the pattern of use. If an individual consumes four drinks per day on average, it will make a considerable difference (at least in the clinical picture) whether he consumes them in an hour or two or whether they are spaced out over the course of the entire day. With the former pattern, the individual is likely to become intoxicated; with the latter pattern, intoxication is unlikely.

The pattern of alcohol use in turn can make a difference in the consequences the individual experiences. In a recent study (Kranzler et al., 1990) it was found that both an increased level of consumption and a pattern of consumption likely to result in intoxication independently increased the risk of consequences. Interestingly, it was found that an increased level of consumption was more likely to contribute to consequences in males, while an intoxication pattern of consumption was more likely to contribute to consequences in females. The authors concluded that “these variables, though related, require independent consideration.”

Beyond the daily pattern of use, it is important to have information about the pattern of use over longer periods of time. Some persons do drink at the same level and in the same daily pattern over prolonged periods of time. Others, however, vary both their level and their daily pattern of use quite considerably. Binge drinking is a well-known long-term pattern of alcohol use. It is likely that such long-term patterns have important implications for consequences as well as prognosis; hence what can be termed a history of use is an important element in the characterization of an individual's use of alcohol. Such a history would include information as to the time in life the individual began to drink and the length and circumstances of periods of nonuse, as well as the pattern of use over the last few years prior to seeking treatment.

Thus, an adequate assessment of an individual's use of alcohol would include information on the level of use, the pattern of use, and the history of use. It might be felt that such a comprehensive consideration of alcohol use might suffice to characterize an alcohol problem because there is a general and positive correlation between the use of alcohol, signs and symptoms, and consequences, a correlation that becomes most evident when aggregate data from large groups of individuals are explored and when the problems themselves are longstanding and severe. But treatment is a clinical process that deals with single individuals, one at a time; among individuals, wide variations may be found in the relationship between use, signs and symptoms, and consequences. The vignettes at the beginning of Chapter 2 of this report include individuals (George, Gregory) with low levels of consumption and serious consequences, as well as one individual (Elizabeth) in whom a high level of consumption was associated for a long period of time with no apparent consequences at all.

Disparities between the level of alcohol consumption and the effects of alcohol are also matters of common experience. Some individuals “can't hold their liquor” and become thoroughly intoxicated on small amounts of alcohol which would not faze most social drinkers. Other individuals drink constantly throughout the day, consuming remarkable quantities of alcohol but without exhibiting the least sign of intoxication (so-called “hollow legs”).

That these phenomena are not mere folklore is well substantiated by research that, for example, documents that similar doses of alcohol differ widely in their effects on different individuals, or that even trained observers are unable to identify individuals with elevated blood alcohol levels in from more than a fifth to more than a half of all cases without the aid of such instruments as breathalyzers (Hartocollis, 1962; M. B. Sobell et al., 1979). A study in adolescents found that use-related problems and intensity of drinking were to some degree correlated in the study population but that they were not sufficiently correlated to constitute a single dimension. Rather, they were most accurately viewed as separate dimensions (White, 1987). There have been similar findings in adults (Sadava, 1985). Indeed, a RAND Corporation study found that consequences from drinking were only weakly related either to the amount of alcohol consumed or to the symptomatology of alcohol use (Polich et al., 1981).

All three dimensions appear to contribute useful and independent information to the overall characterization of an alcohol problem. Taken together, they may be seen as illuminating the important question of the severity of the alcohol problem or problems experienced by a given individual. Therefore in assessing alcohol problems, the committee recommends that information be sought along three specific dimensions: (1) the use of alcohol; (2) the signs and symptoms of alcohol use; and (3) the consequences of alcohol use. This multidimensional classification is outlined in Table 10-1.

Table 10-1

Multidimensional Classification of Alcohol Problems.

One of the principal advantages of a multidimensional classification system of this type is that, by providing a more fine-grained, specific characterization of individuals with alcohol problems, it facilitates communication among workers in the field. To know where an individual stands on any one of these three proposed dimensions provides significant information. To know where an individual stands on all three provides a significantly greater degree of information. A more detailed example of a multidimensional classification system has been given elsewhere (H. A. Skinner, 1985, 1988).

If all patients entering treatment were characterized according to a common multidimensional basis, enormous advantages would be realized. For example, the equivalence (or lack thereof) of patients in different treatment programs could readily be established. If the treatment provided then proved to be effective, it would be much more securely known for whom the treatment was effective. Should a program for whatever reason tend to deal exclusively with individuals manifesting only the most severe kinds of alcohol problems, its treatment outcomes would necessarily be viewed differently from those of a program that dealt largely with individuals manifesting problems that were not severe.

