Clinical Assessment and Substance Abuse Treatment : The Target Cities Experience [1 ed.] 9780791487518, 9780791455937

Citation preview

Clinical Assessment and Substance Abuse Treatment

SUNY series, The New Inequalities A. Gary Dworkin, Editor

Clinical Assessment and Substance Abuse Treatment

The Target Cities Experience

EDITED BY

Richard C. Stephens, Christy K Scott, and Randolph D. Muck

STATE UNIVERSITY OF NEW YORK PRESS

Published by State University of New York Press, Albany © 2003 State University of New York All rights reserved Printed in the United States of America No part of this book may be used or reproduced in any manner whatsoever without written permission. No part of this book may be stored in a retrieval system or transmitted in any form or by any means including electronic, electrostatic, magnetic tape, mechanical, photocopying, recording, or otherwise without the prior permission in writing of the publisher. For information, address State University of New York Press, 90 State Street, Suite 700, Albany, N.Y., 12207 Production by Diane Ganeles Marketing by Michael Campochiaro Library of Congress Cataloging-in-Publication Data Clinical assessment and substance abuse treatment : the target cities experience / edited by Richard C. Stephens, Christy K Scott, and Randolph D. Muck p. cm. — (SUNY series, the new inequalities) Includes bibliographical references and index. ISBN 0-7914-5593-9 (HC : alk. paper) — ISBN 0-7914-5594-7 (pbk. : alk. paper) 1. Drug abuse—United States. 2. Drug abuse—Treatment—United States. 3. Alcoholism—United States. 4. Alcoholism—Treatment— United States. I. Stephens, Richard C. II. Scott, Christy K. III. Muck, Randolph D. IV. Series. HV5825 .C5784 2003 362.29'1860973—dc21 10

9 8 7 6 5 4 3 2 1

2002023126

To my wife Barbara J. Stephens for her wit, wisdom, love and courage. Richard C. Stephens, Ph.D. I wish to express my sincere gratitude to Mark Godley for his unwavering support over the years and to my field team, particularly Floyd, Judy, Kevin, Mick, and Lloyd for their passionate commitment to our work. I would also like to thank the treatment providers in Chicago for all they taught us, and the thousands of clients who participated in the Target Cities project, for their time and patience. Christy K Scott, Ph.D. I want to acknowledge my wife, Bonnie, for her support and encouragement during the many long days, nights, weekends, and the frequent meetings away from home, that were a part of this effort. And thanks to my children (Justin, Rebekah, Jordan, and Joshua), who didn’t completely understand why I had to be away so much, but kept loving me anyway and were always ready to advise me on where to find good music while I was on the road. Randolph D. Muck, M.Ed. We would like to thank the CSAT project officers and their immediate supervisors who worked with the individual sites and assisted in their participation in the development and implementation of the cross-site evaluation: David Thompson, Edith Jungblut, Joyce Johnson, Ray Hylton, Jean Donaldson, Carol Coley, Mady Chalk, Barry Blandford, Jutta Butler, and Mary Lou Andersen. The Editors

This page intentionally left blank.

Contents

Acknowledgments 1.

2.

ix

Background and Overview of the Target Cities Demonstration Program Christy K Scott and Randolph D. Muck Methodological Issues in the Development of the Target Cities Multisite Databases Peter J. Leahy, Richard C. Stephens, Heather K. Huff, and Russell S. Kaye

3.

Participants in the Target Cities Program Ronald E. Claus and René M. Dailey

4.

Identifying Service Needs among Substance Abuse Treatment Participants Mark A. Foss, Nancy Barron, and Cynthia L. Arfken

5.

6.

7.

The Target Cities Participants: From Centralized Intake to Treatment Entry Ronald E. Claus, Nancy Barron, and Kimberly A. Pascual Effects of Centralized Intake on Participant Satisfaction with Treatment and Ancillary Services Christy K Scott, Mark A. Foss, and Richard E. Sherman Implementation of Selected Target Cities Components: Analysis of Matching, Case Management, and Linkages Cynthia L. Arfken, Chris Klein, Elizabeth J. Agius, and Salvatore di Menza

vii

1

17

35

55

73

87

103

viii 8.

9.

10.

11.

Contents Does Centralized Intake Improve Substance Abuse Outcomes?: A Multisite Analysis Joseph Guydish, Alan Bostrom, Sven Klingemann, Christy K Scott, Mark A. Foss, and Nancy Barron Outcomes Before and After Implementing Centralized Intake Services Christy K Scott, Mark A. Foss, and Richard E. Sherman Effectively Assessing and Preparing Inmates for Community Substance Abuse Treatment: The Portland Target Cities Project In-Jail Intervention Michael W. Finigan, Nancy Barron, and Shannon Carey

129

149

165

Lessons Learned from the National Target Cities Initiative to Improve Publicly Funded Substance Abuse Treatment Systems Joseph Guydish, Richard C. Stephens, and Randolph D. Muck

179

References

201

Contributors

215

Index

225

Acknowledgments

The content of this work is solely the responsibility of the authors and does not necessarily represent the official views of the Substance Abuse and Mental Health Services Administration’s Center for Substance Abuse Treatment. The authors wish to acknowledge the leadership and support of the staff at the Substance Abuse and Mental Health Services Administration’s Center for Substance Abuse Treatment during the course of the Target Cities Initiative. Chicago: The Target Cities Project in Chicago and the ensuing studies were made possible by grant number 5 UP95 TI00664 from the Substance Abuse and Mental Health Services Administration’s Center for Substance Abuse Treatment and contract no. PA 00000019 through the State of Illinois Department of Human Services, Office of Alcoholism and Substance Abuse. Cleveland: The Target Cities Project in Cleveland and the ensuing studies were made possible by grant no. I- U95 T100672-02 from the U.S. Department of Health and Human Services, Center for Substance Abuse Treatment, through the Ohio Department of Alcohol and Drug Addiction Services and the Alcohol and Drug Addiction Services Board of Cuyahoga County, Ohio. Detroit: The Target Cities Project in Detroit and the ensuing studies were made possible in part by grant no. 6 U95 T100665-05-5 from the U.S. Department of Health and Human Services, Center for Substance Abuse Treatment and Michigan’s Bureau of Substance Abuse Services, and a grant from the state of Michigan (Joe Young, Sr.). Portland: The Portland Target Cities Project and the ensuing studies were funded through Multnomah County Department of Community and Family Services under a cooperative agreement (no. 93-196) with

ix

x

Acknowledgments

the State of Oregon, Office of Alcohol and Drug Abuse Treatment, Substance Abuse and Mental Health Services Administration. Saint Louis: This Target Cities Project in St. Louis and the ensuing studies were supported by grant no. T100662 from the Center for Substance Abuse Treatment. San Francisco: The Target Cities Project in San Francisco was supported by the Center for Substance Abuse Treatment (grant no. U95T100669) and by the National Institute on Drug Abuse San Francisco Treatment Research Center (grant no. l 9500-9253). The editors would also like to acknowledge the assistance of Heather K. Huff, M.A., and Beth Pfohl, M.S., in the preparation of this manuscript.

CHAPTER

1

Background and Overview of the Target Cities Demonstration Program Christy K Scott and Randolph D. Muck

This volume is an effort to describe the experiences and results of a large, federally funded substance abuse research project known as Target Cities, which was designed to address a multitude of challenges inherent in the country’s approach to treating substance abuse. These challenges included poor treatment infrastructure, accessibility, and quality; service provider capacity; service coordination/integration; and treatment outcome monitoring. Undertaken in 10 cities across the United States, Target Cities projects were funded through the Department of Health and Human Services (DHHS) and its Substance Abuse and Mental Health Services Administration (SAMHSA). SAMHSA is the federal agency charged with improving the quality and availability of mental health and substance abuse treatment and prevention services in the country. Federal funding for the Target Cities projects was made available through SAMHSA’s Center for Substance Abuse Treatment (CSAT). At the project’s outset, CSAT’s administrators recognized that drug and alcohol abuse was a complex, multifaceted phenomenon that impacted on a wide variety of areas in a person’s life, including physical and mental health, family relationships, employment, involvement with the criminal justice system, and housing, as well as other domains. Many treatment facilities across the country were not equipped to deal with the myriad of problems. Accordingly, CSAT issued a call for proposals (e.g., a Request for Applications [RFA]), which invited both public and private substance abuse treatment agencies to design new paradigms for assessing and treating persons struggling with substance abuse (U. S. Department of Health and Human Services [DHHS], 1990).

