The fifth edition of the addiction severity index

Journal of Substance Abuse Treatment, Printed in the USA. All rights reserved.

Vol. 9, pp. 199-213,

1992 Copyright

0

0740-5472192 $5.00 + .OO 1992 Pergamon Press Ltd.

ARTICLE

The Fifth Edition of the Addiction

Severity Index

A. THOMAS MCLELLAN, PhD,* HARVEY KUSHNER, PhD,* DAVID METZGER, PhD,* ROGER PETERS, PhD,t IRIS SMITH, PhD,$ GRANT GRISSOM, PhD,$ HELEN PETTINATI, PhD,II AND MILTON ARGERIOU, PhDf *Penn-VA Center for Studies of Addiction, Philadelphia; iUniversity of South Florida, Department of Law and Mental Health, Tampa, Florida; SGeorgia Mental Health Atlanta, Georgia; §INTEGRA Inc., Radnor, Pennsylvania; )ICarrier Foundation, TStabilization Services Project, Boston, Massachusetts

Florida Mental Health Institute, Center, Pediatrics Division, Belle Meade, New Jersey;

Abstract- The Addiction Severity Index (ASZ) is 12 years old and has been revised to include a new section on family history of alcohol, drug, andpsychiatricproblems. New items were added in existing sections to assess route of drug administration; additional illegal activities; emotional, physical, and sexual abuse; quality of the recovery environment; and history of close personal relationships. No changes were made in the composite scoring to maintain comparability with previous editions. This article discusses the clinical and research uses of the ASZ over the past 12 years, emphasizing some special circumstances that affect its administration. The article then describes the rationale for and description of the changes made in the ASZ. The final section provides “normative data” on the composite scores and severity ratings for samples of opiate, alcohol, and cocaine abusers as well as drug abusing inmates, pregnant women, homeless men, and psychiatrically ilI substance abusers. Keywords-alcohol abuse;

drug abuse; evaluation;

INTRODUCTION

treatment.

budget for the project was small, the data had to be collected by technicians rather than health care professionals. Since the data had to be collected as part of the clinical process in a relatively short period of time, the instrument had to focus on a minimum number of questions, and they had to be relevant to the treatment plan that was to be provided. Finally, since a major purpose of the project was to measure outcome (an issue that still remains open to various definitions), the questions had to cover a broad range of potential areas that could be affected by substance abuse treatment, and the format of these questions had to be suitable for repeat administration at follow-up contacts. Perhaps this set of initial requirements more than any other rationale explains the resulting orientation and construction of the ASI. Since 1980 the AS1 has been widely used. The instrument has been translated into nine languages, including French, Spanish, German, Dutch and Russian. It has been found to be reliable and valid in several different contexts (McLellan, Luborsky, Cacciola, & Griffith, 1985; Kosten, Rounsaville, & Kleber, 1985; Rogalski, 1987; Hendricks, Kaplan, VanLimbeek, &

IN 1980, A NEW INTERVIEW for substance-dependent patients was introduced, called the Addiction Severity Index or AS1 (McLellan, Luborsky, O’Brien, &Woody, 1980). The ASI was created for the special purpose of enabling a group of clinical researchers to evaluate treatment outcome for “alcoholics” and “drug addicts” in a six-program, substance abuse treatment network (see McLellan, Woody, Luborsky, O’Brien, & Druley, 1982; McLellan, Luborsky, Woody, Druley, & O’Brien, 1983). Since the programs were quite different, the AS1 had to be reasonably generic. Since the

Supported by grants from the Department of Veterans Affairs and the National Institutes of Drug Abuse, Alcohol Abuse, and Alcoholism. We are grateful for the efforts of John Cacciola, Ian and Barbara Fureman, Gargi Parikh, Alicia Bragg, David Zanis, Jeff Griffith, Ray Incmikoski, and Louise Bridges in the development, refinement, and training of ASI interviewing procedures. Reprint requests should be addressed to Penn-VA Center for Studies of Addiction, Building 7, Philadelphia VA Medical Center, University Avenue, Philadelphia, PA 19104.

I99

A. T. McLellan et al.

200

Geerlings, 1988), and it has been a significant part of treatment outcome studies for opiate (Ball & Corty, 1988), cocaine (Gawin et al., 1989), and alcohol (Kadden, Cooney, Getter, & Litt, 1990) dependence. The AS1 has also been used to evaluate large scale studies of other related populations such as drug abusing prisoners (Wexler, Falkin, & Lipton, 1988), psychiatrically ill substance abusers (Lehman, Myers, & Corty, 1989), homeless persons with and without substance abuse problems (Lubran, 1990), and more recently, pregnant addicts and addicted mothers (Smith, Moss-Wells, Moeti, 8c Coles, 1990). However, there have been many changes in the field of drug and alcohol dependence treatment since the AS1 was originally developed in 1980. The primary drug problems at that time were heroin, LSD, and amphetamine (Kozel & Adams, 1986; McLellan, McGahan, & Druley, 1979). Cocaine was rarely mentioned by prospective patients in 1980, and the smokable, freebase form (now called “crack”) was essentially nonexistent. Regular use of single substances (particularly alcohol) was more common, and multiple substance use was less pervasive. At the beginning of the decade, alcohol- and “drug’‘-abusing patient samples were clearly differentiable with respect to both demographic and drug pattern variables, and these groups were generally treated in separate types of rehabilitation programs. There have also been major changes in our knowledge about substance abuse over the past decade resulting from research studies. For example, genetic and family therapy studies indicate that alcohol dependence and perhaps some forms of drug dependence are passed across generations within families (Hesselbrock, Stabenau, Hesselbrock, Meyer, & Babor, 1982; Rounsaville et al., 1991), although it is not yet clear whether this intergenerational transmission is primarily genetic, primarily learned, or a combination of these processes. Route of administration (particularly among cocaine users) has been shown to be an important consideration in determining the severity of drug dependence (Gawin & Kleber, 1986). Researchers have also shown that thei-family, criminal, employment, and psychological problems seen among alcohol and drug abuse patients are important predictors of response to treatment, with psychiatrically ill and especially antisocial substance abusers being particularly likely to show poor during-treatment response and early relapse, regardless of the treatment modality or setting (Rounsaville, Dolinsky, & Babor, 1987; Woody, McLellan, Luborsky, & O’Brien, 1985). Despite these significant changes in the field of alcohol and drug abuse over the past years, the AS1 has not been updated to reflect these changes, and the widespread use of the Addiction Severity Index over this period has revealed several shortcomings within the instrument. In an attempt to address these concerns and to improve the general utility and contemporary value of the instrument, we have updated the ASI and added

a significantly more detailed and comprehensive user’s manual. The present paper describes this fifth edition of the ASI and the rationale for the changes that have been made. The first section of the paper discusses the most common general problems with the AS1 and offers cautions and suggestions based on our experience with the instrument over the past 12 years. The second section of this paper introduces and discusses the rationale for the new items that have been added to the fifth edition of the ASI. The third section of this article discusses the use of “norms” in the substance abuse population and provides mean values on the AS1 severity ratings and composite scores for a variety of representative patient samples. SECTION ONE-A DISCUSSION OF COMMON QUESTIONS AND PROBLEMS WITH THE AS1 The Interview Format

Does It Have to be an Interview? In the search for faster and easier methods of collecting data, many clinicians and researchers have asked for a self-administered (either by computer or paper and pencil) version of the instrument. We have not endorsed the use of a self-administered version for several reasons. First, we have tested the reliability and validity of the severity ratings by having raters use just the information that has been collected on the form-without the interview. This has resulted in very poor estimates of problem severity and essentially no concurrent reliability. Second, we have been sensitive to problems of illiteracy among segments of the substance-abusing population. Even among the literate there are problems of attention, interest, and comprehension that are especially common in this population. Finally, since the instrument is often used as part of an initial clinical evaluation, it has been our philosophy that it is important to have interpersonal contact for at least one part of that initial evaluation. We see this as simply being polite and supportive to a patient with problems. We have found that particularly among some segments of the substanceabusing population (e.g., the psychiatrically ill, elderly, confused, and physically sick) the interview format may be the only viable method for insuring understanding of the questions asked. Thus, it should be clear that at this writing there is no reliable or valid version of the ASZ that is self-administered and there is currently no plan for developing this format for the instrument. We would of course be persuaded by comparative data from a self-administered version of the ASI, and this is an open invitation to interested parties. Role of the Interviewer

What Are the Qualifications Needed for an ASI Interviewer? Having indicated the importance of the interview process it follows that the most important part of