An analogous multidimensional classification system has been in place in another field for some time. The tumor-node-metastasis (TNM) system, which was first proposed by P. F. Denoix in 1944, subsequently refined and accepted by the Union Internationale Centre de la Cancer (UICC), and widely implemented from about 1968 on, has been of great utility in dealing with cancer (Harmer, 1977), particularly in evaluating prognosis, planning therapy, and reporting results. It considers three significant events in the life history of a cancer: (1) tumor growth (T), (2) spread to primary lymph nodes (N), and (3) distant metastases (M). Cancers occurring in various regions of the body can be classified according to this system, and the effectiveness of differing modalities of treatment (e.g., surgery, radiation, chemotherapy) can be examined using the classification. Such examinations are predicated on the notion that the effects of treatment will vary depending on the classification of the cancer in the TNM system.

Because of the widespread use of the TNM system, treatment results from multiple tumor registries can be aggregated to provide a vast body of information on which to base judgments regarding optimal intervention. The National Cancer Institute offers a Physician Data Query (PDQ) system in which the TNM classification scheme is used to provide physicians with information on prognosis and with treatment protocols for different cancer sites and stages. The TNM system illustrates how a relatively simple but widely utilized multiaxial assessment scheme can greatly facilitate progress in research and treatment. The potential advantages of developing and using a similar scheme in the treatment of alcohol problems are substantial. Potentially, the enormous fund of clinical experience built up from the multitudes of cases seen every year can be directly brought to bear in a systematic manner upon the disposition of future cases (cf. the earlier discussion of clinical data bases).

To some it will come as a surprise that the characterization of the alcohol problems of individuals seeking treatment is not regularly carried out along multiple dimensions. As has already been indicated, however, assessment itself is not commonly carried out in treatment programs. Even in research studies, which explicitly aim at widespread generalizability, characterization of the alcohol problem is often incomplete.

For example, the much-publicized collaborative studies by Swedish and American investigators that have examined the contribution of genetic factors to alcohol problems (Bohman et al., 1981; Cloninger et al., 1981) have been criticized on these grounds: “[T]he data reflect only obvious and reportable instances of the consequences of insobriety. There are no measures of the quantity or frequency of drinking or of the social and personal consequences of private or unreported drunkenness . . . [I]t is quite likely that the obtained results are an artifact of the criteria of abuse” (Searles, 1988:159-160). Further, in a review of 48 alcohol treatment outcome studies published in the period 1980-1984, it was found that reporting on all three dimensions was deficient: only 56.3 percent quantified pretreatment drinking levels; none reported whether the symptom of physical dependence was present or absent; only 18.8 percent reported on alcohol-related arrests; and only 33.3 percent reported on whether or not there had been prior treatment for alcohol problems (M. B. Sobell et al., 1987).

Although the need for a multidimensional approach has been discussed here largely in terms of the characterization of alcohol problems, multidimensionality is a broad structural principle that ideally should be applied to all stages of assessment. For example, a screening instrument such as the AUDIT, although brief, contains questions that cover all three of the axes suggested above for the problem assessment (Saunders and Aasland, 1987; Babor et al., 1989). In like manner the personal stage of assessment requires that multiple aspects of the individual with the alcohol problem be assessed. Agreement on and consequent uniform adoption of a standard multidimensional problem assessment and personal assessment seem essential to progress in clinical services and in research.

To review briefly, more than a single kind of information is required for adequate assessment at each stage of the assessment process. The dimensions along which problem assessments, for example, should proceed include aspects of the individual's use of alcohol, the signs and symptoms of alcohol use, and the consequences alcohol use. This is the principle of multidimensional assessment.

Content of Screening

Screening is by design a brief process (Saunders, 1988). As noted earlier, it must answer two questions: (1) whether an alcohol problem is present and (2) if so, whether it is likely to require brief intervention or specialized treatment. Examples of instruments that accomplish these purposes are the Alcohol Clinical Index (ACI) (H. A. Skinner et al., 1986; H. A. Skinner and Holt, 1987), the CAGE questionnaire (Ewing, 1984), and the AUDIT (Saunders and Aasland, 1987; Babor et al., 1989). Many alternative screening methods are available (Babor and Kadden, 1985; Babor et al., 1986). The reader is referred to Chapter 9 for a more extensive discussion of screening and brief intervention.