1

2

Christy K Scott and Randolph D. Muck

The specifications for the creation of the Target Cities projects contained a number of elements (which will be discussed in detail later in this chapter). However, the key components required of each applicant site were a management information system, a standardized assessment process that would include a paradigm to match participants to appropriate treatment facilities, case management services, linkages with agencies across the community whose services could be utilized by Target Cities participants, and both process and outcome evaluations of these efforts. CSAT funded Target Cities programs in two separate waves. This book addresses the efforts of the 10 Target Cities programs funded in the final wave. Unique to this generation of Target Cities was the fact that, in addition to local evaluation efforts, there was also a multisite evaluation that culminated in large multisite databases containing information for thousands of individuals. This book includes studies that utilize both multisite and local site data to address the fundamental questions regarding the success of the Target Cities effort. Even though the multisite databases have been used primarily to evaluate specific components of the Target Cities experience, it is important to note that these databases provide one of the richest sources of data on drug and alcohol users in the country. Detailed information is available on over 40,000 cases in the databases. Several research questions, including many that are not related to the Target Cities Program’s concerns, can be investigated using this extremely valuable set of data. Prior to examining the results of this project, it is important to place the Target Cities initiative in a historical context that outlines the increasingly complex and sophisticated evolution of the ways in which problems of drug and alcohol abuse are, and have been, addressed in this country.

The Evolution of Drug Treatment Policy Several concerns regarding addiction treatment in the United States led to the initiation of the Target Cities Program—treatment agencies’ financial and organizational infrastructure, service accessibility, quality of clinical practice, intra- and interfield collaboration, and outcome monitoring. The following discussion briefly outlines the evolution of these concerns and describes the social policy context within which the Target Cities Program was initiated.

Background and Overview

3

Following America’s “discovery” of addiction in the late 18th and early 19th century (Levine, 1978), the definition of alcohol and other drug-related problems emerged for the first time as a newly christened medical disorder: inebriety. A private network of inebriate homes, inebriate asylums, and private addiction cure institutes emerged in the second half of the 19th century. Many of these institutions were linked through their membership in the American Association for the Study and Cure of Inebriety (Baumohl & Room, 1987). Even at the height of their popularity in the 1880s and 1890s, inebriate institutions were plagued by problems of weak organizational infrastructures, inadequate funding, geographical and financial inaccessibility, inconsistent and even harmful care, and poor continuity between the initiation of recovery in an institutional setting and transfer of that recovery process to natural community environments. Problems of poor service coordination (particularly between inebriety and psychiatric institutions) and lack of continuity of care were pervasive and spawned proposals for the creation of what today would be called a “continuum of care” (Crothers, 1893). These proposals for federal and state involvement in the planning, financing, and regulation of inebriate asylums in the 19th century were never fully implemented or sustained. Support for addiction treatment rapidly dissipated in the opening decades of the 20th century as the country fell under the sway of a series of anti-alcohol and anti-drug campaigns that focused, not on helping the victims of drug addiction, but on legally prohibiting and/or controlling access to psychoactive drugs (White, 1998). The dramatic policy changes brought on by the passage and interpretation of the Harrison Tax Act of 1914 and the passage of the Eighteenth Amendment to the Constitution in 1919 led to the virtual demise of most addiction treatment institutions in America. Most private and state-operated inebriate asylums closed. Brief (1919–24) local experiments in morphine maintenance were terminated under threat of criminal indictment, and access to addiction treatment existed only for the most wealthy within a shrinking pool of private hospitals and sanatoria (White, 1998). Viewing addiction as a criminal rather than medical problem led to the transfer of responsibility for these problems from physicians and hospitals to law enforcement and criminal justice institutions. Two treatment trends emerged in post-Prohibition America: (a) the first federal involvement in addiction treatment via the opening of two federal prison hospitals for narcotics addicts in Lexington, Kentucky (1935), and Fort Worth, Texas (1938), and (b) a multibranched

4

Christy K Scott and Randolph D. Muck

“modern alcoholism movement” that sought local, state, and, eventually, national involvement in the establishment of community-based alcoholism treatment services. These two trends reflect the split in social policy toward dealing with alcoholism and narcotic addiction that followed the repeal of Prohibition. Addicts were isolated from the community in prisons and punished, while at the same time new calls emerged for the provision of community-based alcoholism treatment. The incarceration of narcotic addicts escalated dramatically in the late 1920s and 1930s, which led to increased calls for treatment inside the Federal Bureau of Prisons and subsequent opening of the prison hospitals in Lexington and Fort Worth through the U.S. Public Health Service. In contrast to this policy was the opening of hospital-based alcoholism treatment units in cooperation with local Alcoholics Anonymous groups. There were also problems with the lack of continuity of care that characterized both the treatment of alcoholics and narcotics addicts. A perennial problem of the U.S. Public Health Hospitals in Lexington and Fort Worth was the lack of continuity of care that resulted from trying to treat addicts hundreds or even thousands of miles away from their local communities. These concerns eventually spurred experiments in establishing reentry clinics in communities such as New York and Detroit. Marty Mann, a pioneer organizer and researcher in this field, sought to bridge this gulf between the alcoholism treatment institution and the community when she announced the creation of the National Committee on Education on Alcoholism (NCEA) in 1944 (Mann, 1944). Mann called for the creation of • local hospital detoxification units; • local alcoholism education, assessment, and referral centers; • local alcoholism treatment institutions; and • “rest homes” for those who needed extended convalescence and rehabilitation. Mann’s proposals reflected concerns not only to create treatment resources but to guarantee that such services met standards of quality and accessibility and that such services were coordinated to assure some degree of continuity of care between the elements of what she conceived as a service “system.” The service elements proposed by Mann increased in the 1940s and 1950s—most under the

Background and Overview

5

leadership of local NCEA affiliates or through the private actions of members of Alcoholics Anonymous. In spite of the building blocks of a treatment system, Mann’s vision of a community-based continuum of care for alcoholics went unfulfilled in most communities until the 1970s. New local community–level treatment models emerged in the 1950s and 1960s that could be replicated in communities across the United States: outpatient clinic models and the residential-based Minnesota Model for treating alcoholism, therapeutic communities and methadone maintenance for the treatment of narcotics addiction, and outpatient drug-free counseling as a treatment for the growing problem of youthful polydrug abuse. What was needed was a shift in policy that would allow for the wide replication of these new treatments. In the 1950s, reports from joint committees of the American Medical Association (AMA) and the American Bar Association (ABA) recommended that responsibility for the country’s drug problem should be shifted from the criminal justice arena to the medical and public health venues. The AMA/ABA reports called for a renewed emphasis on treatment and medically directed maintenance experiments for narcotics addiction. The 1967 report of the Cooperative Commission on the Study of Alcoholism went even further in outlining the skeleton of what would become the federal-state-local partnership in the management and treatment of addiction (Plaut, 1967). This groundwork led to growing federal and state involvement in addiction treatment in the 1960s and the passage of legislation in the early 1970s that virtually spawned the modern field of addiction treatment. The creation of the National Institute on Alcohol Abuse and Alcoholism and the National Institute on Drug Abuse marked a major turning point in more than a century of efforts to address the problem of addiction. Between 1965 and 1975, the federal government, through new funding initiatives, established a national network of local community–based treatment programs. This shift reflected the movement toward more medicalized models of addressing narcotics addiction and was also fueled by the Nixon administration’s concern about the rise in drug-related crime. This change included some experiments with Centralized Intake Units (CIUs), a single point of entry into the treatment system (e.g., Special Action Office for Drug Abuse Prevention [SAODAP], 1974). President Nixon recruited Dr. Jerome Jaffe in 1971 to head SAODAP. Jaffe had previously been in charge of a pilot program in Chicago to respond to increasing numbers of heroin addicts. In an effort to streamline the service delivery system to deal with the increasing

6

Christy K Scott and Randolph D. Muck

problem, Jaffe had overseen the development of a single point of access model wherein an assessment (including a physical examination) was performed and clients were matched with appropriate treatment programs. Jaffe brought this model of CIUs with him to SAODAP. Shortly after Jaffe’s arrival, contracts were issued for cities to set up CIUs and systems similar to the model developed in Chicago (Scott, Muck, & Foss 2000; Massing, 1998). As a part of the federal war on drugs, treatment capacity was increased exponentially. By 1973 some cities had excess capacity. However, during the 1980s support for treatment services declined, and the focus of federal efforts shifted to law enforcement and interdiction. Drug enforcement budgets were increased by 20%, and treatment funding was reduced by 25% (Besteman, 1990). Although the 1970s saw an increase in funding for treatment, and the federal government was influencing the start-up of CIUs across the country, there was little data produced on how well these systems actually worked. CIUs, while gaining popularity, did not have scientific findings that supported their existence. In concert with the reduced federal funding, the voices that opposed CIUs, mainly local treatment providers, began to be heard. Any funding for a CIU would take away from the dollars available for treatment programs. CIUs, where they existed, held the decision-making authority regarding which programs clients were referred to. Consequently, by the late 1980s, CIUs were shuttered and closed in almost all jurisdictions. A large multimodality treatment system had emerged in the late 1960s and early 1970s, even as concerns were raised anew about problems of weak infrastructure, inconsistent quality, poor accessibility, and a lack of service coordination and continuity. These concerns were magnified as federal spending for drug treatment diminished between 1975 and 1986. Although some states attempted to supplant federal funding, the availability and quality of treatment varied widely across the country. The bottom line was a sharp decline in the stability of the community-based public tier of treatment, paralleled by a period of unprecedented growth for the role of the criminal justice system during the 1980s. This intensified criminalization of addiction in the public sector was paralleled by growth in the private addiction treatment sector. The private sector embraced the medical ideas about addiction, leaving few resources for those dependent on publicly funded substance abuse treatment. In 1990, the Institute of Medicine (IOM) published a report that summarized this period, noting that the public tier of drug treatment