Revised ASI the AS1 is the interviewer who collects the information. The interviewer is not simply the recorder of patient responses. The interviewer is responsible for the integrity of the information collected and must be willing to repeat, paraphrase and probe until the patient understands the question and the answer is consistent with the intent of the question. Despite the range of situations and unusual answers that we have described in the AS1 manual, a new exception or previously unheard of situation occurs virtually each week. Thus, it is critical for prospective interviewers to understand the intent of each question, to probe for the most complete information available from the patient, and then to record the most appropriate answer, including a comment. There are no clear-cut educational or background characteristics that have been reliably associated with the ability to perform a proficient AS1 interview. We have trained a wide range of people to administer the ASI, including receptionists, college students, police/probation officers, physicians, professional interviewers, and even a research psychologist! There have been people from each of these groups who were simply unsuited to performing interviews (perhaps 10% of all those trained) and were excluded during the 2-day training period or on subsequent reliability checks. Reasons for exclusion were usually the inability to form reasonable rapport with the patients, insensitivity to lack of understanding in the patient, or failure to effectively probe initially confused answers with supplemental clarifying questions. Quantity of Use

Why Doesn’t the ASI Assess Quantity of Use? There is no assessment of the quantity of drug or alcohol use in the AS1 for two reasons. First, in over 10 years of examining the relationships among quantity and frequency of alcohol and drug use and treatment response we have found that frequency is very well correlated with quantity and that they provide essentially redundant information regarding the severity of a patient’s problem. A second and more important reason for not reporting quantity of drug or alcohol use is the inherent problem of getting accurate information. In the case of alcohol it is possible to record quantity estimates in terms of absolute levels of alcohol, using drink conversion tables (Cahalan, 1987). We have found that while patients can usually recall the frequency of their drinking episodes, they have much more difficulty in recalling the quantity during each episode. Recording patient estimates of the quantity of their drug use has been virtually impossible. First, the street measures of drug use differ widely (e.g., spoons or bags of heroin, lines of cocaine, vials of crack, hits of speed, etc.). Second, even if it were possible to record quantity accurately, it has often been the case that the drug that was purchased for ingestion (e.g., cocaine) actually turned

201 out to be a different drug (e.g., PCP, methamphetamine, or a mixture). Again, we have found that frequency of use has been the best estimate of overall severity of use. Users who wish additional, detailed information on quantity of drug use or the specific conditions under which drug and alcohol use usually take place are urged to probe further with specific questions. 30-Day Period

Why Have We Used a 30-Day Reporting Periodfor Recent Status? Many have questioned our decision to record the frequency of “problem days” during the onemonth period prior to the evaluation point. There are problems with this (or any other interval) and these bear noting. For example, it can be reasonably argued that the 30-day period prior to admission to a drug or alcohol treatment program may offer a poor representation of the patient’s “true severity” in each problem area. First, this period may over-represent the usual severity of these problems and reflect the most desperate point in the patient’s life. Second, since many patients enter treatment at a point following incarceration, hospitalization or stabilization in a controlled environment, the prior 30-day period may under-represent the “true severity” of problems. While both of these occurrences are common, we have settled on this sampling frame based on the following rationale. It is valuable for both clinical and research reasons to develop a continuous measure of severity in each problem area. In turn, the best candidate for this continuous measure of severity has been the frequency of problem days over a fixed period. This has been historically the easiest and most reliable measure of overall problem severity in each of the seven areas evaluated by the ASI. It could be argued that a longer period of sampling, for example 2 months, would provide an equally continuous and possibly more representative measure of problem status. Unfortunately results of several reliability studies that have specifically focussed on this question suggest that recall accuracy drops drastically for periods beyond 30 days (McLellan et al., 1980; 1985). This longer period would also not eliminate the influence of particularly poor function or of confinement in a controlled environment. From a clinical perspective we have decided that the problem status presented by the past 30-day period does offer an accurate representation of the patient status at the point of admission, regardless of whether this point represents the “typical” pattern of problems shown by the patient. It must be remembered that the admission or baseline AS1 also inquires about the pattern of lifetime symptoms in each problem area. It has been our experience that the lifetime and past 30-day periods may be combined clinically to form the basis of a more representative evaluation. At the follow-up point, the 30-day window has been criticized as not providing an adequate picture of a

A. T McLellan et al.

202

patient’s outcome status over the 3, 6 or 12 months (depending upon the follow-up interval) since leaving treatment. In response to this, many workers have used time-line follow-back procedures to determine the nature and severity of problems throughout the period following treatment discharge. We applaud these efforts and have also used them in a number of research studies. We have found that while some level of detail is sacrificed in the use of these procedures, major events (e.g., use of drugs, employment, crime, hospitalization) may be recalled accurately using these follow-back procedures. Our rationale for the use of the 30-day period at the follow-up point is based on the importance of measuring both outcome status (i.e., the condition of the patient at the time of assessment) and improvement (i.e., change from pretreatment to posttreatment). Despite the problem of the loss of some specificity of recall with the time-line follow-back procedure, it generally does provide a more representative indication of a patient’s outcome status following treatment than the prior 30-day period. However, this measurement procedure does not permit a comparison of a patient’s status pretreatment and posttreatment (due to the unequal time intervals assessed) and therefore cannot assess the degree of improvement shown by the patient. In our use of the AS1 at follow-up, we have relied on random sampling procedures and high follow-up contact rates (typically 88% or higher) to provide a representative (but admittedly not complete) indication of group outcome status at follow-up, and we have regularly attempted to characterize patient status at more than one point following treatment (typically 3 months, 6 months, and 12 months). In addition, the comparison of the two 30-day periods, prior to treatment and follow-up, permit the assessment of patient improvement over that time period. We do not claim that this is the only or even the best method of evaluating treatment efficacy, only that it is one reliable, valid, and direct method. Workers who are primarily interested in the most complete representation of outcome status following treatment would be advised to use a time-line follow-back procedure instead of focusing just on the 30-day period prior to the follow-up point. However, as indicated previously, it is possible for users of the AS1 to do both a standard follow-up and a time-line follow-back evaluation. Severity Ratings

How Important and Useful Are They? It is noteworthy that the severity ratings were historically the last items to be included on the ASI. They were considered to be interesting but nonessential items that were a summary convenience for clinicians who wished a quick general profile of a patient’s problem status. It was surprising and interesting that when interviewers were appropriately trained, these severity estimates were

reliable and valid across a range of patient and interviewer types (McLellan et al., 1985). The severity ratings are still viable as a clinical summary and we continue to recommend their use- but only for initial treatment planning and referral. This means that many potential users of the AS1 would not benefit from using the severity ratings. For example, researchers may be interested exclusively in posttreatment outcome evaluation, and severity ratings have never been used as outcome measures. Other users may not have the time or inclination to check and recheck severity estimates among their various interviewers. For all of these potential users, the severity ratings would not be useful or worth the investment of staff-hours required to train reliability. Therefore, it is entirely acceptable to use the AS1 without the severity ratings. Appropriate

Populations

Can I Use the ASI With Samples of Substance Abusing Prisoners, Psychiatrically Ill Substance Abusers, or Homeless Substance Abusers? Because the AS1 has been shown to be reliable and valid among substance abusers applying for treatment, many workers in related fields have used the instrument with other types of substance abusers. For example, the AS1 has been used at the tiime of sentencing, incarceration, and/or parole/probation to evaluate substance abuse and other problems in criminal populations. In addition, because of the widespread substance abuse in mentally ill and homeless populations, the AS1 has also been used among these groups. While we have collaborated with many workers on the use of the instrument with these populations (see Section Three of this paper), it should be clear that, to our knowledge, there are nopublished reliability or validity studies of the instrument in these populations. This does not mean that the AS1 is necessarily invalid with these groups, only that its testparameters have not been established. In cases where no other suitable instrument is available, the AS1 could be a better choice than the creation of a totally new instrument. However, it is important to note circumstances that are likely to reduce the value of data from the AS1 among groups such as these. For example, when used by a trained interviewer, with a treatment-seeking sample and in a private interview setting there is little reason for a substance abuser to misrepresent (even under these circumstances it still happens). In circumstances where individuals are being “evaluated for probation/parole or jail” there is obviously much more likelihood of misrepresentation. Similarly, when the AS1 is used with psychiatrically ill substance abusers who are not necessarily seeking (and possibly avoiding) treatment, there is often reason to suspect denial, confusion, and misrepresentation. Again, there is currently no suitable alternative research instrument or clinical procedure available that will insure valid, accurate responses un-

Revised ASI

203

der these conditions. While the consistency checks built into the AS1 may be of some benefit in these circumstances these are not substitutes for systematic tests of the reliability and validity of the AS1 in populations of substance abusers within the criminal justice system and within the mental health system. Again, this is an open invitation to interested parties.