Content of the Problem Assessment

In the problem assessment stage, content must address the three dimensions of the multidimensional characterization of alcohol problems. With respect to specifying the prospective client's use of alcohol, care must be taken to look broadly at a variety of aspects, including especially the level of use, the pattern of use, and the history of use. Various techniques have been developed for taking a drinking history. These techniques may be divided into three broad classes: (1) retrospective methods that gather information about drinking over a specified time interval in the past, using a self-report questionnaire or a structured interview; (2) prospective methods in which the individual is asked to monitor alcohol use and record it in a daily diary; and (3) laboratory determinations such as various tests performed on body fluids (blood, urine, saliva, sweat) and breath tests. Extensive reviews of these methods have been published (Babor et al., 1987; O'Farrell and Maisto, 1987; L. C. Sobell et al., 1987); they will be discussed further below.

Multiple methods are also available to survey the signs and symptoms of alcohol use. Signs and symptoms form the basis of both the 10th edition of the diagnostic manual of the International Classification of Diseases (ICD-10) and DSM-III-R. Methods of making diagnoses within the contexts of these systems, such as the DIS (Robins et al., 1981) or its computerized version (Blouin, 1986), may be useful.

A number of self-report questionnaires have also been developed (Edwards, 1986; Davidson, 1987) that systematically review symptoms of alcohol use. The Severity of Alcohol Dependence Questionnaire (SADQ) (Hodgson et al., 1978; Stockwell et al., 1983) has been extensively used outside of North America, and a considerable amount of data on its measurement properties has been generated. The Alcohol Dependence Scale (ADS) has been widely used in North America (H. A. Skinner and Allen, 1982; Horn et al., 1984). Many treatment outcome studies and almost all epidemiological studies have included alternative scales of this kind.

With respect to the third axis, consequences of alcohol use , options abound in this arena as well. The Michigan Alcoholism Screening Test (MAST) (Seizer, 1971; Skinner, 1979), the Alcohol Use Inventory (AUI) (Wanberg et al., 1977; Skinner and Allen, 1983a; Horn et al., 1987) and the Addiction Severity Index (ASI) (McLellan et al., 1980; 1985) are largely concerned with the assessment of this axis. The ASI produces information in the areas of drug abuse, medical problems, psychiatric problems, legal problems, family and social problems, and employment and support problems. Beyond this, there are multiple direct ways of assessing all of these dimensions (e.g., routine medical and psychiatric examinations), as well as scores of psychometric instruments (cf. Lettieri et al., 1985b).

Content of Personal Assessment

Both the screening and problem assessment stages of the comprehensive assessment process may lead to the determination that a problem is present, that it requires a specialized intervention, and that it is of a particular kind. What remains to be specified in order to select the optimum treatment approach is the individual who presents the problem. Some of this information is needed to permit description of the individual and, through the aggregation of individual data, description of the program population. Additional information may be needed to gauge the prognosis, to assist in matching individuals to appropriate treatment, to determine whether prevalent comorbidities are present, to help plan living and working circumstances, to understand certain etiologic possibilities, and for other purposes.

There may be some overlap between information gathered as part of the personal assessment and information gathered as part of the problem assessment when the consequences of alcohol use are being considered. For example, alcohol problems are often manifested in the vocational area. On the other hand, a person with alcohol problems may quite independently have vocational problems, which it may be important from a therapeutic perspective to know about. Fundamentally, the problem assessment looks at those problems that are reasonably attributable to alcohol consumption, while the personal assessment looks at those problems the individual has whether or not they are attributable to alcohol consumption . Whatever area of overlap exists is tolerable and ensures that all existing problems will come to light.

Contingent dimensions in personal assessment As has been noted, personal assessment is a particularly complex area. Multiple dimensions that are descriptive of various aspects of the individual might be relevant. Unfortunately, the assessment of any one of these dimensions is often complex and time-consuming. Therefore (in keeping with the principle of sequential assessment), screening should be an important part of the personal assessment process. The purpose of such a screening is to identify which of the many potential dimensions of personal assessment require a more thorough investigation. For example, are there family problems, marital problems, vocational problems, sexual problems, personal problems (problems with assertiveness, with social skills, with particular situations), medical problems, or psychiatric problems? If screening indicates the presence of these problems, a more intensive assessment can be undertaken; if the problems are absent or minimal, no further evaluation may be necessary (and there is a considerable saving of time and effort over the routine administration of intensive assessment in all of these areas). An example was given earlier of screening for psychiatric problems with the GHQ or the ASI and administering the full DIS only when screening is positive.