Background and Overview

7

had been the neglected front in the drug wars of the 1980s. The report further highlighted the way in which the federal anti-drug abuse legislation of 1986 and 1988 directed funding toward enforcement against traffickers and prevention among nonusers. Publicly funded substance abuse treatment was largely ignored, with the exception of treatment related to stemming the growing epidemic of acquired immune deficiency syndrome (AIDS). There were a number of trends that emerged related to the state of addiction treatment in the 1980s that laid a foundation for the Target Cities Program. First, many treatment agencies were plagued by weak organizational infrastructure, by clinicians’ preferences for particular types of treatment or specific treatment agencies, by isolation from the larger network of health and human service providers, and by clinical practices that did not reflect the latest breakthroughs in clinical scientific research. Second, there was a growing recognition of subpopulations of participants and “special populations” as shifts occurred in primary drugs of choice (from heroin to crack cocaine) and as clinicians came to believe that treatment methods needed to be adapted to the user’s age, gender, culture, and developmental history. The drug-related spread of HIV and the growing numbers of persons with AIDS seeking treatment added further urgency to the need to address multiple and complex problems within addiction treatment institutions. The shift from providing a “program” for all participants to a greater emphasis on differential diagnosis and individualized treatment planning created enormous pressure to elevate the level of clinical practice in most treatment agencies. A third trend in the 1980s was the growing complexity of the clinical profiles presented by participants seeking treatment. Increasing numbers of those seeking addiction treatment presented with multiple problems of great chronicity and acuity and with complicated service histories in multiple systems. Shifting social policies brought ever-increasing numbers of substance abusers into both the criminal justice and child protection systems and generated an enormous flow of referrals to treatment. The recognition that these participants had long histories of exclusion, extrusion, premature service disengagement, and multiple service episodes across many community agencies generated incentives toward increased agency coordination in the short run and visions of integrated service systems. A fourth trend in the 1980s was that rewards were beginning to be reaped by the nation’s investment in addiction research. Research findings were beginning to underscore some basic principles of what

8

Christy K Scott and Randolph D. Muck

worked in treatment, and new treatments were emerging from pharmacological adjuncts to empirically based, manual-guided therapies. These advancements threatened to further widen the gap between the advances of addiction science and the state of clinical practice in mainstream addiction treatment in the country.

The Interface between the Institute of Medicine, the Office of Treatment Improvement, and Target Cities The Anti-Drug Abuse Act of 1986 called upon the Secretary of Health and Human Services to commission an independent study of substance abuse treatment. This landmark study conducted by the Institute of Medicine was published under the title Treating Drug Problems and led to significant changes in the focus of federal treatmentrelated activities. These changes included the establishment of the Office of Treatment Improvement (OTI), which later transitioned to the Center for Substance Abuse Treatment (CSAT). The public tier of addiction treatment began to expand following the passage of the Anti-Drug Abuse Act. Treatment expanded even more dramatically through resources provided under the 1988 AntiDrug Abuse Act and the emergency supplemental appropriation to the Alcohol, Drug Abuse, and Mental Health Services (ADMHS) block grant in 1989. The Office of National Drug Control Policy (ONDCP), which was legislatively authorized and established in March 1989, was assigned a leading role in national strategic planning for drug treatment. Approximately 6 months later, the Alcohol, Drug Abuse, and Mental Health Administration (ADAMHA) consolidated the block grant and many of the treatment demonstration authorities in the Office of Treatment Improvement. The general mission of the Office of Treatment Improvement was to improve the overall quality of drug abuse treatment nationwide. Coordination among local, state, and federal agencies was a major theme within the OTI treatment improvement strategy. Moreover, the IOM report, Treating Drug Problems (Gerstein & Harwood, 1990, p. 195) clearly recommended that the “National Institute on Drug Abuse, in conjunction with its sister agency, the Office of Treatment Improvements [sic], needs to give more adequate, focused attention to the drug treatment delivery system as a whole.” To that end, part of the rationale underlying the OTI philosophy for improving treatment was outlined in the Target Cities Demonstration Program Request for Applications. This document communicated the need for a comprehensive

Background and Overview

9

service system, including centralized intake, in which participants were objectively matched to the most cost-effective treatment, case management, coordination, and outcome monitoring. The model or philosophy outlined in the Target Cities Program was consistent with concurrent work by the Institute of Medicine (Gerstein & Harwood, 1990) that recommended a comprehensive self-correcting system of care that included assessment, matching, outcome determination, feedback, continuity assurance, and clinician training.

Overview of Target Cities Through the Target Cities demonstration project, a total of 19 cities were funded in two 5-year waves. The first wave of cities included Albuquerque, Atlanta, Baltimore, Boston, Los Angeles, Milwaukee, New York, and San Juan, with funding beginning in 1990. CSAT funded Philadelphia between the first and second rounds of award, and in 1993, CSAT funded the second generation of cities, which included Chicago, Cleveland, Dallas, Detroit, Miami, New Orleans, Newark, Portland, St. Louis, and San Francisco. Program Description Program requirements for the first and second generation of cities were similar, yet differed in important ways. CSAT required both generations to target service improvements for at least one of the following: adolescents, minorities, pregnant women, female addicts and their children, and residents of public housing. In 1993, CSAT refined the eligibility requirements to include persons who resided in the target jurisdiction, suffered from alcohol and drug problems, were unemployed or underemployed, and required publicly subsidized treatment. Both waves were mandated to propose models to (a) improve coordination among local drug abuse, health, mental health, education, law enforcement, judicial, correctional, and human service agencies; (b) establish or enhance central intake and referral facilities, (c) develop an automated patient tracking and referral system; and (d) implement measures to ensure the quality of services provided. Discussions between the federal, state, and local project staffs who were involved with the first generation of Target Cities resulted in modifications to the requirements for the second wave of cities and shifts in the guidance that CSAT staff provided to program staff from those cities. During these discussions, staff identified several types of

10

Christy K Scott and Randolph D. Muck

problems specific to implementing a system change of the Target Cities Program’s scope and magnitude. For example, program staff reported that in changing systems as large as those operating in most cities, the number of unforeseen variables and the vast differences in types of programs had to be considered simultaneously. The time required to effect change involving these entities was often much longer than the planning time allowed by CSAT. In addition to a need for sufficient time to implement such a complex program, substantive differences existed in the ways in which treatment delivery systems operated prior to the Target Cities Program. Such vast differences limited the degree to which each city could implement a national model with standardized components (personal communication, Office of Treatment Improvement, Target Cities meeting, July 23, 1992). Modifications that worked in one area of the country would not necessarily be successful in another part of the country. Another challenge in incorporating the lessons learned between the first and second wave of Target Cities was striking a balance between standardization and adequate flexibility to allow for local site differences. For example, the first-wave cities were required to establish or enhance central intake but were not provided with an operational definition of central intake. In the end, while both generations of cities were required to establish or enhance central intake and referral facilities, the goals differed between the two waves. Therefore, one of the overriding differences between the first and second generations of Target Cities was the level of detail CSAT provided regarding the goals, interventions, and requirements. For the second generation of cities, CSAT was considerably more prescriptive, mandating specific activities and more clearly articulating the goals that produced important differences in the nature of the projects between generations. For example, among the first wave of Target Cities projects, central intake was not intended to “replace the existing outreach, case finding, and intake procedures of local treatment programs, but a local program could contract with the Central Intake Units (CIUs) to provide the program with assessment and intake services Employee assistance programs could also contract with the CIUs” (see the RFA in Appendix B in DHHS, 1990). In contrast, second-wave cities were clearly required to provide a comprehensive assessment to all participants. Given that this is often a time-consuming task requiring specialized staff capabilities not readily available in most publicly funded community-based treatment programs, it was assumed that centralizing the assessment function at one or more

Background and Overview

11

sites, referred to as Service Delivery Units (SDUs), would obviate the need to invest resources to create these capabilities in each and every treatment provider location (DHHS, 1993, pp. 14–15). In addition to more clearly defining the role for central intake, CSAT staff identified several goals for the project, and many of the activities deemed as “optional” in the first-wave cities were required activities for the second generation. Below is a list of the goals for the project, followed by a list of required activities for the second generation of Target Cities. Program Goals 1. Increase access to treatment for those in need of treatment. 2. Increase the effectiveness of addiction treatment and recovery services in large metropolitan areas (i.e., to improve treatment outcomes for individuals with alcohol and drug problems and their families). 3. Foster coordination among addiction treatment and recovery programs and related health (e.g., TB/HIV/STDs), housing, welfare, job training, education, community redevelopment, social programs and institutions, and the legal system (e.g., police, courts, jails) as a means of involving alcohol- and drug-involved individuals in treatment and achieving improved treatment outcomes. 4. Develop methods by which metropolitan systems of care can continually improve treatment effectiveness. Required Activities CSAT did not require that all components of the proposed system be new. Specifically, the service system components already existing in an applicant metropolitan area were to be combined with the proposed system enhancements so that the system had the following capability (DHHS, 1993, p. 12): 1. To conduct an assessment of treatment staff training requirements, followed by the design and implementation of continuing professional education and other staff training programs and the evaluation of these activities.

