Addiction Severity Index (ASI)

The ASI is a multidimensional clinical interview instrument that measures drug and alcohol abuse. It may be used shortly after admission to treatment as well as during follow-up visits. The developers suggest using the ASI to match patients with appropriate treatments and to "promote greater comparability of research findings." (Ref: 1) Introduced at the 1975 National Institute of Drug Abuse Conference on Treatment Efficacy, the ASI is based on the theory that addiction should be addressed by concentrating on the treatment problem issues which contribute to or result from addiction. The seven treatment issues addressed by the ASI are: medical condition, employment, alcohol use, drug use, illegal involvement, family/social relations, and psychiatric status. The instrument consists of a total of 161 items.

The ASI produces a problem severity profile for a patient across the seven problem areas. "Severity" in the ASI is defined as "need for additional treatment." (Ref: 1) The administrator assigns a severity score (on a scale of 0-9) for each problem area based on objective data (patient responses to questions about number, intensity and duration of problem symptoms during lifetime and over the past 30 days, data from physical exams, lab reports, psychological test results, etc.) and patient self-assessments (patient uses a 5-point scale to rate how much he/she has been bothered by each problem area over the past 30 days, and the extent to which he/she feels treatment would be beneficial). There is no assessment of quantity of drug or alcohol use. (Ref: 4)

The authors caution that the ASI is best administered in person-to-person interview format by a trained medical interviewer. When administered in phone-interview or written format, reliability drops to 0.74 or 0.55, respectively. The ASI is inappropriate for use with three groups of substance abusers: cognitively impaired older patients, younger patients with criminal histories, and adolescents younger than 16 years old. (Ref: 2,4)

Free, through the senior author, Dr. A.T. McLellan.

Average is 50-60 minutes; range is 20-70 minutes. (Ref: 4)

1980

Some of the items are technical, but appropriate for clinicians (the target administrators).

1) Objective information

• open-ended
• objective data questions

• 5-point scale

• 10-point scale

A trained medical interviewer should administer the ASI. (Ref: 1,4) A computerized version has been established. (Ref: 9)

Information is computer-coded. Drawing upon the objective patient medical data, the interviewer develops a 4-point preliminary range of problem severity for each problem area. Then, considering the patient’s self-assessment responses, the interviewer hones his/her preliminary severity rating range to a point estimate within a 2-point bracket. Severity ratings (point estimates) range from zero (0) to nine (9) on an unanchored scale. A composite score for the patient for each problem area is computed by summing individual item scores across sets of interrelated items corresponding to a problem area; items are standardized and summed to produce an estimate of patient status in each area. (Ref: 2) Composite scores vary from 0 to 1, with higher values indicating greater severity of problems. (Ref: 7) Further detail on scoring and interpretation is not provided in the articles, but is available on request from the primary author. (Ref: 2)

Multiple languages including French, Spanish, German, Dutch and Russian. (Ref: 4)

A. Thomas McLellan, Lester Luborsky, George E. Woody, Charles P. O’Brien

Drug Dependence Treatment Service, Veterans Administration Hospital, University and Woodland Avenues, Philadelphia, PA 19104.

Developer’s current address: Building 7, PVAMC, University Avenue, Philadelphia, PA 19104; (214) 399-0890

tmclellan@tresearch.org

No information found.

top

No information found.

The authors of the ASI recognized the need to validate the ASI in various populations, but also acknowledged the large variability within each patient population. Therefore, rather than creating a single set of norms, they opted to present mean values of composite scores and severity ratings for several different substance abuse patient groups with individual and treatment variations. (Ref: 4) Three groups of data are provided:

1. Alcohol, opiate and cocaine abusers: drawn from Philadelphia public and private, inpatient, partial or outpatient treatment programs (55% from Philadelphia VA, all female subjects from non-VA programs), greater than three years of problematic use.
2. Alcohol, cocaine and alcohol/cocaine abusers: drawn from the VA and Philadelphia city public treatment programs, Carrier Foundation and Integra Employee Assistance private treatment programs.
3. Other abusers: Pregnant substance abusers (women in the Emory University Treatment Program for Mothers in Atlanta, GA.), “out of treatment” opiate abusers, incarcerated abusers drawn from admission information of the Hillsborough County Sheriff’s office S.A. Treatment Program for drug-dependent inmates, psychiatrically ill substance abusers admitted to outpatient treatment programs, “employer coerced” patients forced into treatment due to random drug testing, and homeless male substance abusers drawn from a community NIAAA stabilization project.