Noncontingent dimensions in personal assessment With regard to other dimensions of personal assessment, detailed examination should not be contingent on screening. For example, the collection of demographic information is important both to identify the individual and to help understand prognosis; age, marital status, and social class are examples of such data. Demographics are also among the principal descriptors of populations; hence, there is a particular need for completeness and uniformity in their compilation.

Yet experience suggests that the collection of demographic data is not as simple a matter as it may seem. Many programs collect age by category, and the categories may not be consistent from one program to another (e.g. persons aged 35-45 vs. persons aged 30-40). Where there is an option to designate that a person is unmarried but living with someone in a stable relationship, this information will be recorded; if the option does not exist, however, the person will probably be designated as single. If data categories vary from program to program, data cannot be aggregated among different programs, nor can different programs be compared.

The use of tobacco and other drugs by persons seeking help for alcohol problems is another area of personal assessment that should be surveyed as a matter of routine rather than on the basis of contingent screening. Available data suggest that the correlations between use of alcohol and use of tobacco and other drugs are regularly impressive. With regard to tobacco, one study found that 80 percent of patients in an addiction treatment setting were daily cigarette smokers (Kozlowski et al., 1986); another study of the same population by different means found that 78.3 percent of patients presenting for new episodes of service qualified for the DSM-III diagnosis of tobacco use disorder (Ross et al., 1988). An earlier inpatient study on an alcohol and drug unit found a correlation of 0.74 between amount of alcohol drunk and number of cigarettes smoked for males and females diagnosed as “alcoholics”; the correlation did not hold for individuals not so diagnosed (Maletzky and Klotter, 1974). In an ambulatory medical care setting, 85.1 percent of new patients who answered five or more of the MAST questions affirmatively were smokers, whereas only 47 percent below this cutoff point were smokers (Cyr and Wartman, 1988).

In the large Epidemiologic Catchment Area survey, in which approximately 3,000 persons were diagnostically interviewed in each of five sites, almost one in every five persons (18 percent) who met criteria for alcohol problems also met diagnostic criteria for some type of drug problem (Helzer and Burnham, in press). In aggregate data from three of the five sites, the proportion was much higher (28 percent) among young men (Robins et al., 1984). A study of all persons seeking help from a large alcohol and drug treatment center in Toronto found that exactly the same proportion as in the ECA study (18 percent) of individuals qualified for diagnoses of both alcohol and drug dependence (Ross et al., 1988). A study of a large number of adult women admitted to inpatient treatment for alcohol and drug problems in the United States (N = 1,776) found that, although 61.1 percent of women over the age of 30 reported using only alcohol, the corresponding proportion for women under 30 was 31.2 percent (Harrison and Belille, 1987).

Thus, there is a great deal of evidence that persons with alcohol problems are highly likely to have problems with tobacco use and quite likely to have problems with drug use. It also appears, from the data cited here and from the observations of clinicians in the field, that the concurrent use of alcohol and drugs is a more prevalent phenomenon among younger persons entering treatment. These findings suggest that the use of tobacco and of other drugs should be assessed routinely in those seeking treatment for alcohol problems.

The personal stage of assessment should also examine data that are relevant to matching. For example, knowing what goals are seen as important by each individual entering treatment can be quite useful in the matching process. In one program that made consistent attempts to match, a treatment goals inventory was extensively utilized (Glaser and Skinner, 1984:76-77, see page 291). Demographics are another dimension that have utility for matching purposes; information collected by AA, for example, suggests that it is a less attractive option for the young and for women (Ogborne and Glaser, 1981; J. K. Jackson, 1988; but see Emrick, 1987). Matching individuals to appropriate treatments is discussed in some detail in Chapter 11.

Personality may be another area that should be routinely assessed. The intent is not to search for evidence of an “addictive personality” (i.e., the one personality pattern that characterizes persons with alcohol problems) because this population is heterogeneous in its personality characteristics (Syme, 1957; Mogar et al., 1970; Partington, 1970; H. A. Skinner et al., 1974; Stein et al., 1977; Barnes, 1979; Cloninger, 1987). Rather, in keeping with the assessment goal of obtaining treatment-relevant information, the intent is to identify personality factors or patterns that may affect in a significant way which treatment programs individuals are likely to find suitable and which they are likely to reject.

Certain pathological personality types, especially antisocial personality but also borderline personality, have been found to have prognostic value in this regard (Schuckit, 1973; Vaillant, 1983; Nace et al., 1986; Cloninger, 1987). Normative factors or traits of personality may also have value in predicting appropriate matches and outcome (see the discussion earlier in this chapter and also Beutler [1979]). It is difficult to assess personality efficiently as a general construct, simply because the scope of the concept is extremely broad. However, potentially suitable structured instruments of reasonable length are available—for example, the Personality Research Form (PRF) (Jackson, 1974), and the Sixteen Personality Factors Inventory (16PF) (Institute for Personality and Ability Testing, 1986).