12

Christy K Scott and Randolph D. Muck 2. To design and implement one or more central intake, assessment, and referral facilities wherein (a) a standardized, comprehensive intake assessment process is utilized to include a physical, screening for HIV, TB, STDs, and other infectious disease; alcohol and drug use history; psychosocial evaluation and, where warranted, a psychiatric evaluation; (b) standardized protocols are used for matching individuals with a continuum of appropriate treatment, recovery, and support services; (c) a case management system is implemented that is capable of tracking individuals across SDUs. 3. To establish a process whereby individual economic and social welfare needs are thoroughly assessed and addressed, including a determination of eligibility and subsequent registration for AFDC, food stamps, SSI, and so forth. 4. To establish linkages and formal referral processes whereby the ongoing preventive and primary health care needs can be met. 5. To case manage criminal justice participants through the various stages of treatment and legal case processing. 6. To develop a management information system capable of capturing current program characteristics, intake, assessment, referral and outcome data, financial data including charges and costs of treatment, and capacity utilization for every participant. 7. To incorporate an evaluation or management unit capable of utilizing the MIS to determine how successfully referrals were made, the length of stay, and which patients benefited from which programs. 8. To integrate a quality-assurance mechanism and provision of targeted technical and financial assistance designed to ensure that the quality of service delivery in participating SDUs is continually enhanced.

Evolution of the Multisite Evaluation Local program evaluation was considered an integral part of the Target Cities demonstration project. The combined interest on the

Background and Overview

13

part of both CSAT and local evaluators in each city led to a multisite evaluation. This effort began in February 1995 with a small group of evaluators meeting with CSAT staff to determine the feasibility of such an effort. At this point, data collection had begun at several sites, causing flexibility in instrumentation at some sites to be limited. Moreover, because cities could rely on a combination of already existing system components as well as new ones, variation in implementation across cities differed greatly. For example, Centralized Intake Units already existed in Detroit, whereas Chicago opened new ones and required participants to access treatment through them. In Cleveland, new CIUs were opened, but participants could still access treatment directly from a treatment program. In spite of the variation across sites, evaluation designs, and instrumentation (described in more detail in chapter 2), the evaluators and CSAT staff determined that enough similarity existed across sites to develop several multisite databases that would capture important information about the demonstration project. During the next 5 years, the evaluators and CSAT staff met several times each year. During the initial stages of the collaboration, staff from the University of Akron completed a crosswalk to help identify the common data elements across sites and operated as a depository and distributor of the data on an ongoing basis (a more detailed description of the process is included in chapter 2). Given the magnitude and complexity of the system changes resulting from the Target Cities Program, not every aspect of the change could be evaluated. Moreover, the variation and financial constraints across sites in many ways drove the focus of the local evaluations. For example, while case management became the focus in Detroit, Portland focused on increasing access to treatment for the criminal justice population. The chapters in this book are based on data collected during implementation of the second generation of Target Cities. They are not intended to serve as a comprehensive compilation of all of the outcome data from the programs. Various topics are covered in other publications, including the implementation process at many sites, described by Guydish and Muck (1999a) in a special issue of the Journal of Psychoactive Drugs, the Management Information System (Hile, 1998), and other measures of treatment access and participant satisfaction (Scott, Muck, & Foss, 2000). Many research questions remain that can be explored using the Target Cities data sets. It is important to note, as will be discussed in chapter 2, that the Target Cities data set is quite large and complex.

14

Christy K Scott and Randolph D. Muck

Attempts to use these data in isolation from a full explication of their construction and the understanding of the complexity and diversity of the programs will likely lead to specious conclusions.

Book Overview The chapters in this book move between full multisite data comparisons, outcomes from subsets of sites, and single site outcomes. Due to the nature of initial development of the multisite data sets, relying on a combination of data sources provided a good method for addressing important issues related to the goals and objectives of the Target Cities Program. This also speaks to the complexity of the program. Within each Target Cities project, implementation was shaped by parallel forces including, but not limited to, the federal requirements, political agendas, and the ecological realities operating at each site. Although considerably more statistical power would have resulted from uniform implementation across sites, this requirement would have represented an inappropriate “cookie-cutter” approach to a highly complex project that now has the ability to inform many jurisdictions on a multiplicity of differing approaches to systems change. It should not be surprising that much more detail can be provided from single sites concerning particular issues that were the focus of their local evaluations. However, each of these levels provides a perspective on this program that is important for understanding the implementation and outcomes and for gleaning information that will be useful for others planning modifications to large service delivery systems. In chapter 2, following this present introductory chapter, Leahy, Stephens, Huff, and Kaye provide a description of the methodologies used across the sites and the procedures used to develop the multisite databases. In chapter 3, Claus and Dailey describe the CIU participant population as a whole as well as for each site. This chapter clearly illustrates the richness of the data and provides a snapshot of the more than 40,000 participants represented in the databases. As recommended by the 1990 Institute of Medicine report and called for by CSAT, the sample, and therefore the intervention, included a large number of pregnant women and women with children, as well as other subpopulations. It is important to note that this multisite sample is not representative of any participating site; it is, however, one of the largest samples of substance-abusing persons presenting for treatment assessment nationally.

Background and Overview

15

In chapter 4, using cluster analysis for a random sample of the participants, Foss, Barron, and Arfken identify seven ways in which problems and patterns of service needs differed by site. The approach used in this chapter can be a critical component of a comprehensive needs assessment. The resulting need profiles allow for a clearer projection of the ways in which the type, environment, and modality of substance abuse services and linkages may be configured for a system most responsive to participants with multiple needs. Issues pertaining to improved access to treatment, one of the major goals of the Target Cities Program, are reviewed in chapter 5. Claus, Barron, and Pascual present results based on data from one of the multisite databases as well as from single site evaluations. One of the issues raised during the planning phase of the project was whether or not adding the additional step of centralized intake would negatively impact participants’ access to treatment. The authors concluded that despite the magnitude of the system change (including the added step of centralized intake, often located at a different location from treatment), the interventions increased access to treatment for underserved populations, decreased time to treatment entry, and maintained participant satisfaction with the intake process. In chapter 6, Scott, Foss, and Sherman address a related issue by looking at a treatment satisfaction survey that was administered during a 6-month postintake follow-up interview. Levels of treatment satisfaction reported by participants who entered treatment prior to the opening of the Central Intake Unit were compared with levels of satisfaction reported by participants who accessed treatment through the CIU. The outcome of the comparison shows that centralizing intake did not negatively impact participants’ perceptions of treatment services. In chapter 7, Arfken, Klein, Agius, and diMenza explore the degree to which the critical interventions of matching, case management, and linkages were successfully implemented at the program sites. The achievements and downfalls of these implementations are analyzed utilizing a policy analysis framework adapted from Sabatier and Mazmanian (1979). This in-depth analysis provides a context in which to evaluate some of the results reported in other chapters and serves to inform administrators considering similar complex system changes. The remaining chapters focus on participant outcomes using either multisite or single site databases. In chapter 8 Guydish et al. compare treatment outcomes for participants who accessed treatment before centralized intake with treatment outcomes for those who accessed it through centralized intake. Data from three cities—

16

Christy K Scott and Randolph D. Muck

Portland, San Francisco, and Chicago—contributed to these findings. The data compared in this chapter were collected during interviews conducted at intake and at either 6 months (Chicago) or 12 months (Portland and San Francisco) later. Based on ASI composite scores, participants in the CIU cohort did not demonstrate significantly better outcomes than participants who accessed treatment directly from treatment programs. It is also shown that outcomes differed by city. Chapter 9 (Scott, Foss, & Sherman) includes a more detailed analysis of participant outcomes in Chicago. Centralized intake was established at two separate locations as part of the Target Cities project in Chicago. Outcomes for participants in the pre-CIU cohort were compared with participant outcomes in the CIU cohort. Participants in the CIU cohort demonstrated lower rates of drug use and better employment outcomes than participants in the pre-CIU cohort. Chapter 10 (Finigan, Barron, & Carey) focuses on the development of a pretreatment In-Jail Intervention Program (IJIP) for substance-abusing criminal justice clients in the Portland Target Cities Project. The analysis focused on arrest and days of incarceration after participants completed the program. Results indicated a reduced number of subsequent rearrests. The final chapter (Guydish, Stephens, & Muck) provides a synthesis of the findings, explores lessons learned, and discusses policy implications that may be drawn from the experience gained during evolution of this program. Prior to exploring the local and multisite outcomes in the Target Cities Program, it is critical to understand the breadth of the data collection efforts and the constraints in utilizing the data. The following chapter, “Methodological Issues in the Development of the Target Cities Multisite Databases,” provides the detailed description necessary to this understanding and subsequent interpretation of the results of the several studies conducted within the program.