Based on a three-day test-retest time interval, the correlation of composite scores is reported as 0.92. (Ref: 2) With a two-day interval, agreement rates of 76-100% and kappa coefficients of ³0.83 were achieved. (Ref: 4) For each individual problem area, Pearson correlations of composite scores (across three- to five-day intervals) are reported as ranging from 0.14 to 0.95, with a mean of 0.55, and Spearman correlations of 0.25 to 0.97, with a mean of 0.65). The subscales representing the Medical and Legal problem areas produced correlations insufficient to support adequate reliability. (Ref: 7) For a computerized version of the ASI, score correlations range from 0.68 to 0.95 for composite scores and 0.62 to 0.84 for severity ratings (test-retest interval not reported). (Ref: 9)

Inter-rater coefficients for judgment of severity ratings ranged from an average of 0.89 to 0.92 in developmental testing. (Ref: 1) In another study, ASI interviews given by an independent interviewer were videotaped and then watched by seven other judges, who then made their own assessments. Ratings correlated on average from 0.74 for the Employment problem area to 0.91 for the Drug problem area (Pearson), and from 0.94 for the Family problem area to 0.99 for the Drug problem area (Spearman). Other evidence shows that, 10 percent of the time, eight judges will agree within two points of each other on the 10-point rating scale. (Ref: 2) Correlations of composite scores (CS) and interviewer severity ratings (ISR), as measured by Intraclass Correlation Coefficient (ICC), support good inter-rater reliability for the Medical (0.80), Alcohol (0.96) and Other Drugs (0.86) sections of the ASI. Other results follow: (Ref: 3)

Scale intercorrelations for ASI Severity Ratings range from 0.06 to 0.27, with the exception of the correlation between the Family/Social scale and the Psychological scale (0.41). (Ref: 1) Internal consistency of composite scores, as measured by Cronbach’s alpha, ranged from 0.48 (Legal) to 0.88 (Medical) with a mean alpha of 0.68. (Ref: 3) Another study reported higher alphas for composite scores, with a mean alpha of 0.80 and a range of 0.70 (Employment) to 0.89 (Medical). (Ref: 6) In administration of the ASI under ‘nonideal’ conditions (i.e., procedures may not have been followed according to the ASI manual and clinicians were minimally trained), Cronbach’s alpha for composite scores was still good, ranging from 0.65 to 0.89. (Ref: 8)

No information found.

Face validity was demonstrated with a sample of 524 male veteran substance abusers. Scale scores were correlated with independent factors that were related to each scale’s problem area. (Ref: 1) Results for scale and independent measure correlations are as follows: Abuse scale r=0.54-0.72, Medical scale r=0.58-0.69, Employment/support scale r= -0.64-0.56, Family/social scale r=0.43-0.52, Legal r=0.62-0.71, Psychological scale r=0.58-0.64.

The ASI severity ratings correlate with other instruments as follows: CAGE (0.45), SMAST (0.52), CUAD alcohol score (0.73), and drug urinalysis screening (0.50). The ASI composite scores correlate with other instruments as follows: CAGE (0.50), SMAST (0.59), and CUAD (0.72). Composite drug scores were related to the CAGE-AID (0.64), DAST (0.73) and CUAD (0.70). (Ref: 6)

As a measure of concurrent validity, patients were divided into severity groups (Low, Mid, High) based on their severity scores in the ASI problem areas and then compared on items which were indicators of each problem area (items from the lifetime and past 30 days portions of each problem area). There were significant (p<=0.01) between group differences on all comparisons except one (number of times treated for alcohol use). (Ref: 2)

Researchers assessed criterion validity by comparing the ASI drug and alcohol scales to the SCID-P. Results showed that a rating of ³1.0 had a sensitivity of 84% in people with lifetime alcohol problems and 93% in recent abuse. On the drug scale, a score of 1.0 had a sensitivity of 91% and a specificity of 92%. Sensitivity was 93% for diagnosis of current drug abuse, but only 55% of nonabusers were detected. (Ref: 6)

Another study assessed concurrent validity by comparing ASI self-report results with urine drug screens. Four patients tested positive for cocaine, but only two of them reported it. Eleven patients tested positive for marijuana, but only 5 reported it. (Ref: 7)

Criterion validity of a computerized version of the ASI was found by comparing it to the original ASI. Intraclass correlations ranged from 0.54 to 0.95 for composite scores and –0.12 to 0.64 for severity ratings. (Ref: 9)

No information found.

No information found.

Yes. (Ref: 1-2,4)

Yes. (Ref: 3,5-9)