Intelligence has some influence both on matching and on outcome (Gibbs, 1980). Fortunately, there is a high correlation between intelligence and some measures that can be efficiently administered, such as vocabulary tests. Full-scale intelligence testing with such standard instruments as the Wechsler Adult Intelligence Scales (WAIS) (Wechsler, 1981) is quite exacting, however, and if used at all should be used sparingly.

Cognitive functioning is a related and important area to be assessed, because many kinds of treatment require the ability to process information at a high level of abstraction. Some persons with alcohol problems prove to be cognitively impaired (Wilkinson and Carlen, 1981; Wilkinson, 1987), and this information should enter into the process of treatment selection (Wilkinson and Sanchez-Craig, 1981). The precise evaluation of cognitive functioning is complex and time-consuming, but it is possible to screen briefly for cognitive deficit before embarking on a full-scale evaluation (Barrett and Gleser, 1987; Kiernan et al., 1987; Wilkinson, 1988).

Some aspects of the assessment of the individual may be partly contingent and partly noncontingent. The area of the family is an example. Family circumstances may be an important determinant of alcohol problems and, even in instances in which they do not figure in the etiology of alcohol problems, they may nevertheless affect the outcome of treatment (Cronkhite and Moos, 1978, 1980; Moos et al., 1979). Family history is also important. A significant minority of individuals with serious alcohol problems come from families in which a parent has a similar problem (Cotton, 1979). In some samples potentially treatment-relevant differences have been found between individuals whose families do and do not have a history of alcohol problems (Penick et al., 1987). Recent work in the genetics of alcohol problems (Cloninger, 1987) has enhanced interest in familial factors.

Consequently, data on the family are relevant to the assessment of individuals with alcohol problems. It seems reasonable to recommend that a family history of alcohol problems be taken on all persons entering treatment. A full-scale family assessment process, on the other hand, is sufficiently complex (Jacob, 1988) that it should probably be contingent on the results of screening. Although it is often confidently asserted that, in cases in which a member of a family has an alcohol problem, the family itself has a problem, the assertion lacks strong empirical support.

That drinking, and perhaps especially episodes of heavy drinking, may be related to specific life events (deaths, births, marriages, job losses, new jobs, etc.) is a matter of common knowledge. Yet individuals may not report these events unless they are specifically questioned about them. Moreover, it may be the combined weight of such events, whether positive or negative, rather than a specific event that acts as the trigger to particular coping behaviors, including the consumption of alcohol (Holmes and Rahe, 1967; Paykel et al., 1971; Holmes and Masuda, 1973). A systematic attempt to assess the number and kinds of life events may therefore be useful, and a number of instruments have been developed for this purpose.

Related to but different from life events are situational factors. These tend to be more complex constellations of individuals, feelings, and circumstances than are life events, but may also be closely related to individual variations in alcohol consumption. Recent studies (Marlatt and Gordon, 1985; Annis, 1986b; Annis and Davis, 1989) have provided both a theoretical and an empirical basis for understanding and dealing with situational factors, including the development of specific assessment devices. The ubiquity of situational stress suggests that careful consideration should be given to the inclusion of such instruments as the Inventory of Drinking Situations (IDS) (Annis, 1982; Annis, Graham, and Davis, 1987) as a noncontingent element of personal assessment.

The preceding review of the potential content of assessment, which is intended to be more heuristic than exhaustive, nevertheless suggests an embarrassment of riches: many instruments are available to assess the relevant dimensions. The development of reliable and valid new instruments should in no way be discouraged, as the development of assessment must be an evolving process. But a pressing need already exists for shaping what is now available into a reasonable and practical assessment procedure. To facilitate this development, an outline of the assessment discussed above may be helpful (Table 10-2).

TABLE 10-2

Comprehensive Assessment of Persons Seeking Treatment for Alcohol Problems.

The translation of such an outline into concrete reality—a fully specified comprehensive assessment battery in use across the universe of treatment programs—has both scientific and political aspects. The publication of two monographs on assessment research (Lettieri et al., 1985a, 1985b) which summarize much of the literature to date and provide examples of instruments is a step in the right direction, and every effort should be made to secure the widespread distribution and serial updating of these useful compendia. A caution should be mentioned, however, that the standard means of conveying information (publication in journals and presentation at scientific conferences) will not suffice to bring about the development of comprehensive assessment batteries. Advantage should be taken in this effort of all that is known regarding the dissemination of innovations (cf. Backer et al., 1986).

