off score from one to four positive responses, and differences in population samples. The major drawback of this test is that it does not assess current problems, levels of consumption, or episodes of heavy drinking.

T-ACE and TWEAK

Two instruments specifically developed for pregnant women were tested with 4,000 patients attending a prenatal clinic in Detroit (Russell

et al. 1991). The T-ACE (Sokol et al. 1989) is a modification of the CAGE in which a question on TOLERANCE (i.e., how many drinks it takes to make you high?, with an alternative tolerance question asking how many drinks the woman can hold) is substituted for the item on GUILT. The TWEAK (Russell et al. 1991) is a similar modification that assesses (T) TOLERANCE; (W) WORRIED (Have close friends or relatives worried or complained about your drinking?);

(E) EYE opener (Do you sometimes take a drink in the morning to wake up?); (A) AMNESIA (Has a friend or family member ever told you things you said or did while you were drinking that you could not remember?); and (K) CUT (Do you sometimes feel the need to cut down on your drinking?). TWEAK items are weighted, and a score of 3 is considered to be a positive test. In a sample of nonabstaining women attending an inner-city prenatal clinic, the TWEAK exceeded the T-ACE in predictive value relative to riskdrinking during pregnancy when tolerance was defined as the number of drinks one could hold (Russell et al. 1991).

MAST

The MAST and its variations continue to be one of the more frequently used tests. The MAST assesses lifetime dependence symptoms, alcoholrelated problems, medical consequences, and previous treatment. It has no questions on consumption and does not differentiate past alcohol problems from more recent ones. The full MAST appears in the Seventh Special Report (U.S. Department of Health and Human Services 1990). The original 24-question MAST (Selzer 1971) (see also Hedlund and Vieweg 1984; Ross et al. 1990) has been modified by a number of investigators. Modifications include the brief 10question B-MAST (Pokorny et al. 1972), the 13-question shortened SMAST (Selzer et al. 1975) (see also Cleary et al. 1988; Fleming and Barry 1989b, 1991; Harburg et al. 1988), and the 35question Self-Administered Alcohol Screening Test (SAAST) (Swenson and Morse 1975) (see also Davis and Morse 1987; Hurt et al. 1980; Loethen and Khavari 1990).

AUDIT

The Alcohol Use Disorder Identification Test (AUDIT), a brief two-part multicultural screening tool for the early identification of problem drinking (Babor and Grant 1989), was developed by a working group of the World Health Organization (WHO). This group selected questions that identified high-risk drinkers in a six-nation study (Saunders, Aasland, Amundsen et al. in press; Saunders, Aasland, Babor et al. in press). The core of the AUDIT is shown in Figure 2. It contains a series of 10 questions that include 3 items on alcohol use, 4 on dependence, and 3 about problems. The 10 questions can be administered in an interview or as a paper-and-pencil questionnaire. Several studies on the validity of this

measure have been conducted (Barry and Fleming 1992; Fleming et al. 1991). Using a cutoff score of 8, the AUDIT displayed an overall sensitivity of 92 percent, and an overall specificity of 93 percent (Babor and Grant 1989). The second part of the AUDIT records alcoholrelated physical measures and laboratory findings.

Lifestyle questionnaires

The Health Screening Questionnaire (HSQ) (Cutler et al. 1988; Wallace and Haines 1985) and its modification, the Health Screening Survey (HSS) (Fleming and Barry 1991a) were developed as lifestyle questionnaires. These screening tests contain parallel questions on smoking, exercise, weight, and alcohol use. The alcohol portion of the HSQ and the HSS contain questions on the use of three types of alcoholic beverages in the previous 3 months, the four CAGE questions, and two questions on problem drinking. Although these types of questionnaires do not appear to increase the validity of the responses (Plant and Miller 1977), such tests give additional health-related information and may be more acceptable to clinical staff and patients since the focus is not limited to alcohol. These tests were developed for two large brief-intervention studies in England (Wallace and Haines 1985) and in the United States (Fleming and Barry 1991a). The Life-Style Risk Assessment questionnaires developed by the Addiction Research Foundation (Skinner et al. 1985a, 1985b) and the Life Style Test developed by Graham (1991) are two other examples of screening questionnaires that combine alcohol questions with items on other health issues.