CHAPTER

2

Methodological Issues in the Development of the Target Cities Multisite Databases Peter J. Leahy, Richard C. Stephens, Heather K. Huff, and Russell S. Kaye

Unlike small, localized data collection efforts of the past, the Target Cities Program offered researchers an unprecedented opportunity. Through substantial federal funding over a period of 5 years, in 10 major U.S. cities, CSAT was poised to collect demographic, treatment, and outcome data from thousands of participants in substanceabuse assessment and treatment programs. While not the original design of the Target Cities Program, local evaluators and CSAT administrators quickly recognized the power and potential inherent in creating a database across all Target Cities sites. Such a database would not only yield information regarding the results of the demonstration program but also would provide a rich source of data for future study. The decision to create the database was made in the second year of operations of the second-wave Target Cities sites. Data collection was already well underway in many of the cities, obviating the opportunity to collect standardized data from the beginning of the project. A sequence of steps was undertaken to create a series of databases for which the sites could provide data: 1. All sites sent copies of their data collection instruments to the University of Akron, which became the central coordinating center and data repository for the multisite databases.

17

18

Peter J. Leahy et al. 2. Staff at the University of Akron undertook an inductive analysis of all research instruments and enumerated those variables which appeared to be common across many, if not all, of the nine sites. 3. A multisite planning team (composed of evaluators and, when appropriate, MIS directors, project managers, principal investigators, and CSAT personnel) met on a regular basis to agree on which variables should be included in the multisite databases and how each of these variables was to be operationalized. 4. Each site had to make its own decision about whether a variable could be collected locally. Not all sites could contribute data to the multisite effort for all variables. In determining whether a variable should be included, the primary decision rule was whether the data element was substantively close enough in meaning to the multisite level variable to be included. Often the local data on a variable were submitted to the multisite database, but the site also included an explanation of that variable. All of these comments were appended to the databases as annotations (see University of Akron, 1998, 1999a, 1999b). Potential users of these databases are strongly advised to carefully consult the annotations when conducting analyses. 5. After data were submitted, the University of Akron conducted internal validity and other quality checks. Multisite data were then sent to all participating sites, and it is these data that form the basis for many of the analyses presented here.

This chapter details the development of the multisite databases that were used in many of the analyses to be presented in the following chapters. Ten of the second-wave Target Cities sites (Chicago, Cleveland, Dallas, Detroit, Miami, Newark, New Orleans, Portland, St. Louis, and San Francisco) agreed to collaborate in this effort. It is important to recognize that the development of data sets and the requisite Management Information Systems needed to capture these data were a crucial but difficult part of the implementation of the multisite effort. For example, the Newark site was able to enhance services and develop a new model of collaboration and cooperation; however, staff were unable to implement a viable MIS until after data collection had

Methodological Issues

19

ceased for the multisite project. For that reason Newark is not discussed in the site-by-site review that follows. The Miami site also dropped out early in the process. To help readers understand the complexity involved in creating this multisite database and to highlight the evaluation design variability that existed across these sites, it is important to briefly review each site’s evaluation research design and implementation experience. This review is also included to assist researchers in understanding the strengths and limitations of the Target Cities Multisite Databases in future studies. Table 2-1 provides a summary of the evaluation process at each multisite Target Cities project. Included in Table 2-1 are the evaluation research designs, sample sizes of CIU and comparison groups (if any), number of waves of data collection, baseline data collection timetables, and follow-up completion rates.

Chicago A nonequivalent control group design (Campbell & Stanley, 1963) involving two sequential cohorts was used for evaluation. The earlier comparison cohort consisted of individuals seeking treatment directly at the participating treatment agencies prior to the implementation of a centralized intake unit (CIU). The later cohort consisted of individuals seeking substance abuse treatment directly through the CIU. The sites consisted of substance abuse treatment agencies that agreed to turn over their program assessment functions to a CIU for assessment and referral. One site was located on the north side of Chicago and another on the west side. Participants in each cohort were followed up 6 months after the original interview at each site (Scott, Foss, Dennis, & Godley, 2000). In both sites, participants were interviewed at intake and again 6 months later for the outcome evaluation study. Participants can be considered an “intent to treat” sample except for the pre-CIU cohort on the west side that had to appear for treatment at least once. Approximately half the participants in each network were sampled before the implementation of the CIU (i.e., were a pre-CIU cohort) and the other half after the CIU opened (i.e., were a CIU cohort). The levels of care represented in Chicago included outpatient drug-free, methadone, intensive outpatient, halfway house, short-term residential, and long-term residential. Recruitment continued until approximately 100 cases per level of care were interviewed. The desired

20

Peter J. Leahy et al.

number of participants was not reached in every case (see Scott et al., 2000 for details). Participants were volunteer adult Chicago residents seeking publicly funded services who had used alcohol or drugs in the past 6 months. Individuals who had spent the last 6 months in a controlled environment were not recruited if their abstinence was considered involuntary. Baseline data collection for the north side comparison (pre-CIU) group in Chicago began in March 1995 and concluded in December 1995. The CIU group baseline collection began in August 1996 and concluded in June 1998. For the west side baseline data collection began in July 1996 and concluded in July 1997 for the comparison group (pre-CIU). For the CIU group data were collected from August 1997 through June 1998. As Table 2-1 reveals, Chicago had an excellent follow-up record, achieving completion rates ranging from 87.7% to 98.9%. An Augmented Addiction Severity Index (A-ASI) (Scott, Dennis, Godley, & Foss, 1995) was the primary data collection instrument used at baseline and follow-up to measure outcomes. The ASI, fifth edition (McLellan et al., 1992) is a semistructured interview designed to measure problem severity in seven areas: medical, legal, alcohol abuse, drug abuse, employment, family, and psychiatric functioning. The psychometrics have internal consistency ranging from .69 to .93 and testretest reliabilities ranging from .51 to .92 (Kosten, Rounsaville, & Kleber, 1985; McLellan, Luborsky, Woody, Druley, & O’Brien, 1983). Chicago utilized several other measures in their evaluation (see Scott et al., 2000). Of interest to readers of this volume were the Treatment Satisfaction Questionnaire (TSQ) and the 8-item Client Satisfaction Questionnaire (CSQ-8) (Nguyen, Attkisson, & Stegner, 1983). Evaluation staff trained interviewers and CIU assessment staff on the ASI. Evaluation staff conducted the intake assessments for the pre-CIU, cohort and CIU staff conducted intake assessments for the CIU cohorts. The assessment at the CIU took at least 2 hours. All follow-up interviews were done by evaluation staff and took approximately 120 minutes.