Leadership of a high order will also be required to produce the desired result of comprehensive and at least partly uniform assessment across all programs. Widely subscribed consensus conferences devoted to the issue are a possible mechanism. The example of large and prestigious programs and of leaders in the field undoubtedly would have a positive influence. Ultimately, the financing of treatment could be made contingent upon the appropriate classification of individuals who are accepted for treatment; a reasonable aspect of accountability is to require a detailed documentation that a type of problem requiring treatment is being dealt with. The present practice of utilization management operations is directed in part toward this end. Surely it would be advisable for the alcohol treatment field itself to assume responsibility for such accountability.

Obtaining Valid Assessment Data: Problems in Providing Information

Validity, as discussed above, is a desirable property of information. Most of the information obtained during the assessment of an individual seeking assistance for an alcohol problem comes from the individual himself. Although self-reports are a standard procedure in all human therapeutics, they take on a special coloring in the alcohol field. The use of alcohol may negatively affect the validity of self-reported information, especially regarding alcohol consumption.

Contemporary research provides some reason for concern about this issue. Persons with elevated blood alcohol levels tend characteristically to underreport their alcohol consumption (M. B. Sobell et al., 1979). Significant discrepancies have been noted in some recent studies between self-reports of alcohol consumption and alternative lines of evidence, such as laboratory tests or information obtained from collateral informants (spouses, relatives, and close friends) (Orrego et al., 1979; Watson et al., 1984; Peachey and Kapur, 1986; Fuller et al., 1988). These findings have given additional impetus to efforts to find alternative or supplementary methods to self-reports for determining an individual's alcohol consumption.

Methods of obtaining data on alcohol consumption were reviewed in Chapter 9 in relation to their potential use in screening. To recapitulate briefly, there is general agreement that in many circumstances screening questionnaires are significantly more accurate than physical examination findings and currently available laboratory measures (including body fluid examinations and biological markers). Questionnaires also have numerous practical advantages. On the other hand, the validity of the self-reports involved in questionnaire methods remains a matter of concern, and there is some indication that more recently proposed biological markers may possess significant advantages over those that are currently available.

There are particular aspects of assessment that favor the use of laboratory methods in conjunction with other methods. One is the need for a validity check on the assessment process itself. Assessment is lengthy and complex, and its validity might be adversely affected by a high blood level of alcohol. For example, the validity of tests of cognitive functions could be confounded by alcohol intoxication. The use of a breathalyzer test as a preliminary to assessment, in recognition of the fact that high blood alcohol levels are not clinically apparent in a significant proportion of cases (M. B. Sobell et al., 1979), may be advisable.

Because tolerance, physiological parameters (e.g., lean body mass), pharmacodynamic parameters (e.g., whether the blood alcohol curve is rising or falling) and other factors make the exact equation of a particular blood alcohol level with actual impairment in a particular individual quite difficult, an arbitrary level—often the legal limit for impaired driving in the jurisdiction of the assessment program—is generally used. Those achieving a level above the legal limit are not assessed but given a return appointment. If, as a matter of usual practice, persons given an assessment appointment are cautioned that they will be tested and that elevated blood alcohol levels will require a postponement, few individuals will need to be turned down. Breathalyzers are well suited for this role, as they are generally quite accurate and easy to use, do not require body fluid collection, are inexpensive to operate (although an initial investment is required), and provide immediate results.

A second consideration is that, although their generally low sensitivity makes biological markers less useful than questionnaires for screening, they may be very useful for monitoring (Kristensen et al., 1983; Schuckit and Irwin, 1988). Thus, establishing a baseline level for one or more markers may be a reasonable assessment procedure, although it should be entertained primarily if there is a specific intent to repeat the measure subsequently. Under these circumstances the time delay involved in a laboratory determination is not a significant problem, and the expense of the test may be quite justifiable.

A sensitive, specific, responsive marker would be a most welcome supplement in assessing individuals with alcohol problems. Some of the newly proposed markers (desialated transferrin, hemoglobin adducts, etc.) are promising, and work continues. At present, however, there is no viable alternative to reliance on verbal self-report for the assessment of alcohol consumption in clinical settings. Fortunately, current research has shed some light on how the validity of self-reports may be enhanced.

Such research (H. A. Skinner, 1984; Babor et al., 1987; O'Farrell and Maisto, 1987; L. C. Sobell et al., 1987) suggests that the concern over whether verbal reports are either valid or invalid has been somewhat misleading. Verbal reports are inherently neither valid nor invalid; rather, their validity varies, depending on circumstances. Table 10-3 contrasts those circumstances in which valid and invalid self-reports are likely to be forthcoming.