A Special Note on Adolescent

Populations

Despite the fact that we have repeatedly published warnings for potential users of the ASI regarding the lack of reliability, validity, and utility of the instrument with adolescent populations, there remain instances where the ASI has been used in this inappropriate manner. Again, the ASI is not appropriate for adolescents due to its underlying assumptions regarding self-sufficiency and because it simply does not address issues family problems from (e.g., school, peer relations, the perspective of the adolescent, etc.) that are critical to an evaluation of adolescent problems. At this writing, there are four versions of the AS1 that have been developed for adolescent populations and have shown evidence of reliability and validity in this population. Interested readers may contact the following individuals directly for more information about these instruments: David Metzger, Penn-VA Center for Studies of Addiction, Philadelphia, PA 19104 “Adolescent Problem Severity Index” (APSI)

plement what was, by design, the minimum number of questions possible to evaluate patient status in each problem area. This suggestion remains in the latest (fifth edition) version of the instrument. At the same time, this revision reflects an effort to add information in areas that were inadequately covered in earlier versions of the instrument. No changes or additional items have been added in the Medical, Employment, or Psychiatric sections of the ASI. These sections continue to offer valid basic information that may be easily supplemented with particular questions or additional instruments to provide more detailed evaluations of patient status in these areas. Changes have been made in the alcohol and drug use and particularly in the family sections, reflecting our view that these areas were poorly covered in prior editions and that advances in our knowledge in these areas warrants revision. The new items are discussed generally below by problem area, but the specific instructions for asking these questions and for interpreting the answers are discussed in the Fifth Edition Users Manual (availably from the senior author). Please note, there are many scales, questionnaires, and other assessment instruments that have been developed over the past 10 years that may be used instead of or in addition to the AS1 for a more complete and specific assessment. The interested reader is encouraged to consult the tests and measurements literature for more information.

ALCOHOL

or Kathy Meyers, Carrier Foundation, Belle Meade, NJ 08502 “Comprehensive Addiction Severity Index for Adolescents” (CASI-A) OI

Yifrah Kaminer, Bradley Hospital, East Providence, RI 02915 “Teen Addiction Severity Index” (Teen ASI) or All Friedman, Philadelphia Psychiatric Center, Philadelphia, PA 19108 “Adolescent Drug and Alcohol Diagnostic Assessment” (ADAD).

SECTION

TWO-NEW EDITION

ITEMS IN THE FIFTH OF THE ASI

AND DRUG USE PROBLEM

New Items

Route of Administration.

The Fifth Edition of the AS1 has added a column to the “drug grid” portion of the Alcohol and Drug Problem Area that codes the usual or most recent route of administration for each drug listed in the grid as follows: 1 -oral, 2-nasal, 3smoking, 4 -non-IV injection, 5 - IV injection. This is an item that should have been included in earlier versions but is particularly important presently with the widespread use of smokable, freebase cocaine and with the significant threat of HIV and other infectious illnesses associated with IV injection of drugs. Note: The third edition of the AS1 used a “drug grid” that was divided into years and months. In the current version, the months portion of the grid has been removed and users are instructed to round off duration of alcohol and drug use to the nearest full year. LEGAL PROBLEM

Additional

Questions

AREA

AREA

for the AS1

From the tiime of the initial publication of the AS1 (McLellan et al., 1980) we have suggested that users of the instrument should add items of special interest or local concern in each of the AS1 problem areas to sup-

New Items

Criminal Charges. The charges of prostitution and contempt of court have been added since they are common in most patient populations. However, users are en-

A.

204

couraged to add other charges that are locally common or of special interest in their local populations. FAMILY/SOCIAL

PROBLEM AREA

General Changes in the Family Social Section of the AS1

Although the family and social problems of substance abusers are among the most important parts of any clinical evaluation, there has been no satisfactory method for fully capturing the important aspects of a patient’s family and social relationships (McLellan et al., 1985). However, there has been substantial work in this area over the past 10 years, and four important aspects of family and social life have been shown to be related to outcome following treatment. These four areas have been addressed in the Family/Social and Family History sections of the Fifth Edition of the ASI. Family History of Alcohol/Drug Psychiatric Problems. Regardless of whether the influence of the family alcohol and drug problems on the patient’s alcohol and drug use is primarily learned, primarily genetic, or is multiply determined, information on the number of relatives who have experienced psychiatric, drug, and alcohol problems may be of potential value in determining treatment options and in predicting outcome from treatment (Hesselbrock et al., 1982; Rounsaville et al., 1987,199l; Woody et al., 1985). For this reason, the Fifth Edition of the AS1 contains a new, separate section called Family History. This section consists entirely of a grid designed to locate and detail the number and types of substance abuse and psychiatric problems in the patient’s biological family. This section was purposely placed at the beginning of questions in this area to provide an introduction to the more sensitive and personal experiences that are discussed in the revised Family/Social Relationships section. Abusive Relationships. This revised section now contains questions about past and present episodes of physical, emotional, and sexual abuse. We had intended to include questions of this nature in earlier versions of the ASI, but had finally decided against it based on the scarcity (at that time) of trained personnel who could use the revealed information to actually develop and carry out a targeted intervention. Safety and Support of Living Situation. It is now well recognized that the level of safety in the home or residential environment and the level of support for continuing sobriety are significant factors in predicting relapse following treatment. For this reason, additional questions tapping these issues have been added in the Family/Social section of the Fifth Edition of the ASI.

T. McLeIIan et al.

Antisocial Personality Disorder. One clear result from both alcohol and drug abuse treatment studies over the past 10 years has been that patients meeting diagnostic criteria for antisocial personality disorder have a particularly poor treatment prognosis, regardless of the treatment attempted (Rounsaville et al., 1987; Woody et al., 1985). One of the hallmarks of this disorder is the inability to develop or to honor close personal relationships. Because the majority of treatments currently available for substance abusers rely heavily upon a patient’s capacity to form a “helping” or “therapeutic” relationship with a counselor, therapist, and/or a therapy group, we felt it would be important to provide this type of information in the current edition of the instrument. Therefore, the Fifth Edition of the AS1 contains a series of questions that examine the patient’s history of relationship formation. While the AS1 was never designed to provide psychiatric diagnoses, AS1 information on the types of drug use, work, and crime histories, and relationship patterns can be useful for clinically trained personnel in assessing the extent of antisocial behaviors and traits.

SECTION THREE - TEST PROPERTIES THE AS1 FIFTH EDITION

OF

The Fifth Edition of the AS1 has been tested over the past 18 months in the methadone maintenance, cocaine day-hospital, and inpatient and outpatient alcohol treatment programs of the Philadelphia VA Medical Center and in five private treatment programs for alcohol and cocaine abusers. The field tests of the AS1 Fifth Edition have focused exclusively on the assessment of reliability in the new items. We have chosen not to focus our efforts on assessment of the validity of the instrument for two reasons. First, there is ample indication of the validity of the earlier Edition of the AS1 in several settings and languages (McLellan et al., 1985; Kosten et al., 1985; Rogalski, 1987; Hendricks et al., 1988), and these items comprise 135 of the 161 items in the Fifth Edition, Secondly, the new items that have been added have been selected expressly because there has been significant clinical research information indicating that they are valid predictors of treatment response in this population. Initial Testing

Forty-two patients were randomly selected from the detoxification, methadone maintenance, and cocaine rehabilitation programs of the Philadelphia VA Medical Center. These patients participated in early, developmental testing of the instrument to refine the wording of the questions to insure patient understanding, to develop the most efficient interview format, and to finalize the instructions contained in the administration