Cleveland Cleveland used a concurrent nonequivalent comparison group design. Evaluation group participants came from two CIUs in Cleveland, one located on the east side and the other on the west side. To be included in the CIU group, a participant had to be a first-time visitor to the CIU who was assessed with a drug or alcohol problem and

Methodological Issues

21

referred to a treatment provider. Comparison group participants were assessed with a drug or alcohol problem at one of 17 (of 50 total) service delivery units (“SDUs,” i.e., treatment agencies) and referred to treatment. The participating agencies were part of the publicly funded Alcohol and Drug Addiction Services Board of Cuyahoga County (ADASBCC) treatment system and were located throughout the city of Cleveland. Approximately 75% of the alcohol and drug treatment participants seen in Cleveland entered through those agencies. One agency discontinued operations during the course of the study (Stephens, Kaye, & Chen, 1999). Two additional evaluation subgroups were defined in Cleveland: treatment and dropout. Treatment participants in either evaluation condition appeared for treatment at an ADASBCC service agency after referral, and remained in treatment for 14 or more days. Dropout participants either failed to appear at a treatment program to which they had been referred or stopped attending treatment within the first 13 days. In the Cleveland evaluation study, treatment excluded detoxification services. All participants were adult volunteers. The original design called for selecting 300 cases in each of four quasi-experimental conditions: CIU treatment, CIU dropout, SDU treatment, and SDU dropout. When fewer participants appeared at the two CIUs than projected, the smaller number of CIU participants in a given month set the ceiling for the random selection of comparison cases from the 17 SDUs using the ADASBCC management information system. As a result, sample sizes for the four conditions were as follows: CIU treatment (269), CIU dropout (402), SDU treatment (378), and SDU dropout (216). All groups were interviewed at baseline and again at 6 and 12 months postbaseline using the Comprehensive Intake Assessment Instrument–Cleveland (CIAI-C) (University of Akron, 1997). The CIAI-C is a modified version of the CIAI first developed by CSAT. It is a comprehensive assessment of drug and alcohol use, conducted by an interviewer on a laptop computer, and includes extensive self-report data on criminal history, health and mental health, demographics, housing, family relationships, treatment, social service use, and employment. It also included DSM-IV diagnostic assessment in the baseline version. Akron worked closely with a group of clinicians from the ADASBCC system to refine the instrument for use in Cleveland. Cleveland was the only multisite program to use this instrument for evaluation. The success of the CIAI-C as an assessment tool in Cleveland has led to subsequent adoption by the entire ADASBCC system. Table 2-1 reveals that the follow-up rates ranged from 68.9% to 75.8% at 6 months, and from 69.1% to 76.2% at 12 months.

22

Peter J. Leahy et al. Dallas

Dallas used a sequential nonequivalent comparison group design with a CIU evaluation group and a comparison group that had a less intensive CIU experience. Both groups were followed up 6 months after baseline assessment. The CIU group included participants who were assessed and referred to treatment in Dallas under the full CIU experience. Participants receiving the full CIU experience had significant access to medical and mental health services. By November 1997, budget cuts resulted in less agency staff time for assessment and a reduction in the availability of medical and mental health services. Participants were assessed after this time with a shorter assessment instrument and became the comparison group participants under what could be termed a partial, or “CIU lite,” experience (Krepcho, Snell, Coletta, & Olsen, 1999). Both CIU and comparison group participants were randomly selected from adults who were assessed and referred to treatment. It is important to note that in Dallas a large number of the Target Cities participants were referred solely to detoxification. Although CSAT did not consider detox as a treatment modality for the Target Cities Program, nonmedical detoxification was a major treatment referral made for many participants in the Dallas Target Cities project. For this reason the data sets contributed by Dallas to the multisite effort exclude large numbers of Target Cities participants. Participants were assessed at baseline using a computerized version of the ASI. Baseline data collection began for the CIU group in November 1996 and concluded in February 1997. Comparison group baseline data were collected over the period from November 1997 through February 1998. Table 2-1 reveals that Dallas had a 6-month completion rate of 60.9% for the 221 participants in the CIU condition, and a rate of 53.4% for the 209 comparison participants. Follow-up participants were assessed with the same computerized Addiction Severity Index.

Detroit Detroit was the only city with a functioning CIU prior to the start of Target Cities. Assessment by staff at the CIU had been required prior to Target Cities for all adults seeking detoxification, outpatient, intensive outpatient, or residential substance abuse treatment services in the publicly funded system in Detroit. Persons with other

Methodological Issues

23

sources of payment could also be assessed at the CIU, but it was not mandatory. Mandatory assessment for Medicaid recipients at the CIU was not required until 1999. Persons whose treatment was paid for with mental health funds also were not assessed at the CIU, nor were people of Hispanic or Native American heritage (Zold-Kilbourn, Tucker, & Berry, 1999). The Detroit evaluation study initially used a two-group nonequivalent comparison group design to evaluate a new wraparound case management evaluation. In addition, the Detroit evaluation examined changes following the initiation of MIS and enhanced CIU case management using a pretest/posttest evaluation design. For this latter analysis, the evaluation group consisted of all persons assessed at the CIU during the baseline data collection period of February 1996 through December 1997. The ASI was used at the CIU for measurement. Users of the Detroit multisite data need to be aware that frequent changes in management, evaluation, and MIS staff during the project restricted Detroit’s ability to contribute to several of the multisite databases. While data for Detroit are included, their reliability should be viewed cautiously. For that reason Detroit’s data are not included in multisite databases 2 and 3 (discussed later in this chapter). Researchers interested in studying the Detroit Target Cities experience should consider using the Detroit data separately and can contact the Detroit Target Cities project for appropriate limitations.

New Orleans New Orleans planned a one-group pretest/posttest design with a single follow-up interview of Target Cities Program participants conducted 12 months after discharge from treatment. The comparison group in New Orleans consisted of participants who bypassed the central intake process. These participants self-selected treatment providers and received their intake assessment at the SDUs. Sample selection was based on a disproportionate stratified selection process designed to ensure sufficient cases to generalize to both small and large agencies. New Orleans utilized a battery of instruments developed or modified within the Target Cities Program. The New Orleans Target Cities Intake Assessment (NIA), a modified version of the Global Appraisal of Individual Needs (Dennis, Rourke, Lennox, Campbell, & Caddell, 1995; Dennis, 1999) was used at baseline. The New Orleans Target Cities

24

Peter J. Leahy et al.

Quarterly Assessment (NQA) was used for follow-up (Becnel et al., 1999). This is an abbreviated version of the NIA, without questions on “lifetime” drug use, criminal justice, and physical and mental health. The NIA and NQA primarily included the Michigan Alcoholism Screening Test (MAST) (Zung, 1979) and the Drug Abuse Screening Test (DAST) (Skinner, 1982), and also included DSM-IV diagnostic criteria. Clinical staff at Target Cities sites conducted baseline data collection on adult participants between February 1996 and April 1998. Follow-up interviews were conducted by telephone one year after discharge from treatment on a random sample of 797 of the original 1,659 adult Target Cities participants. Ninety-five percent were interviewed at the follow-up (McDermeit & Dennis, 2000). Of the 754 follow-up interviews conducted, 707 matched appropriately with baseline interviews.

Portland Portland used a nonequivalent comparison group evaluation design with the CIU and comparison groups followed up at 6- and 12month postbaseline assessment. The CIU group comprised 344 adults who appeared for alcohol or drug assessment at any of three CIUs during the period of March 1997 through July 1997. CIU1 was a major outpatient assessment center that focused on DUI and general participant assessment, and other locations including probation and parole offices and health clinics. CIU2 was located in an in-jail (county) intervention program and included assessment and pretreatment services (described in Finigan, Barron, & Carey, chapter 10 in the present volume). CIU3, a small operation, was an administrative designation. (Assessments at CIU3 were done by county CIU staff for in-jail inmates on regular wards, and these participants received no additional in-jail services.) The two comparison subgroups were collected over two different periods of time. The first comparison subgroup included 406 adults who presented for treatment at an SDU or the preexisting county DUI intake unit during the pre–Target Cities period of November 1994 through April 1995. The other comparison subgroup comprised 88 participants who presented for assessment at an SDU during a time interval roughly concurrent with the CIU evaluation group (March 1997 through July 1997). The CIU and comparison groups were measured at baseline and follow-up intervals using the Addiction Severity Index (ASI) that was

Methodological Issues

25

included within Portland’s local assessment instrument, the Multnomah Clinical Assessment (MCA) (Barron et al., 1999). For both cohorts, baseline interviews were conducted by county employees trained as interviewers, and follow-up interviews were conducted by trained interviewers from a contract research firm. Table 2-1 shows that the follow-up completion rate for the earlier comparison subgroup was 67.9% at 6 months and 72.2% at 12 months. For the later comparison subgroup, the 6-month completion rate was 70.5% and the 12-month rate was 75.9%. The completion rate for the CIU group was 73.4% at 6 months and 72.9% at 12 months.

St. Louis St. Louis used a nonequivalent comparison group design with follow-up waves at 3-, 6- and 12-months postbaseline assessment for treatment. Two sequential cohorts of participants, all recruited from CIU participants seeking assessment and referral to a substance abuse treatment agency, were used in the study. In the first cohort, participants were randomly assigned to the full CIU experience (evaluation condition) or to one of two “CIU-lite” comparison conditions. In the second cohort all participants were assigned to the full CIU experience (evaluation) condition. In both cohorts the full CIU experience consisted of a complete baseline assessment with a clinician’s report accompanying the participant to a treatment agency. Participants in the full CIU experience condition also benefited from use of a computerized algorithm to better match an assessment to a referral. One group of comparison participants in cohort 1 assessed at the CIU had no matching algorithm, and these participants were referred to a treatment agency of their choice at a level of care compatible with their assessed need. The CIU assessment also accompanied them to the treatment agency. The other group of comparison participants in cohort 1 also had no matching algorithm. Although they, too, were assessed at the CIU, the clinical report was not transmitted to the CIU; each treatment agency carried out its standard intake assessment. The latter condition was introduced to approximate “treatment as usual.” Participants eligible for inclusion in the evaluation study were those over 18 years of age who were diagnosed with active substance abuse or dependency by a CIU staff psychologist and who were present at the CIU during either time period when the evaluators were on site.