TABLE 10-3

Factors Influencing The Validity Of Self-Reports.

Attention to the matter of accuracy is clearly of the greatest importance. Not only is an inaccurate assessment not worth doing but, insofar as treatment planning may be based on it, it is potentially harmful. A considerable amount of effort must be expended to ensure that assessment information is accurate and remains so over time. In many clinical settings minimal effort is currently expended toward this end, either because the information the patient gives is presumed to be completely accurate or because the opposite presumption is made. Neither of these presumptions is correct. Programs must come to see themselves as active collaborators with individuals seeking their services to produce the most accurate information possible.

Obtaining Valid Assessment Data: Problems in Gathering Information

Problems with the validity of assessment information can arise not only from those who provide the information but also from those who gather it. Two kinds of factors that operate on assessors, ideological factors and financial factors, have been identified as posing particular threats to the validity of assessment data. They arise from the circumstance, common because it is practically convenient, in which assessment data are obtained by therapists who also provide treatment in a given treatment program.

Those who provide treatment often are, and perhaps ideally ought to be, enthusiastic about the services they provide. It is likely that their enthusiasm will both sustain them in the arduous task of providing treatment and will contribute to positive treatment outcomes. But their enthusiasm may at the same time adversely affect their ability to perceive other aspects of treatment objectively. One such aspect is treatment outcome. Therapists' perceptions of the results of their own efforts are traditionally considered suspect, and it is felt that obtaining objective data on outcome requires that such judgments be made by others. In technical language, this is referred to as the necessity for those who determine outcome to be “blind” to the treatment received by each individual.

The enthusiasm that colors perceptions of treatment outcome may also color perceptions of the suitability of a given patient for a given treatment. Invested in their work, therapists may tend to perceive more individuals as suitable candidates for the treatments they provide than is actually the case. Thus they may not conduct valid assessments.

In a pertinent study (Savitz et al., 1973), hypothetical patient profiles were constructed from actual patient data and were submitted to three groups of staff for disposition. One staff group was not connected with any particular treatment program but ordinarily made dispositional decisions on court-referred cases to a variety of treatment programs. This group assigned 45 percent of the hypothetical patients to treatment A and 35 percent to treatment B. But staff from treatment A assigned 93.8 percent of patients to treatment A, and staff from treatment B assigned 55 percent of patients to treatment B. That is, staff committed to a particular treatment program were significantly more likely to refer patients to that program than staff who were not so committed.

Another potential source of bias is the prospect of financial gain. Programs are often reimbursed on a fee-for-service basis. Under this arrangement the more service that is provided, the greater the profits of the provider. To refer a potential patient elsewhere under this type of reimbursement schedule is to turn away income. Financial incentives operating in the opposite direction may influence treatment provided under fixed-price prepayment schemes; here, the more service provided, the smaller the profits of the provider. Although many individuals involved in providing treatment do not deliberately base their treatment dispositions on financial considerations, it may be that such considerations influence them in subtle ways. An exception is the utilization management company, in which the influence of financial considerations upon the provision of services is explicit and paramount. Contracted to ensure that effective services are provided at the least cost, such companies are of increasing importance in alcohol treatment (Korcok, 1988; Lewis, 1988).

In a pertinent study of possible financial influences on assessment (Hansen and Emrick, 1983), five individuals were carefully trained to give histories about their use of alcohol and their concerns about that use; the levels of use and concern chosen by the investigators, although not negligible, did not objectively indicate either the presence of serious alcohol problems or a need for inpatient treatment. Each of these individuals then presented himself or herself to six local agencies that were selected on the basis of their aggressive recruitment practices.

In 59 percent of the ensuing assessments, a diagnosis of “alcoholism” was made. In 100 percent of the diagnosed cases, total lifetime abstinence from alcohol was recommended, and in 53 percent inpatient treatment was recommended. The only agency that made consistently correct dispositional decisions was also the only not-for-profit agency of the six. The researchers concluded that “people who seek out someone to evaluate their drinking behavior may be told they are ‘alcoholic' when they are not and are subsequently influenced to receive very expensive, unnecessary, and perhaps harmful treatment” (Hansen and Emrick, 1983:177).