Revised ASI manual. Interviewers indicated that there were no general problems with the administration of the new items and that all patients test-interviewed with the Fifth Edition showed the ability to comprehend the questions and to use the answer formats correctly. The average time of administration is now 50 minutes to one hour, but this varies with the number and severity of problems presented (range 20 to 70 minutes). Reliability Studies. Following this developmental testing, formal reliability testing of the instrument was begun focused on the test-retest properties of the new items. Interviewers. Six members of the clinical research staff were involved with the testing of the new questions. All were experienced users of the earlier versions of the AS1 and all were provided a 3-hour, manual-based training session in the use of the new version prior to testing. Patient Samples. Four groups of 25 patients each were selected from representative treatment settings and drug problem backgrounds (inpatient opiate and alcohol detoxification patients; inpatient, partial hospital, and outpatient alcohol and cocaine rehabilitation patients; methadone maintenance patients). Sixty of these subjects were uninsured, lower socioeconomic strata subjects receiving public treatment at the VA programs; 40 were privately insured, middle to upper-middle socioeconomic strata subjects recruited from private treatment facilities. Of these subjects 73 were male and 27 (15 cocaine abusers, 12 alcohol abusers) were female; 62 patients were black, 37 were white, and one was Puerto Rican-Hispanic. Although formal diagnostic testing was not performed, it was clear that all patients met DSM-III-R criteria for dependence on their primary drug, and a majority met dependence or abuse criteria on a second drug. All patients volunteered for the studies and signed informed consent statements. Procedure. The test-retest paradigm was approved by our Human Subjects Committee and has been described in our earlier work (McLellan et al., 1985). It involved an initial interviewer completing an entire interview, followed by a 2-day interval, followed by a second (different) interviewer approaching the patient with the following story. “I’m sorry but we seem to have mistakenly destroyed the data from your first interview. Could we go through this again? I will be glad to give you 5 dollars for your extra time. Let’s just go through the whole interview again. You don’t have to try to remember the answers you gave before.” All patients were paid 5 dollars and all were apprised of the real reason for the second interview following the procedure. No patient indicated any significant annoyance when apprised of the deception.

205 Results. The reliability results were very clear and did not differ between male and female subjects or among the different treatment settings, socioeconomic classes, or drug preference subgroups. All new items on the AS1 Fifth Edition showed test-retest reliability coefficients of .83 or higher (Cohen’s kappa; Cohen, 1988) and exact agreement rates of 76% to 100% over the two day test-retest interval. These results are quite consistent with the data from similar studies conducted with previous editions of the AS1 (McLellan et al., 1985) and indicate very satisfactory reliability. The few areas of discrepancy were typically due to interviewer confusion in the interpretation of the new questions, and the results of the reliability testing were used to improve the instructions in the revised AS1 User’s Manual. SECTION FOUR - COMPARATIVE DATA ON COMPOSITE SCORES AND SEVERITY RATINGS Composite

Scores

In program evaluation and clinical research where there are usually multiple measures of patient characteristics (often in multiple measurement domains), it is often useful to create summary measures such as composite or factor scores. These summary scores offer distinct statistical advantages such as greater reliability of measurement and greater statistical power when measuring change (e.g., from admission to follow-up). In addition, probability adjustments (e.g., Bonferroni corrections) that would be necessary for multiple comparisons among large sets of single variables can often be avoided by using fewer summary measures (Cohen, 1988). A separate manual describes the rationale for the composite scores used in the AS1 and shows the formulas for calculating them (McGahan, Griffith, & McLellan, 1986). The composite scores have been developed from combinations of items in each problem area that are capable of showing change (i.e., based on the prior 30-day period, not lifetime) and that offer the most internally consistent (Cronbach’s alpha) estimate of general problem status in each area. The complicated formulas used in the calculation of these composites (e.g., log conversions, multiple divisions, etc.) are necessary to insure equal weighting of all items in the composite. A Special Note Regarding the Use of the New Items in Composite

Scores

It should be clear that we have not used any of the new items in the calculation of the composite scores or in the determination of severity estimates. Obviously, the use of additional information for these purposes would reduce the comparability of the resulting scores across

206 sites and time points. Thus, it is important to stress that earlier ASI versions will still provide comparable data on the composite scores and on the majority of items in the current version. While these composite scores have been very useful to researchers as measures of change, they have had almost no value to clinicians as indications of current status in a problem area, since there have been no “standards” or “normative values” against which to compare individual or group scores. It seems logical and useful to have a single set of mean values (i.e., “norms”) for the AS1 Composite scores that would reflect the nature and types of problems shown by the “average patient” at the start of treatment. Further, this type of “information” might have clinical value for patients (so they might see their problems compared with others like them) and for staff (possibly as a means of assigning patients with more severe problems to more intensive treatments. The major problem with the development of “norms” is the selection of the standard group upon which these will be based. We have of course considered this problem over the past several years and have discussed the use of at least three types of comparison groups. First, it would be possible to develop AS1 mean values on nonproblematic or “normal” individuals. This is unsatisfying because of the inherent lack of information provided by such a comparison. The composite scores and severity ratings are measures of severity defined as “need for treatment” and scores of “0” on either of these measures were purposely designed to indicate “no problem, treatment not necessary.” While it is likely that some individuals among the population of nonpatients could actually have some need for treatment in one or more areas, the mean of the severity ratings and composite scores among true “normals” would almost certainly approach “0.” Further, the distributions of these scores for such a population would likely be skewed, or at least not normally distributed, thus presenting problems for statistical comparisons between samples and that population. Finally, the information that would be provided by such a comparison is actually already available. That is, it is already possible to test whether the mean of a distribution of scores from any group of individuals (e.g., a patient sample) differs from zero, the assumed value for a nontreatment-seeking, “normal” population, using simple statistical procedures (e.g., t tests). A second “standard” comparison group considered was a proportional representation of substance abuse patients currently in treatment on a national level. The Office of Technology Assessment (OTA) has recently surveyed all public and private alcohol and drug abuse treatment facilities in the United States and has published a demographic and “drug of choice” profile of this population (Office of Technology Assessment, 1990). Given the availability of ASIs from patients ad-

A. T. McLellan et al. mitted to a wide range of treatment facilities, it would be possible to proportionately represent the OTA population and thereby create a standard “in-treatment” group that would provide at least an approximation of a representative population. Again, this type of comparison group would offer useful information, but such a comparison group would only reflect the problem profiles of patients in treatment during 1989-1990, and this will likely change as new drugs enter the arena (e.g., “ice”), as other drugs become less popular (e.g., LSD); or if there are changes in the age, race, sex distributions, or socioeconomic status and drug use chronicity of patients in treatment over the coming years. A Suggested Solution. It is apparent that substantial variability exists within the alcohol and drug-abusing patient populations. Variables such as primary drug of abuse, socioeconomic status, age, race, and type of treatment program entered (public vs. private; inpatient vs. partial hospital vs. outpatient) are not uniformly distributed across patient samples and yet are major contributors to treatment assignment, treatment prognosis, and posttreatment adjustment. Thus, rather than attempt to find or create single set of “norms,” we have concluded that the maximum comparative information possible might be derived from the presentation of mean values for a number of different patient groups representing different types of individuals and different treatment settings. Further, the data on these subgroups may be helpful in providing a general comparative overview of the nature and severity of treatment problems across the subgroups. Thus, in the remainder of this paper we report the background characteristics, lifetime data, and mean values for ASI composite scores and severity ratings among selected samples of substance abusers who we feel are representative of many of the subject populations presently assessed by the ASI. It should be noted that there are still limitations associated with this approach. For example, not all possible patient and treatment subgroups are available at this time. Further, the great majority of patients in the samples available are male. This does reduce the information value of the data, but we feel it is at least a reasonable start toward the compilation of a “directory” of AS1 scores for standard forms of treatment cross referenced by sex and type of major drug problem. This is an invitation to interested parties to share ASI admission data from samples (N 2 50) of patients that are not shown here. This information will be compiled into a regularly updated directory and made available at no charge through our office. Demographic and Background Characteristics. Table 1 presents demographic and background characteristics as well as lifetime patterns of substance use and treatment histories of 7 (4 male and 3 female) subject groups

Revised ASI

207 TABLE 1 ASI Background Variables for Selected Subgroups of Substance Abusers

Opiate Abusers

Mean age (SD) Race (Percent) Black White Years education, Mean (SD) Marital status (Percent) Never married Married Divorced/Separated Living situation (Percent) WlSexual partner W/Family W/Friends Alone No stable arrangement Mean years of (SD): Alcohol abuse Heroin abuse Cocaine abuse More than one drug Criminal convict., Mean (SD) Prior treatments, Mean (SD): Alcohol abuse Drug abuse Psychiatric disorder Medical condition

Alcohol Abusers

Cocaine Abusers

Multiple Abusers

Male !-I=205

Female r/=113

Male n= 129

Female

Male

n = 76

n= 164

39 (7)

36 (7)

40 (6)

36 (6)

34 (7)

31 (6)

36 (6)

33 (6)

72 25 11.3 (2)

36 56 11.4 (1)

65 30 12.1 (2)