26

Peter J. Leahy et al.

CIU staff conducted baseline interviews, and project staff of the Missouri Institute of Mental Health conducted follow-up interviews. Interviews were conducted by phone, in the field, and in the office. Baseline data were collected with the Initial Standardized Assessment Protocol (ISAP) (Hile, Callier, Schmoock, Adkins, & Cho, 1998). The ASI was included as part of the ISAP and was repeated at all follow-up waves. Baseline data collection for the full CIU experience cohort 1 participants began in September 1996 and concluded November 1996. Baseline data collection for both “CIU lite” groups in cohort 1 began in September 1996 and ended no later than January 1997. Baseline data collection for the full CIU experience group in cohort 2 began in June 1997 and concluded in September 1997.

San Francisco San Francisco used a nonequivalent comparison group design with two follow-up waves. The first follow-up wave occurred approximately 1 month after treatment entry and the second at 12-month posttreatment entry. CIU (evaluation) participants were adults who reported to a treatment program with a CIU referral. Comparison group participants were adults who reported to a treatment program without a CIU referral. Seven larger treatment programs (of 30 within the community) were selected to identify and recruit evaluation study participants. These seven agencies were believed to have received higher numbers of referrals from the CIU during the evaluation recruitment period. The programs were also selected to represent residential, day treatment, and outpatient modalities. Two adult gender-specific programs for females were selected to ensure adequate representation of women. Efforts were made within each group to recruit participants in roughly equal proportions between residential, day, and outpatient treatment modalities (Guydish, Woods, Davis & Bostrom, 2000). The CIU and non-CIU groups were recruited during overlapping time frames. The baseline data collection period for the comparison non-CIU group was March 1995 through July 1996. Baseline data for the evaluation CIU group occurred over the period May 1995 through February 1997. Research staff coordinated with CIU staff weekly to identify CIU participants who entered one of the seven treatment agencies. Research staff also coordinated with intake workers at the seven agencies to identify and recruit new admissions not referred by the CIU. Re-

Methodological Issues

27

search staff attempted to complete all baseline interviews within 2 weeks of admission. The ASI was used for all baseline and follow-up data collection. Other instruments were also used, including the Beck Depression Inventory, Basic Symptom Inventory subset of the Global Severity Index (SCL-90), and a homegrown measure of social support. These are not included in the multisite databases, however. Table 2-1 shows that the completion rates for the evaluation group were 92.0% at the 1-month follow-up and 87.0% at 12 months. The analogous rates for the comparison group were 87.0% and 80.0%, respectively.

The Target Cities Multisite Databases Table 2-2 represents the three multisite databases developed from the Target Cities evaluation effort. Table 2-2 identifies the sites that contributed data to a particular database and shows the number of records contributed by each site. Database 1 This database contains the multisite sample of all CIU baseline assessments collected over the period November 1994 through December 1997. Baseline data continued to be collected on Target Cities participants at some sites after December 1997; to maintain consistency, however, a decision was made to stop the entry of participants into Database 1 after December 1997. Database 1 reflects all persons who underwent a “CIU experience” (as individually defined by each of the eight Target Cities multisites) who were assessed during this period. Participants who received multiple CIU assessments appear in Database 1 as often as they received a CIU assessment. There are 259 comparable multisite variables in the data set. Variables include participant demographics, extent of alcohol and other drug use at assessment, primary drug of use, self-reported treatment experiences and measurement of self-reported criminal involvement, living arrangements and family relationships, mental health, employment, and income, among others. A total of 43,624 records are included in the file from the eight sites. As Table 2-1 shows, there is significant variation in the extent to which each site could contribute to Database 1. St. Louis and Detroit, for example, together contributed just over half of all cases. Portland contributed the fewest cases. Nevertheless, because of the richness of data collected and its

28

Table 2-1 Evaluation Design Characteristics of Target City Participants in the MultiSite Databases

City/Evaluation Condition

Sample Size at Baseline

Baseline Collection Period

Number of Follow-up Waves

Completion Rates at Follow-up Interval 3 Months

6 Months

12 Months

CHICAGO Site A Evaluation: CIU Site A Comparison: pre-CIU Site B Evaluation: CIU Site B Comparison: pre-CIU

(2862) 692 842 753 575

8/96–6/98 3/95–12/95 8/97–6/98 7/96–7/97

1 1 1 1

– – – –

– – – –

93.3 87.7 96.5 98.9

– – – –

CLEVELAND CIU Evaluation CIU Dropout Comparison Evaluation Comparison Dropout

(1265) 269 402 378 216

10/96–10/98 10/96–10/98 10/96–10/98 10/96–10/98

2 2 2 2

– – – –

– – – –

75.8 73.2 68.9 71.5

76.2 75.5 69.1 73.3

DALLAS CIU Evaluation Partial CIU Comparison

(430) 221 209

11/96–2/97 11/97–2/98

1 1

– –

– –

60.9 53.4

– –

DETROIT CIU Evaluation

(2211) 2211

10/96–9/97

0

–

–

–

–

Peter J. Leahy et al.

1 Month

NEWARK CIU Evaluation SDU Comparison

– –

– –

– –

– –

(797) 707

2/96–4/98

1

–

–

–

95.0

PORTLAND CIU Evaluation Comparison: pre-CIU Group Comparison: SDU Group

(838) 344 406 88

3/97–7/97 11/94–5/95 3/97–7/97

2 2 2

– – –

– – –

73.4 70.5 67.9

72.9 75.9 72.2

ST. LOUIS Evaluation Cohort 1: CIU Evaluation Cohort 2: CIU Comparison Cohort 1: partial CIU Comparison Cohort 1: partial CIU

(310) 38 197 38 37

9/96–11/96 6/97–9/97 9/96–1/97 9/96–1/97

3 3 3 3

– – – –

76.3 74.1 81.6 73.0

60.5 79.2 68.4 64.9

73.7 81.7 73.7 81.1

SAN FRANCISCO CIU Evaluation SDU Comparison

(470) 212 258

5/95–2/97 3/95–7/96

2 2

92.0 87.0

– –

– –

87.0 80.0

Methodological Issues

NEW ORLEANS CIU Evaluation

29

30

Peter J. Leahy et al.

large size, Database 1 represents one of the most significant sources of information ever collected on such a large number of substance abusers. Since each site was required by CSAT to include an outcome evaluation component, it was logical for the Multisite Planning Team to develop outcome evaluation databases as part of their effort. Database 1e (where the “e” refers to the “evaluation group database”) contains baseline data on 259 variables for all CIU participants included as the evaluation condition in any of the participating cities (refer to Table 2-1). Database 1c (where the “c” refers to the comparison group database) contains similar, or in some sites identical, baseline data for all participants included in the comparison group conditions. The variables contained in these databases mirror the variables listed for Database 1. Since evaluation follow-up data were collected in several cities after December 1997, the cutoff date for including data in Database 1, not all of the cases in Databases 1e or 1c will also be found in Database 1. Chicago, Cleveland, and New Orleans, for example, have cases contained in Database 1e that are not in Database 1. Cleveland, Dallas, and New Orleans have cases contained in Database 1c that are not in Database 1. Taken together with Databases 3e and 3c (discussed below), the databases provide an extremely rich and unusually large sample of data on outcomes for substance abusers. The difficulty of developing a set of comparable multisite measures from evaluation studies already underway has been noted. Table 2-1 and the ensuing discussion is intended to give the reader insight into the design variation existing across these sites. The reader is once again strongly advised to consult the annotated codebooks for Databases 1, 1e, and 1c (and the other databases discussed below) prior to working with these data (University of Akron, 1999a). This is important, since each site produced annotations for any variable whose coding departed from the multisite codebook. When a local code deviated too dramatically from the multisite codebook, the variable from that site was not included in the multisite database. Database 2 This database can be referred to as the “treatment history database.” Most sites, as previously noted, were required to develop a MIS component that could link Target Cities participants to their subsequent treatment history. Data for Database 2 were largely de-

