increases with severity of dependence (given that greater dependence usually means increased alcohol intake and diminished responsiveness to social controls), such disabilities can occur in the absence of dependence.

Concepts of alcohol abuse and dependence have continued to evolve since the formulation of ADS. For example, the DSM-III (American Psychiatric Association 1980) diagnosis of alcohol abuse required a pattern of pathological alcohol use and related impairment of social or occupational functioning over a period of 30 days; diagnosis of alcohol dependence required the presence of tolerance or withdrawal symptoms in addition to a pattern of pathological alcohol use or impaired social and occupational functioning. In DSM-III-R (American Psychiatric Association 1987), alcohol abuse is the central diagnosis; alcohol abuse is indicative of an ongoing but subthreshold alcohol problem. The DSM-III-R diagnosis of alcohol dependence does not require the presence of alcohol withdrawal syndrome or tolerance. All or most of the proposed DSM-IV

criteria for alcohol abuse are expected to be distinct from dependence criteria (Schuckit et al. in press); in addition, the proposed dependence criteria would again emphasize physiological indicators of dependence (Grant and Towle in press).

Findings from a study by Hasin et al. (1990) support the proposed distinction between DSMIV alcohol abuse and dependence. Using a data set for males aged 21 through 49 years, the investigators found that 70 percent of the subjects who initially met DSM-III-R criteria for alcohol abuse also met criteria for current alcohol abuse or alcohol abuse in remission 4 years after the original diagnosis. Diagnoses of the remaining 30 percent of the patients changed from alcohol abuse to dependence during the same period. The investigators concluded that alcohol abuse and dependence represent distinct diagnoses and that abuse need not be simply a prodromal state of alcohol dependence.

As diagnostic criteria have continued to evolve, interviews used to formulate diagnoses

have also evolved. Such diagnostic interviews as the Diagnostic Interview Schedule (Robins et al. 1981) and the Schedule for Affective Disorders and Schizophrenia (Endicott and Spitzer 1978), were historically used to assist diagnosis according to the now-superseded DSM-III. More recently, the Structured Clinical Interview for DSM-III-R (SCID) (Spitzer et al. 1988) has been used to formulate diagnoses according to DSMIII-R criteria. For the current round of DSM and ICD revisions, new diagnostic interview schedules have also been developed and continue to be tested (Grant and Towle 1990). Among these are the Composite International Diagnostic Interview: Authorized Core Version 1.0 (CIDI-Core) (World Health Organization 1990), the Schedule for Clinical Assessment in Neuropsychiatry (SCAN) (Wing 1989), the Alcohol Use Disorders and Associated Disabilities Interview Schedule (AUDADIS) (Grant and Hasin 1990), and the Structured Clinical Interview for DSM-III-R and DSM-IV—Alcohol and Drug Version (SCID-A/D) (Hasin and Grant 1991)—all of which yield diagnoses according to the new nomenclature. Of these, AUDADIS and the CIDI-Core were developed primarily for epidemiologic research purposes, whereas the SCAN and SCID-A/D were constructed primarily for use in clinical settings. Unlike the SCID, the SCAN, SCID-A/D, and AUDADIS distinguish ICD-10 diagnoses of "harmful use of alcohol," "alcohol dependence syndrome," the alcohol withdrawal state, and alcohol intoxication, together with the DSM-III-R diagnoses of alcohol abuse and dependence (Grant and Towle 1990).

(usually questionnaires or tests), sometimes supplemented by laboratory tests. The information obtained through patient assessment is used to characterize the severity of alcoholism and to determine the appropriate setting and intensity for treatment, establish treatment goals and strategies appropriate to individual patient needs, and facilitate outcome measurement (Allen 1991). Table 3 outlines the role of assessment in alcoholism treatment according to a continuum of clinical care and lists examples of instruments appropriate for different stages of the continuum.

Several of the many instruments used in patient assessment, including the Addiction Severity Index (ASI) (McLellan et al. 1980), have been the subject of recent analyses. The ASI evaluates seven domains of life functioning: alcohol consumption, other drug use, physical health, psychiatric adjustment, family and social functioning, legal status, and employment. For each of the domains, an interviewer rates the severity of a patient's adjustment problems on a 10-point scale, incorporating both objective and subjective data provided by the patient. Differentially weighting the items, the examiner then formulates a composite score. The ASI can be readministered periodically following and after treatment to assess progress.

Screening (the process of identifying persons who are developing or are at risk for developing alcoholism) and diagnosis are preliminary steps toward alcoholism treatment.

Patient Factors:
Assessment and

Assessment Instruments

Screening (the process of identifying persons who are developing or are at risk for developing alcoholism) (see chapter 13) and diagnosis are preliminary steps toward alcoholism treatment. Conducted by both epidemiologic researchers and clinicians in both community and clinical populations, screening and diagnosis usually precede patient assessment-the determination of patient characteristics that have implications for choice of treatment and treatment prognosis.