46 51 13.1 (3)

76 26 12.6 (2)

56 42 13 (3)

76

41

E.7

37 30 33

34 30 36

30 20 50

29 34 37

32 36 32

44 22 34

17 44 39

37 27 36

40 35 4 15 6

45 33 6 6 6

51 21 5 20 3

45 39 4 6 6

39 40 5 12 4

37 44 3 12 4

36 32 22 6 2

45 43 2 6 4


2 1 3 5

(4) (3) (3) (4) 1 (3)

4 4 3 6

3 3 2 3

2 (3) 3 (3) 2 (3) 5 (11)

7 14 4 13 4

(6) (6) (6) (9) (7)

3 9 2 6 3

(4) (6) (3) (6) (4)

12 (6)
1 (3) 7 (6) 1 (3) 5 (3)

1 5 1 7

(1) (4) (3) (7)

6 (7) (3)


Female

n=53

7 (7) (4) 4 (4) 5 (6) 4 (4)

7 (7) 0 2 (4) 3 (6)

1 (1)

1 (2)

3 (3)

3 (4)

1 (2)

1 (1)

4 (5)

2 (6)


1 (1)

1 (2) 1 (1)

1 (2) <1 (2)

1 (2) 1 (3)

3 (3)

2 (2)

2 (3)

1 (2) 2 (2)

representing major types of substance use problems. Table 2 presents background data on samples of subjects from public and private programs, divided into three common treatment settings: inpatient, partial hospital, and outpatient. Table 3 presents the same characteristics on 4 “special populations” of subjects upon which the AS1 has often been used: substanceabusing prisoners, homeless substance abusers, psychiatrically ill substance abusers, out-of-treatment opiate users, and pregnant substance abusers. Space limitations prevent exhaustive descriptions of these samples and the programs from which they were obtained. Only brief descriptions are provided here, but interested parties may request additional information from the senior author.

Table 1 Subjects The Alcohol, Opiate, and Cocaine groups were drawn from admissions to Philadelphia area public and private, inpatient, outpatient or partial hospital treatment programs. A majority of these (approximately 55%) were recruited from the Philadelphia VA Medical Center. Female subjects were all recruited from non-VA treatment programs. Subjects in each of these groups

(1)

1 (1)
(1)

Male n=lOO

(3)

(2) (2) (I) (3)

Female

n = 74

z.2

1

(2)

(5) (5) (3) (5) (3)

had greater than 3 years of problematic use (i.e., intoxicating doses at a frequency of 3 times per week or more) of a single drug and less than one year of problematic use of any other drug. Subjects in the Multiple Drug group were drawn from the same population of patients but had no clearly differentiable single drug problem using the criteria above. The modal patterns of drug use in this group were alcohol and cocaine, alcohol and heroin and alcohol, cocaine and marijuana.

Table 2 Subjects The public treatment subjects were drawn from male and female, alcohol, cocaine, and (primarily) alcohol/ cocaine patients treated in the VA and Philadelphia City programs. Private treatment subjects were drawn from male and female, alcohol, cocaine, and (primarily) alcohol/cocaine patients treated at the Carrier Foundation and in five Philadelphia programs through referral from the Integra Employee Assistance Program. Inpatient subjects were admissions to 20- to 30-day rehabilitation programs in hospital and nonhospital residential settings. Partial hospital patients were in 30to 60-day rehabilitation programs that met a minimum of 25 hours per week and usually 5 days per week. Out-

208

A. T. McLellan et al. TABLE 2 ASI Background Variables for Selected Subgroups of Substance Abusers

Type of Program Inpatient Public n= 116 Mean age (SD) Race, Percent Black White Years education, Mean (SD) Marital status, Percent Never married Married Divorced/Separated Living situation, Percent W/Sexual partner W/Family W/Friends Alone No stable arrangement Mean years of (SD): Alcohol abuse Heroin abuse Cocaine abuse More than one drug Criminal convict., Mean (SD) Prior treatments, Mean (SD): Alcohol abuse Drug abuse Psychiatric disorder Medical condition

Partial Hospitalization Private

n = 81

Public n= 150

Outpatient

Private

Public

n = 99

n = 70

Private n=155

36 (6)

36 (9)

39 (7)

40 (8)

38 (8)

41 (9)

76 21 12.1 (2)

42 57 13.2 (1)

81 16 12.1 (2)

61 39 12.5 (2)

71 23 11.4 (2)

82 18 12.1 (2)

39 16 43

40 30 30

26 23 50

29 40 30

31 13 50

20 52 28

48 25 5 17 5

53 26 4 16 1

50 25 2 19 4

54 25 8 12 1

48 23 3 25 1

59 23 4 10 4

14
(7) (2) (3) (4) (1)

12 1 2 5

1 1 1 2

(1) (1) (2) (2)

1 1 1 2

(9) (1) (3) (6)

1 (1) (2) (2) (1) (2)

patients were in 30- to 60-day rehabilitation programs that met less than 20 hours per week and usually one or two days per week. The great majority of all treatment programs (regardless of setting) were traditional, abstinence oriented, 1Zsteps and AA/NA-oriented and used group therapy, education, and social services as the primary therapeutic components.

Table 3 Subjects The Pregnant Substance Abusers group included women who were admitted to the Emory University Treatment Program for Mothers in Atlanta, Georgia. All of these women were pregnant at the time of the interview and many had additional children at home. The “Out-of-Treatment” group consists of male and female IV opiate abusers who had not been in treatment for at least one year. These patients were recruited by asking methadone maintenance patients to refer associates from their same neighborhoods who also used opiates but were out of treatment. Data on the incarcerated groups derive from the admission information taken by the Hillsborough County

(8) (2) (2) (6) (1)

10 1 1 4 1

(9) (4) (3) (6) (1)

1D
2 (2) (2) 1 (1) 3 (2)

1 1 1 2

(2) (2) (1) (2)


15

(6) (1) (2) (6) (1)

2 (3) (2) (1) 1 (2)

9 (16) 1 (3) 1 (2) 3 (5) 1 (4) 1 1 1 1

(1) (1) (5) (1)

(Florida) Sheriff’s Office Substance Abuse Treatment Program, a 5-week, in-jail program for drug-dependent inmates. This population had been incarcerated for l-3 weeks prior to the evaluation. The Psychiatrically Ill Substance Abuse group was comprised of male and female patients admitted to outpatient psychiatrically oriented substance abuse treatment programs in the Philadelphia area. The majority of these patients had a diagnosis of schizophrenia. Only those subjects who actually received a substance abuse or dependence diagnosis (by DSM-III-R criteria) were included in the sample. The “employer-coerced” patients were drawn from the population of employed subjects (usually union transportation workers) who were forced into substance abuse treatment based on the finding of a positive test result from an on-site, random urinalysis (drugs were usually marijuana and/or cocaine). The sample of homeless substance abusers were drawn from admissions to the Stabilization Services Project, one of nine Community Demonstration Projects funded by NIAAA. These male subjects were recruited from seven public detoxification centers located

Revised ASI

209 TABLE 3 ASI Background Variables for Selected Subgroups of Substance Abusers

out of

Mean age (SD) Race, Percent Black White Years education, Mean (SD) Marital status, Percent Never married Married Divorced/Separated Living situation, Percent W/Sexual partner W/Family W/Friends Alone No stable arrangement Mean years of (SD): Alcohol abuse Heroin abuse Cocaine abuse More than one substance Criminal convict., Mean (SD) Prior treatments, Mean (SD): Alcohol abuse Drug abuse Psychiatric disorder Medical condition

Pregnant Abusers

Trt. Abusers

Female n= 100

Male l-l = 88

Psych

Incarcerated

III Abusers

Abusers

Coerced Thru Urine Test

Homeless Abusers

Male n = 260

Female

Male

Male

Male

n = 92

n = 75

n = 99

n= 150

28 (4)

38 (8)

30 (7)

29 (6)

46 (6)

41 (9)

35 (10)

92

78 22 11.3 (1)

57 39 11.3 (2)

47 48 11.3 (2)

62 36 11.4

82 18 12.2 (2)

40 56 11.2 (2.6)

66 12 21

46 17 35

54 16 30

45 13 42

61 17 22

20 52 28

64 3 33

35 42 8 10 5

30 33 6 16 15

44 34 5 12 5

32 40 8 10 10

6 38 12 32 12

61 14 10 15 0

16 17 10 19 31

1:.6

3 0 2 2 1.5

(2)

(4) (0) (3) (4) (5)

<1 (
10 (10)

11

(9)

5 12 3

(5) (8) (4)

2 4 1 3

(4) (5) (2) (3)

4 1 5 5 4
(5) (4) (5) (6) (4) (1)

1 (2) 1 (2) 2 (3)

3 1 5 4 3
(5) (3) (4) (6) (3)

(1) 1 (2) 1 (3) 4 (3)

6 0 2 6
(2)

(5) (0) (3) (5) (2)

9 (10) 1 (1) 1 (2) 3 (5)

1 (1)

13 2 2 7 2.5

(4) (2) (4) (6)

1 1 1 1

14 (31) 3 (9) 1 (3) 4 (11)

(1) (1) (1) (1)

(8) (4) (2) (4) (4.9)

All data except severity ratings reflect the 30 days prior to the interview. Severity ratings are lifetime estimates of problem severity. Factor scores vary from 0 to 1, with larger values indicating greater severity.

in and around Boston. All these participants stated that their usual residence was the “streets.” As can be seen from an inspection of the three tables, there are marked differences in demographic and background characteristics among these groups. While no statistical comparisons were performed since this was not pertinent to the aims of the paper, it is obvious that the groups differ in almost all of the measures selected, particularly the lifetime patterns of drug use and the history of prior medical, psychiatric, and substance abuse treatments.