Other data corroborate the potential effect of financial considerations on the provision of treatment. A Citizens League study in Minnesota found that treatment selection was closely related to reimbursement arrangements. Even though inpatient treatment was more than 3.5 times more expensive than outpatient treatment, programs that were reimbursed on a fee-for-service basis treated the great majority of their patients in inpatient treatment programs. The one program that was reimbursed on a fixed prepayment basis treated the majority of its patients in its outpatient treatment program. The Citizens League concluded that “treatment funding is an important factor affecting where people are referred to for treatment, and what services are provided them” (Boss, 1980:15).

That financial or ideological biases may be reasonably widespread is suggested by data on the level of cross-referral among alcohol treatment programs. In two studies that looked at cross-referral in large numbers of persons seeking treatment, no instances could be documented among 600 persons in treatment (Pattison, 1974:640) nor among 6,805 persons in treatment (Glaser et al., 1978:209-210). The reasons for the lack of cross-referral were not explored in either study on a systematic basis. Further studies are needed. Cross-referral probably does exist, but the rate of cross-referral may not be high.

If the purpose of assessment is to provide objective data to serve as a basis for the appropriate referral of those seeking treatment to the variety of treatment programs that are available, the foregoing information is not encouraging. An obvious remedy for ideological and financial biases would be to follow the lead proposed by the common practice in outcome determination and to ensure that assessment and treatment selection were not in the hands of therapists who might be ideologically or financially tied (or both) to particular treatment programs. In other words, perhaps the assessment (and consequent referral) process should be functionally independent of the treatment process.

Experience with independent assessment has been limited but positive. Following its recommendation by Ontario's Addiction Research Foundation in 1978 (Marshman et al., 1978), the establishment of independent assessment/referral centers has been a policy of the province of Ontario in Canada. As of November 1987, 35 such centers were in operation. In comparison with treatment programs in which assessment is not independent of treatment, such centers:

• attract a younger population with less serious problems;
• attract clients from a wider array of referral sources;
• are more likely to attract persons with serious alcohol problems;
• are more likely to provide continuity of care; and
• are more likely to involve family members in the therapeutic process (Malla et al., 1985; Rush, 1987).

In the state of Minnesota, the financial independence of assessment from treatment is mandated by law. Partly as a result of the Citizens League report quoted earlier (Boss, 1980), Minnesota Statute 254B was enacted by the state legislature and went into effect on January 1, 1988. This law created a consolidated treatment fund for public assistance recipients, the monies of which were exempted from various federal and state entitlement provisions. As a consequence these funds could follow the patient to whatever treatment was indicated, rather than their availability determining where and how treatment would be provided. From an assessment standpoint the law required that the county shall provide a chemical use assessment “. . . for all clients who seek treatment or for whom treatment is sought for chemical abuse or dependency. The assessor shall complete an assessment summary . . . for each client assessed for chemical dependency treatment services. The form shall be maintained in the client's case record . . . An assessor under contract with the county shall have no direct financial interest or referral relationship resulting in shared financial gain with a treatment provider ” [emphasis supplied].

Although Minnesota's approach is of recent origin, the preliminary results are encouraging. There is already evidence that individuals seeking treatment are being directed more frequently to low-cost alternatives and that more treatment is being provided at less overall cost (Minnesota Chemical Dependency Program Division, 1989). These findings are consistent with experience elsewhere in the provision of medical care (Feldstein et al., 1988). In neither instance has it been determined that equivalent effectiveness necessarily accompanies a lower overall cost of treatment, but Minnesota has also mandated a uniform evaluation system and plans to address this question as well. Another state, Michigan, is moving toward a similar approach to treatment, involving independent pretreatment assessment and outcome monitoring (Allo et al., 1988).

The problems of financial and ideological bias are well known and at the very least require attention. At issue is whether the independence of the assessment process from the treatment process is required to prevent these biases from operating in a significant manner. The broadly accepted model of complete independence of outcome determination in efficacy studies suggests that such independence may be important, particularly if reassessment is used to determine outcome. On the other hand, objections have been raised on such practical grounds as the delays that may occur in entering treatment and the imposition of an additional structure (independent assessment) between the individual and the treatment he or she is seeking. Rather than settling the issue on a prima facie basis, a pilot study involving comparisons of referral patterns and of the results of treatment between independent and nonindependent assessment would be instructive.

In the interim it seems possible that an administratively distinct assessment unit within an overall treatment program could provide some level of assurance that due protection is available from the effects of ideological and financial biases. This assurance would be more credible if the treatment program provided a multiplicity of treatments that were diverse in their treatment philosophies and costs. The arrangement could also provide built-in safeguards such as the regular and independent monitoring of disposition patterns and structural provision for cross-referral to other treatment programs.