Patient assessment, which is usually conducted after a patient's admission to treatment, is accomplished through the use of diagnostic and clinical interviews and assessment instruments

Substantial evidence has been offered (McLellan et al. 1985) for the validity and general reliability of the ASI and a recent study (Hodgins and El-Guebaly 1992) reported favorably on the ASI's validity and reliability for assessment of alcohol and other drug abuse patients with concurrent psychiatric disorders. The findings of Hodgins and El-Guebaly support those of Kadden et al. (1989), who demonstrated that scores on the psychiatric severity scale could assist in assigning patients to treatment. More recently, Litt et al. (1992) found that patients high in psychiatric severity who were assigned to coping skills training achieved better outcomes than those who received interactional therapy. These results indicate that the ASI may be especially useful for addressing the needs of patients with

concurrent psychiatric problems a group that historically has responded poorly to traditional interventions (McLellan et al. 1981, 1983).

Another useful instrument in determining treatment intensity is the Clinical Institute Alcohol Withdrawal Scale (revised) (CIWA-Ar) (Sullivan et al. 1989), used to assess whether a patient requires medication to manage the alcohol withdrawal syndrome and, if so, the dosage required. This instrument rates 10 withdrawal features according to clinician's estimates of frequency or severity. Administration of the CIWA-Ar requires approximately 2 minutes and is often performed by nursing staff. The findings of recent studies (Sullivan et al. 1991; Wartenberg et al. 1990) point to the utility of the CIWA-Ar for identifying patients who require benzodiazepines and for determining suitable doses. In addition, the findings of Johnson et al. (1991) suggest that the CIWA-Ar may clarify diagnostic factors: The investigators found that the CIWA-Ar detected more severe withdrawal in patients with both alcoholism and co-occurring anxiety disorders than in patients with alcoholism only.

Another aid to treatment planning, the Alcohol Use Inventory (Horn et al. 1987; Wanburg et al. 1977), is a 40- to 60-minute self-administered test used to assess perceived benefits of drinking, drinking-related consequences, and drinking style. More recently, the Outcome Expectancy Scale (OES) was developed to assess a patient's expectations of the advantages and disadvantages of modifying drinking behavior (Solomon and Annis 1989). Although OES scores have not been found to predict posttreatment alcohol relapse (Solomon and Annis 1990), the scale may be useful both for treatment planning and for modifying beliefs antagonistic to the recovery process.

The Alcohol Expectancy Questionnaire (AEQ) (Brown et al. 1980) and the Inventory of Drinking Situations (IDS) (Annis et al. 1987) have been used to identify emotional, cognitive, and social factors that may precipitate drinking, thus enabling clinicians to better target relapse prevention efforts. Using the AEQ, Brown (1980) also found that patients with fewer expectancies of drinking-associated benefits (e.g., tension reduction and relaxation, social and physical pleasure) had greater overall treatment success than patients with greater expectancies; in particular, fewer expectancies of relaxation effects were associated with greater likelihood of abstinence or nonproblematic drinking.

Several new instruments have been developed to assess more thoroughly factors believed to be associated with relapse. The Situational Confidence Questionnaire (SCQ) (Annis and Graham 1988; Solomon and Annis 1990), which approximates the IDS in content, is used to assess patients' confidence in their ability to avoid drinking in certain high-risk situations. SCQ scores have been shown to discriminate between long-term and short-term abstinent alcoholics (Miller et al. 1989), to predict heavy drinking episodes (Annis and Davis 1988), and to predict length of treatment retention (Burling et al. 1989). In addition, Burling et al. found that changes in SCQ scores during treatment may predict posttreatment drinking status, and Sandahl et al. (1990) found that combined scores on four of the SQC scales can predict treatment outcome at 6 months posttreatment.

Another useful instrument in determining treatment intensity is the Clinical Institute Alcohol Withdrawal Scale (revised), used to assess whether a patient requires medication to manage the alcohol withdrawal syndrome and, if so, the dosage required.

Validity of Patient Self-Report

Traditionally, patient self-report of alcohol use has been the most commonly used measure of pretreatment alcohol use and posttreatment abstinence. However, several research teams (Fuller et al. 1988; Orrego et al. 1979; Peachey and Kapur 1986; see also Midanik 1988) have demonstrated that patient self-reports are often an inadequate measure of alcohol consumption. From a review of estimated alcohol use based on selfreports, collateral reports, patient diaries, and other methods, Midanik (1988) concluded that self-report validity can be influenced by patient and interviewer characteristics, type of data assessed, and context of the assessment. Nevertheless, self reports are an important source of data (Maisto et al. 1990; Sobell and Sobell 1990) (see chapter 13).

Fuller et al. (1988) compared patient and collateral reports and laboratory measures of alcohol use throughout a clinical trial of disulfiram administration (Fuller et al. 1986). When collateral reports were defined as the basis of