Status During the 30 Days Prior to Admission. Tables 4 through 6 present AS1 composite scores, interviewer severity ratings, and representative items from each of the AS1 problem areas for the subjects described in Tables 1 through 3. As in the case of the background information, the information in Tables 4-6 shows major differences among the groups in the patterns of drug and alcohol use and also in the nature and severity of problems in the “addiction related” areas measured by the ASI. Again, no statistical tests were performed to

compare the groups, but it is evident that substantial differences exist. All data were obtained from trained AS1 interviewers, usually as part of a pretreatment evaluation. Thus the information represents the 30-day periods prior to the start of treatment. The “Out-of-Treatment” group was simply interviewed at a time convenient to them. Incarcerated inmates were asked about the 30 days prior to their incarceration (approximately l-3 weeks prior to interview). Interviewer severity ratings were estimated using the two-step strategy described in the AS1 Administration Manual. Severity ratings were not used in the homeless sample and therefore are not included in Table 6. The Composite scores were calculated using the published formulas (McGahan et al., 1986) on the computer scoring disk that is part of the ASI training package (available from the senior author). No information was used in problem areas where the interviewer had rated the information as being significantly distorted by misrepresentation and/or lack of understanding. In cases where more than 3 of the 7 problem areas were so rated, the entire AS1 was elimi-

A. T. McLeIlan et al.

210 TABLE 4 ASI Scores in Selected Subgroups of Substance Abusers Alcohol Abusers

Opiate Abusers Female n=113

Female

Male

Female

n= 129

Female n = 78

Male

n = 205

n= 164

n = 43

n= 100

n = 53

Medical composite (#) Medical severity (’ ) Days of med. probs.

,220 (.330) 2.3 (2.2) 5 (10)

.313 (.354) 3.0 (2.5) 7 (9)

.255 (.327) 2.4 (2.3) 8 (10)

.120 (.236) 0.4 (0.9) 4 (9)

,267 (.333) 2.8 (2.1) 6 (10)

,186 (.295) 1.6 (2.0) 4 (9)

,226 (.188) 1.8 (1.9) 5 (5)

,138 (.294) 1.1 (1.5) 4 (8)

Employment composite Employment severity Days worked Money earned ($)

,820 3.8 6 244

(.228) (1.7) (9) (449)

,803 3.8 5 157

(.267) (2.1) (8) (347)

,702 3.4 7 325

(.269) (2.1) (9) (425)

,392 1.2 10 771

(.285) (2.1) (10) (821)

,697 3.0 IO 410

(.284) (2.2) (10) (478)

,483 2.0 13 780

(.346) (2.6) (9) (653)

.608 4.1 11 605

(.311) (3.3) (4) (314)

,448 2.1 11 610

(.309) (2.5) (9) (580)

Alcohol Alcohol Days Days

,226 3.0 13 8

(.255) (2.3) (13) (12)

.098 1.6 7 3

(.174) (1.9) (10) (8)

,643 6.4 15 15

(.226) (1 .O) (8) (8)

,631 5.5 17 14

(.177) (1 .O) (9) (10)

,257 3.2 9 7

(.263) (2.3) (9) (8)

.241 3.4 8 6

(.322) (2.6) (9) (9)

,289 3.3 9 5

(.169) (3.7) (2) (3)

,308 3.9 10 8

(.287) (2.3) (8) (9)

Drug use composite Drug use severity Days opiate use Days cocaine use Days depressant use Days marijuana use

,265 5.3 19 10 5 4

(.131) (1.2) (14) (12) (9) (8)

.314 5.5 12 8 6 4

(.147) (1.4) (13) (12) (10) (7)

.031 1.2 <1 1 2 1

(.067) (2.0) (1) (2) (6) (3)

,023 0.7 0 0 1
(.061) (1.7) (0) (0) (5) (1)

.245 5.2 0 12 <1 3

(.082) (0.8) (0) (8) (1) (6)

.256 5.4 2 9 1 1

(.064) (1 .l) (6) (8) (7) (1)

,284 6.8 4 8 3 6

(.211) (4.7) (2) (3) (1) (2)

,155 5.2 1 8 1 3

(.103) (2.0) (6) (9) (4) (5)

Legal composite Legal severity Crime days Illegal income ($)

.152 2.6 7 352

(.200) (1.6) (11) (833)

,103 1.5 2 305

(.214) (1.9) (6) (566)

.074 1.4
(.174) (1.9) (1) (94)

,050 0.2 0 0

(.134) (0.8) (0) (0)

.089 1 .o 2.3 1 10

(.146) (1.5) (6.5) (308)

.041 0.5 <1 45

(.126) (1.5) (4) (196)

.046 1.3 1 47

(.069) (1.5) (1) (31)

,074 0.3 2 105

(.153) (1.5) (9) (743)

Family/Social composite Family/Social severity Conflicts w/family Conflicts w/others

.157 3.0 1 1

(.189) (1.9) (5) (4)

,300 4.0 6 2

(.245) (2.1) (10) (5)

,202 3.1 2 1

(.235) (2.0) (4) (2)

.348 3.6 2
(.255) (2.5) (4) (1)

,304 3.9 3 2

(.226) (2.0) (6) (6)

.224 3.6 2 3

(.243) (2.9) (4) (7)

.502 4.7 5 2

(.329) (3.5) (3) (2)

.272 3.7 2 3

(.224) (2.4) (4) (5)

Psychiatric composite Psychiatric severity Days psych. probs.

,155 (.213) 2.8 (2.3) 5 (9)

Male

use composite use severity drinking intoxicated

,379 (.194) 4.3 (2.5) 6 (10)

Male

Multiple Abusers

Cocaine Abusers

,236 (.212) 3.4 (2.3) 9(ll)

.242 (.222) 3.7 (2.3) 10 (12)

,245 (.223) 3.3 (2.4) 9 (10)

,258 (.213) 3.9 (2.7) 11 (12)

.243 (.217) 3.4 (4.4) 11 (4)

.225 (.217) 3.5 (2.7) 10 (11)

All data except severity ratings reflect the 30 days prior to the interview. Severity ratings are lifetime estimates of problem severity. Factor scores vary from 0 to 1, with larger values indicating greater severity.

nated from the data base. In this regard, the data on recent illegal activity from the incarcerated inmates were not used, as this information was sometimes collected by prison officials prior to sentencing. The majority of the information was missing, and the interviewers had reason to believe that much of the remaining data on criminal activity was inaccurate. However, these interviewers also indicated that the data from the remaining sections of the ASI had satisfactory validity. DISCUSSION There have been many changes in the field of drug and alcohol dependence treatment since the ASI was originally introduced in 1980. The emergence of new drugs such as “crack” and the profound increase in the concurrent abuse of multiple substances are merely the most obvious among a series of important changes in the patient population. There have also been major changes in our knowledge about substance abuse and substance abuse treatment since 1980. Information has

accumulated suggesting the importance of factors such as a family history of substance abuse, antisocial personality disorder, and general psychiatric symptomatology in predicting the response to substance abuse treatment (Hesselbrock et al., 1982; Rounsaville et al., 1987,1991;Gawin&Kleber, 1986; Woodyetal., 1985). Although the Addiction Severity Index (ASI) has been found reliable and valid in a number of different settings over the past 12 years, there have also been problems with the instrument. Some of the most common problems with the AS1 were discussed in the first section of the paper. In summary, the instrument cannot be self-administered and does require an interview format. The instrument cannot be used with adolescents, although there are now a number of specially developed adolescent versions. The interview was developed for and tested exclusively with treatment-seeking populations. Although it has now been used with “special populations” such as homeless substance abusers, psychiatrically ill substance abusers and substance abusing prisoners, there is clear need for reliability and validity testing with these groups.