Tests for Screening Adolescents

A number of new instruments have been developed for screening adolescents (Klitzner et al. 1987). The POSIT (Problem-Oriented Screening Instrument for Teenagers) is a promising new instrument developed by Tarter and an expert panel sponsored by the National Institute on Drug Abuse (1990). Developed as part of the Adolescent Assessment Referral System, the POSIT discriminates between adolescents in treatment and those drawn from a school population (Klitzner and Rahdert 1991). This questionnaire assesses 10 problem areas, including a 14-item subscale measuring alcohol and other drug use and abuse. One or more positive responses is considered a positive result. A number of studies

are assessing the validity of the POSIT in general population samples. Other instruments specifically developed for adolescents include the Personal Experience Inventory (Winters 1990a, 1990b), Drinking and You (Harrel and Wirtz 1988), and Perceived Benefit of Drinking and Drug Use Scales (Petchers et al. 1988). Initial validity testing of these instruments is encouraging.

There are limited data on the effectiveness of screening instruments with women, adolescents, the elderly, the mentally ill, and racial and ethnic minorities. The tests discussed in this chapter are screening tools that were not developed as comprehensive assessment instruments. Since screening tests are designed to be as sensitive as possible, false-positive tests are a concern, especially in the general population. To minimize potentially adverse effects, such as denial of insurance, positive screening tests should be followed up with a diagnostic assessment.

Methods of Administering
Screening Tests

Methods of administering screening tests include (1) a face-to-face interview by a nonclinical interviewer, (2) a face-to-face interview by a clinician, (3) a paper-and-pencil questionnaire, and (4) a self-guided computerized interview. A number of studies have established the comparability of these methods. Robins (1985) found a high degree of agreement between the interviews conducted by nonclinical researchers and expert findings using the Diagnostic Interview Schedule (DIS). Barry and Fleming (1990) compared paper-and-pencil questionnaires with computerized administration of the CAGE and SMAST in a primary care sample and found no significant differences in sensitivity and specificity between the two methods.

Using the CAGE, SMAST, and an alcohol consumption test, Bernadt and colleagues (1989) measured the degree of agreement among three administration methods (self-report via computer, face-to-face interview with a nurse, and face-to-face interview with a psychiatrist). The kappa values varied from .79 to .86, with results from the CAGE and S-MAST having the highest levels of agreement. (The kappa statistic measures the degree to which agreement, beyond chance agreement, occurs; the higher the number, the greater the agreement [Sackett 1992].) The findings are consistent with the work of Skinner and colleagues (1985a) and Erdman et al. (1985) but not those of Duffy and Waterton

(1984), who found higher consumption levels by computer report than by face-to-face interview. Griest et al. (1987) compared computer and faceto-face administration of the DIS to a group of 150 clients in an outpatient psychiatric clinic and found a high degree of agreement for the alcohol and drug subscales (kappa = .83).

In summary, the four methods appear to achieve comparable results. Although the validity of the responses obtained by these four methods is similar, paper-and-pencil and computerized methods are the least expensive. Patients also seem to prefer the computerized methods because of their novelty and ability to give immediate feedback (Skinner et al. 1985a).

Methods of administering screening tests include (1) a face-to-face interview by a nonclinical interviewer, (2) a face-to-face interview by a clinician, (3) a paper-and-pencil questionnaire, and (4) a self-guided computerized interview.

Screening Based on Laboratory
Tests

Three types of laboratory screening tests are currently available. The first includes measurement of alcohol levels in the breath, urine, and blood. Alcohol levels have been successfully used for screening by law enforcement agencies (Gijsbers et al. 1991) and emergency departments (Chang and Astrachan 1988; Cherpitel 1989), and in corroborating self-reports (Fuller et al. 1986). Hand-held breathalyzers and urine dipstick testing are technologies that have not been widely employed by health care professionals to assess alcohol levels. Both of these techniques are inexpensive, and the results correlate well with blood alcohol levels. However, alcohol levels measure only current alcohol use and are not necessarily useful in identifying an ongoing alcohol problem.

The second type of laboratory screening measures cellular injury to the liver and to the cells that manufacture red blood cells. The plasma

glutamyl transferase (GGT) and mean corpuscular volume (MCV) are the primary measures used to screen for alcohol problems. The sensitivity of these tests varies from 20 to 60 percent, depending on the chronicity and severity of alcohol use. Specificity is higher, but false-positives