Revised ASI

211 ASI Scores

TABLE 5 in Selected Subgroups

of Substance

Abusers

Type of Program Inpatient Public n= 116

Partial Hospitalization Pirvate

n = 81

Public n= 150

Outpatient

Private

n = 99

Public

n=

Private

70

n=

155

Medical composite (#) Medical severity ( * ) Days of med. probs.

,244 1.9 5.4

(.331) (2.2) (8.1)

,112 0.9 2.3

(.212) (1.5) (4.3)

,240 1.7 5.1

(.343) (2.1) (6.1)

.087 0.9 3.4

(.201) (1.6) (7.4)

,150 1.4 4.4

(_ 143) (2.3) (6.2)

.142 1.1 4.0

(.281) (1.4) (4.3)

Employment composite Employment severity Days worked Money earned ($)

.744 3.5 7.4 267

(.264) (1.9) (9.1) (424)

,334 1.7 17.8 1344

(.267) (2.2) (7.0) (834)

,647 3.0 9.2 419

(.271) (2.2) (9.1) (522)

.276 0.9 19.1 1377

(.190) (1.9) (5.5) (831)

,447 3.0 11.2 459

(.291) (2.5) (9.1) (412)

,309 0.5 17.2 1193

(.196) (0.9) (5.3) (944)

Alcohol Alcohol Days Days

,484 4.7 11.7 9.8

(.241) (1.5) (8.3) (8.4)

,645 5.8 20.1 18.9

(.323) (1.7) (10.9) (1 1.5)

,500 4.7 12.5 10.1

(.255) (1.6) (7.8) (8.2)

,278 3.2 10.1 9.0

(.275) (2.1) (9.8) (9.8)

,510 4.3 12.3 9.3

(.235) (2.7) (9.4) (6.2)

.174 2.1 6.3 5.4

(.160) (1.8) (7.1) (6.7)

Drug use composite Drug use severity Days opiate use Days cocaine use Days depressant use Days marijuana use

.137 3.1 0.1 6.3 1.1 2.6

(.075) (1.1) (0.2) (5.3) (2.2) (5.1)

.178 4.8 2.3 7.8 1.2 6.7

(.156) (2.5) (7.0) (9.7) (5.4) (10.5)

.141 3.4 0.1 6.0 1.8 2.2

(.075) (1.3) (0.5) (4.7) (4.0) (4.6)

.056 2.4

(.087) (2.4)

.I04 3.0 1.1 4.4 1.1 2.2

(.125) (3.3) (1.5) (5.7) (3.1) (2.6)

.046 2.5

(.044) (1.6)

Legal composite Legal severity Crime days illegal income ($)

.075 1.1 1.1 89

(.151) (1.5) (3.8) (193)

,090 1.4 1.2 173

(, 178) (1.5) (5.3) (697)

.086 1.3 1.1 37

(.166) (1.6) (2.6) (35)

.024 0.5 1.2

.044 0.7 0.5 27

(.lOl) (1.6) (1.3) (65)

.OlO 0.4

Family/Social composite Family/Social severity Conflicts w/family Conflicts w/others

.277 3.3 2.7 1.5

(.241) (2.1) (5.6) (3.6)

.290 3.9 4.5 1.1

(.267) (2.3) (7.8) (1.7)

.232 3.0 2.3 0.8

(.238) (1.9) (5.4) (4.1)

.124 2.4 2.3 1.3

(.197) (2.6) (5.6) (3.2)

.152 1.5 2.4

(.158) (1.5) (2.4)

,101 1.3 1.3

Psychiatric composite Psychiatric severity Days psych. probs.

,224 3.1 9.1

(.220) (2.4) (10.6)

.213 2.8 8.7

(.223) (2.6) (10.8)

.256 3.6 9.5

(.21 1) (2.2) (11.2)

.l 13 (.189) 2.1 (2.7) 5.6 (9.8)

.116 2.2 6.3

use composite use severity drinking intoxicated

All data except severity ratings reflect the 30 days prior to the interview. scores vary from 0 to 1, with larger values indicating greater severity.

Given these problems and shortcomings, it became important to update the interview to keep pace with the significant changes in the patient population and to assist clinicians and researchers in their attempts to advance our understanding in this field. In order to improve the general utility and contemporary value of the ASI, we have updated the instrument, added a significantly more detailed and comprehensive user’s manual, and added new items to capture important aspects of the patients lives that were previously not covered. No new items have been added in the medical, employment, alcohol use, or psychiatric sections. This is because there are already a number of excellent, specialized supplemental instruments available for detailed evaluations of AIDS risk behaviors, depression and antisocial personality disorders, alcohol and drug diagnoses, and other important areas of concern to clinicians and researchers working with these populations.

Severity

0 2.4 1.3 2.1

(0) (4.9) (1.5) (5.2) (.103) (1.2) (3.0)

3 (32)

0 (0)

ratings are lifetime estimates

(.201) (2.4) (4.1) of problem

0 1.7 1.2 4.1

(0) (3.1) (2.4) (5.5) (.045) (0.7)

0 (0) 0 (0) (.113) (2.1) (1.3)

0 (0) ,103 1.7 5.2 severity.

(.092) (1.9) (8.1) Factor

Items coding route of administration have been added to the drug use section, and additional charges have been added to the legal section. The most significant changes have occurred in the Family/Social section of the ASI, corresponding to the major developments in this area over the past 12 years and the relatively unsatisfactory level of information in that area of the previous editions of the instrument. A “Family History” section has been added describing the patterns of alcohol, drug, and psychiatric problems in the patient’s biological family. Items were also added to the Family/Social Relationships problem area describing the patient’s patterns of supportive relationships; as well as past and recent history of emotional, physical, and sexual abuse in the home environment. It is important to point out that the changes described

have not changed the composition or scoring of the seven composite scores. We feel this is important for

A. T. McLellan et al.

212

TABLE 6 ASI Scores in Selected Subgroups of Substance Abusers

out of

Psych Ill Abusers

Incarcerated

Coerced Thru Urine Test

Pregnant Abusers

Trt. Abusers

Female n=lOO

Male

Male

Female

Male

Male

n = 68

n = 260

n = 92

n = 75

n = 99

Medical composite (#) Medical severity ( l) Days of med. probs.

,266 (.307) 2.2 (2.8) 3 (7)

,152 (.295) 1.4 (2.0) 3.5 (a.11

,232 (.317) 1.9 (2.8) 5 (9)

,215 (.296) i .a (2.2) 5 (9)

,416 (.360) 4.8 (3.3) 11 (6)

,098

Employment composite Employment severity Days worked Money earned ($)

,632 (.465) 7.6 (3.2)

,633 (.220)

,673 4.7 12 407

(.267) (3.0) (12) (578)

,797 7.3 5 139

(.250) (2.4) (10) (374)

,811 (.606) 7.5 (4.8) <1 (1)

,300 (.346)

Alcohol Alcohol Days Days

,123 (.143) 1.5 (1.4)

,237 2.9

(.247) (3.2)

,152 3.5

(.225) (3.6)

,260 3.5

use composite use severity drinking intoxicated

2 (6)

31 (100)

3 (6) 2 (5)

Drug use composite Drug use severity Days opiate use Days cocaine use Days depressant use Days marijuana use

.169 6.3

Legal composite Legal severity Crime days Illegal income ($)

,039

Family/Social composite Family/Social severity Conflicts w/family Conflicts w/others

,460 4.6 3 2

Psychiatric composite Psychiatric severity Days psych. probs.

,220 (.226) 3.7 (2.7) 6 (10)

(.107) (2.2)

1 (2) 3 (6)
1.7 (1.7) 1 (3) 7 (54) (.227) (2.4) (7) (6)

6.8

(1.7)

4 (7.5)

i a4 (373) ,307 4.1

(.270) (2.3)

16 (13) 13 (13) .271 7.9 17 16 3 5 ,220

(.113) (1 .O) (9) (12) (6) (6) (.203)

2.7 (1.6) 12 (12) 547 (964) .155 2.9 1 1

(.171) (1.6) (5) (5)

.135 (.210) 2.3 (2.2) 4 (9)

Abusers

9 (7) 5 (9) .252 7.5 1 16 2 6

7 (11) 3 (7)

(.116) (2.1) (4) (12) (6) (10)

.244 (.106) 7.5 (1.5)
NA 5.6 (2.4) NA NA .329 3.7 4 3

11 (36)

(.236) (2.7) (6) (7)

.162 (.201) 2.6 (2.7) 7 (11)

(3) (ii) (4) (5)

NA 5.4 (3.0) NA NA ,305 4.3 1 1

(.221) (3.0) (12) (4)

,267 (.234) 4.6 (2.7) 11 (12)

(.255) (2.9)

4 (6) 4 (6) .232 6.6
(.122) (3.4)

(3) 2 (4) 4 (4) 6 (6)

Male n=150 ,208

1.2 (2.2) 3 (6)

3.6 19 (4.6) (6) 1305 (765) .357 4.1 12 9

(.i69)

NA 5 (10) .649 (.169) 5N(:) 165 (347)

(.269) (3.7) (6) (6)

.566 (.247) NA 20(11) 16 (12)

.092 (.132) 4.5 (3.6)

,135 (.154)


.021 0.7 <1 5

(.lll) (1.2) (1) (10)

,027

,236 3.1 3 2

(.201) (2.2) (3) (4)

,140

.366 (.233) 6.2 (2.2) 16 (6)

(.289)

Homeless Abusers

NA

(1) 5 (4) 0 (0) 3 (6) (.154)

1.01 (1.3) (2) 34 (121) (.169)

2.41 (3) (1.7)
2 (7) 7 (11) 1 (5) 3 (6) ,074

(.145)

3N(;) 172 (622) ,257

(.226) 3Np8) 2 (6)

,249

(.223) NA 10 (12)

All data except severity ratings reflect the 30 days prior to the interview. Severity ratings are lifetime estimates of problem severity. Factor scores vary from 0 to 1, with larger values indicating greater severity. NA = not applicable.

maintaining continuity in data collected with earlier versions of the AS1 and in maintaining the reliability and validity of these measures. We feel these additional items and the improved instruction manual will add significantly to the utility of the instrument. In the final section of this article we discussed the use of “norms” in the substance abuse population and provided mean values for the AS1 severity ratings and composite scores for a variety of representative patient samples including opiate, alcohol, cocaine, and mixed abusers from public and private programs and from inpatient, partial hospital, and outpatient treatment settings. Representative data were also presented for incarcerated substance abusers, pregnant addicted women, homeless males, and psychiatrically ill substance abusers. We hope the data on the composite scores and the severity ratings will be helpful to those in the field who have been interested in comparative

information on their patient samples. We see this as merely the beginning of an accumulating “catalog” of ASI information, cross referenced by age, sex, drug of choice, and type of treatment environment. Interested readers who have samples of 50 or more admission or baseline ASIs to contribute to this “catalog” are encouraged to contact the senior author. We will continue to collect this type of information and to make it available to those interested. Copies of the Fifth Edition of the AS1 and the revised Instruction Manual are available at no charge from the senior author.

REFERENCES Ball, J.C., & Corty, E. (1988). Basic issues pertaining to the effectiveness of methadone maintenance treatment. In Carl G. Lukefeld & Frank M. Tims (Eds.), Compulsory treatment of drug

Revised ASI abuse: Research and clinicalpractice (NIDA Research Monograph 86). Rockville, MD: National Institute on Drug Abuse. Cahalan, D. (1987). Understanding America’s drinking problems: How to combat the hazards of alcohol. San Francisco: JosseyBass. Cohen, J. (1988). Statisticalpower analysis for the behavioral sciences (2nd ed.). Hillsdale, NJ: Erlbaum. Gawin, F.H., & Kleber, H. (1986). Abstinencesymptomatologyand psychiatric diagnosis in cocaine abusers: Clinical observations.

Archives of General Psychiatry, 43, 107- 113. Gawin, F.H., Kleber, H.D., Byck, R., Rounsaville, B.J., Kosten, T.R., Jatlow, P.I., & Morgan, C. (1989). Desipramine facilitation of initial cocaine abstinence. Archives of General Psychia-

try, 46, 117-121. Hendricks, V.M., Kaplan, C.D., VanLimbeek, J., & Geerlings, P. (1989). The Addiction Severity Index: Reliability and validity in a Dutch addict population. Journal of Substance Abuse Treat-

ment, 6, 133-141. Hesselbrock, V.M., Stabenau, J.R., Hesselbrock, M.N., Meyer, R.E., & Babor, T.F. (1982). The nature of alcoholism in patients with different family histories of alcoholism. Progress in Neuropsy-

chopharmacological and Biological Psychiatry, 6, 607-614. Kadden, R.M., Cooney, N.L., Getter, H., & Litt, M.D. (1990). Matching alcoholics to coping skills or interactional therapies: Post treatment results. Journal of Clinical Consulting Psychol-

ogy, 57, 851-863. Kosten, T.R., Rounsaville, B.J., & Kleber, H.D. (1985). Concurrent validity of the Addiction Severity Index. Journalof Nervousand

Mental Diseases, 171, 606-610. Kozel, N., & Adams, E. (1986). Epidemiology of drug abuse: An overview. Science, 234, 970-974. Lehman, A.F., Myers, C.P., & Corty, E. (1989). Assessment and classification of patients with psychiatric and substance abuse syndromes. Hospital and Community Psychiatry, 40, 1019-1025. Lubran, B. (1990). Alcohol and drug abuse among the homeless population: A national response. Alcoholism Treatment Quarterly, 7(l), 11-23. McGahan, P., Griffith, J., & McLellan, A.T. (1986). Composite

scores from the Addiction Severity Index: Manual and computer software. Philadelphia, PA: Veterans Administration Press. McLellan, A.T., Luborsky, L., Cacciola, J., &Griffith, J.E. (1985). New data from the Addiction Severity Index: Reliability and va-

213 lidity in three centers. Journal of Nervous and Mental Diseases,

173, 412-423. McLellan, A.T., Luborsky, L., O’Brien, C.P., & Woody, G.E. (1980). An improved evaluation instrument for substance abuse patients: The Addiction Severity Index. Journal of Nervous and Mental

Diseases, 168, 26-33. McLellan, A.T., Luborsky, L., Woody, G.E., Druley, K.A., & O’Brien, C.P. (1983). Predicting response to alcohol and drug abuse treatments: Role of psychiatric severity. Archives of Gen-

eral Psychiatry, 40, 620-625. McLellan, A.T., McGahan, J., & Druley, K.A. (1979). Changes in drug abuse clients 1972-77: Implication for revised treatment.

American Journal of Drug and Alcohol Abuse, 6, 151-162. McLellan, A.T., Woody, G.E., Luborsky, L., O’Brien, C.P., & Druley, K.A. (1982). Is treatment for substance abuse effective?

JAMA, 247, 1423-1427. Assessment. (1990). Theeffectiveness of drug abuse treatment: Implicationsfor controlling AIDS/HIV infection (Publication No. 052-003-01210-3) Washington, DC: US

Office of Technology

Government Printing Office. Rogalski, C.J. (1987). Factor structure of the Addiction Severity Index in an inpatient detoxification sample. International Journal

of the Addictions, 22, 981-992. Rounsaville, B. J., Anton, S.F., Carroll, K.M., Budde, D., Prusoff, B.A., & Gavin, F.I. (1991). Psychiatric diagnosis of treatment seeking cocaine abusers. Archives of General Psychiatry, 45,

43-51. Rounsaville, B.J., Dolinsky, Z.S., & Babor, T.F. (1987). Psychopathology as a predictor of treatment outcome in alcoholism. Ar-

chives of General Psychiatry, 44, 505-513. Smith, I.E., Moss-Wells, S., Moeti, R., & Coles, C. (1990). Characteristics of non-referred cocaine abusing mothers. In L.S. Harris (Ed.), Problems of drug dependence 1989 (NIDA Research Monograph #95, p. 330). Rockville, MD: National Institute on Drug Abuse. Wexler, M.K., Falkin, G.P., & Lipton, D.S. (1988). A modelprison

rehabilitation program. An evaluation of the stayh out therapeutic community. New York: NDRI Press. Woody, G.E., McLellan, A.T., Luborsky, L., &O’Brien, C.P. (198.5). Sociopathy and psychotherapy outcome. Archives of General Psy-

chiatry, 42, 1081-1086.