Methodological Issues

31

rived from these local MIS systems. Database 2 contains assessment and referral experience data for participants contained in the multisite databases for all cities except Detroit for the reasons mentioned previously. Database 2 is subdivided into two files, depending upon whether participants were in the CIU (evaluation) or comparison condition in the site’s evaluation design. Database 2e contains 3,744 cases of evaluation participants, and Database 2c contains 3,203 comparison group participants. The treatment data in Databases 2e and 2c link to the evaluation baseline data for participants contained in Databases 1e and 1c, respectively. Each data set contains 33 variables, including measures of treatment entry, referrals, level of care, admissions by agency, and additional types of referrals made (University of Akron, 1999b). Database 3 As previously detailed in Table 2-1, all sites followed up their CIU group sample in some manner. Except for Detroit, all sites also included a comparison sample of participants in their evaluation designs. While the evaluation cohorts were followed for varying lengths of time, the most frequent follow-up interval was 6 months after initial assessment. Databases 3e and 3c contain multisite follow-up information for participants in these evaluation studies. Specifically, they contain the follow-up interview information on all participants in the CIU and comparison conditions, respectively. The variables contained in databases 3e and 3c are similar to those contained in Databases 1e and 1c, so that comparisons could be made between participants at baseline and at follow-up. Database 3e, the CIU group data file, contains 5,414 records. Database 3c, the comparison group data file, contains 4,711 records. Information is available on 253 variables similar to the variables in Database1, such as drug and alcohol use in the last 30 days, number of times in treatment in the last “site specific” months, criminality, employment, and ASI scores in sites where used. Databases 3e and c also contain follow-up variables such as wave of follow-up interview and evaluation design condition (University of Akron, 1998). Taken together, Databases 1e, 1c, 2e, 2c, 3e, and 3c provide researchers a rare opportunity to examine outcomes on a very large sample of substance abusers. The data offer an extremely rich source of information about changes in substance use, treatment experience, criminal activity, economic and family change, and use of social services. However, use of these data sets in an evaluation

32

Peter J. Leahy et al. Table 2-2 Target City Multisite Databases (Number of Records by Site) Database

Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Total

1

1e

1c

2e

2c

3e

3c

5,037 3,989 5,763 11,329 1,506 787 12,564 2,635 43,624

1,445 657 226 2,211 655 344 235 212 5,985

1,421 613 219 0 142 494 75 258 3,222

1,445 637 221 0 655 344 235 207 3,744

1,421 605 209 0 142 494 75 257 3,203

1,445 1,296 221 0 655 688 705 404 5,414

1,421 1,210 209 0 142 988 225 516 4,711

Notes. Database 1 = Cross-site CIU Sample Database 1e = Baseline CIU Group Database Database 1c = Baseline Comparison Database Database 2e = Treatment History – CIU Group Database 2c = Treatment History – Comparison Group Database 3e = Follow-up Database – CIU Group Database 3c = Follow-up Database – Comparison Group

context must be based upon a full understanding of their diversity. The variations across sites are not insignificant and will be important to researchers, depending upon the nature of their investigation. The most salient include 1. differences in the meaning of “CIU” and “comparison” conditions 2. differences in the nature of the comparison groups 3. differences in whether participants were followed up after assessment or discharge 4. differences in the number of cases included in the comparison or treatment groups 5. differences in the number and timing of follow-up data collection episodes 6. differences in the follow-up completion rates

Methodological Issues

33

7. differences in whether the comparison condition was concurrent in time or sequential in time with the evaluation condition 8. some evidence of possible “creaming” of less-severe participants by the SDUs in San Francisco and Cleveland (not discussed in this chapter) 9. differences in Target Cities interventions (such as case management, matching to services, and linkages) 10. differences in instrumentation The latter difference is probably the smallest of this list, since all sites, save Cleveland, used ASI measures. Cleveland was the only site to use the CIAI-C. An understanding of the creation and structure of these databases and the implementation of the Target Cities Program and evaluation in each of the sites provides only a part of the context for the studies that were generated by the program. A fuller appreciation is derived when this information is coupled with a description of the individuals who participated in Target Cities projects across the country. Participants’ demographic characteristics, comorbid conditions, employment status, and involvement with the criminal justice system are explored in the next chapter.

This page intentionally left blank.

CHAPTER

3

Participants in the Target Cities Program Ronald E. Claus and René M. Dailey

An aim of the Target Cities Program was to provide comprehensive assessment of individuals seeking referral to public sector treatment for substance abuse problems. Local Target Cities projects, driven by the common goal of centralizing the admission process, served a wide variety of individuals who sought assessment or treatment. In this chapter, a broad overview is provided of the Target Cities participants. The findings reported here assist in the understanding of the multisite databases previously described, and provide a point of reference for the remaining chapters. Attention is focused upon areas commonly investigated in substance abuse research: alcohol and drug use, psychiatric problems, employment and financial support, and criminal justice involvement.

Participants The multisite CIU sample (Database 1, N=43,254) includes data on all persons assessed at each local site’s Central Intake Unit (CIU) from its opening through December 1997. Sample size varied widely across cities, a reflection of the way in which each site implemented centralized intake and collected its data (Leahy, Stephens, Huff, & Kaye, chap. 2 in this volume). The cities contributed the following proportion of participants: Chicago, 11.6% (n=5,036); Cleveland, 8.9% (n=3,846); Dallas, 13.3% (n=5,763); Detroit, 26.1% (n=11,283); New Orleans, 3.5% (n=1,506); Portland, 1.8% (n=787); St. Louis, 29.0% (n=12,564); San Francisco, 5.7% (n=2,469). Sample sizes varied slightly

35

36

Ronald E. Claus and René M. Dailey

across analyses due to missing data or instrumentation differences between sites. Some sites are not described in all tables because of data or instrumentation issues. Fewer than 10% of the cases in the multisite CIU sample were due to repeat assessments on the same individuals. St. Louis had the most repeat participants (19.6% of its total), while other sites included 6% or fewer repeat participants. Examination of key demographic and descriptive variables revealed only minor differences by repeat status.

Demographics The demographic characteristics of the multisite sample are summarized in Table 3-1. Overall, the sample was comprised of 69.2% (n=29,908) men and 30.8% (n=13,332) women. The proportion of men across sites ranged from 55.0% (Chicago) to 78.2% (St. Louis), with two-thirds men or more observed at most sites. Among women, nearly two thirds (64.5%) of those surveyed had one or more children under the age of 18; in addition, 5.8% were pregnant at the time of assessment. Overall, two thirds of the sample (67.0%) was African American and approximately one fourth (26.5%) was White, with smaller portions of the participants identifying themselves as Hispanic (4.9%), Native American (0.4%), Alaska Native (0.01%), Asian/Pacific Islander (0.4%), or of other racial/ethnic background (0.8%). Similar compositions of race/ethnicity were observed by gender. Among men, 66.1% were African American, 26.6% White, and 5.7% Hispanic, and 1.6% identified another racial background. Among women, 69.0% were African American, 26.4% White, and 2.9% Hispanic, and 1.6% identified another racial background. Race/ethnicity varied broadly among the sites, reflecting both the geographic diversity and differences in implementation across Target Cities. The proportion of African Americans at each site ranged from one in five participants (Portland, 19.7%) to approximately four in five participants (Detroit, 86.6% and New Orleans, 78.0%). The highest proportion of Caucasians was assessed in Dallas (53.7%) and Portland (66.4%). Chicago (18.2%), Dallas (11.3%), and San Francisco (9.1%) assessed the highest proportion of Hispanic participants, while more Native Americans, Alaska Natives, and Asian/Pacific Islanders were seen in Portland (8.5%) and San Francisco (6.4%) than other sites.

Gender

Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall

% 55.0 61.2 63.2 70.5 68.7 68.2 78.2 72.9 69.2

Men (n) (2,772) (2,354) (3,640) (7,954) (1,035) (537) (9,820) (1,796) (29,908)

Race/Ethnicity

Women % (n) 45.0 38.7 36.8 29.5 31.3 31.8 21.8 27.1 30.8

(2,264) (1,485) (2,123) (3,329) (471) (250) (2,744) (666) (13,332)

African American % (n) 61.3 73.4 33.3 86.6 78.0 19.7 71.2 45.1 67.0

(3,083) (2,688) (1,917) (9,766) (1,174) (155) (8,879) (1,103) (28,765)

% 17.9 21.5 53.7 12.1 19.1 66.4 27.9 39.4 26.5

White (n) (902) (787) (3,090) (1,360) (299) (522) (3,476) (963) (11,399)

Age

Hispanic % (n) 18.2 4.3 11.3 0.0 1.7 5.3 0.6 9.1 4.9

(918) (159) (650) (0) (26) (42) (69) (222) (2,086)

Othera % (n) 2.6 0.8 1.8 1.4 0.4 8.5 0.4 6.4 1.6

(130) (30) (102) (157) (6) (67) (50) (157) (699)

(n = 43,104) Mean (SD) 35.5 34.8 33.5 37.4 31.8 33.7 33.0 40.2 35.0

(8.6) (8.7) (8.6) (8.9) (9.8) (8.9) (9.1) (10.0) (9.2)

Participants in the Target Cities Program

Table 3-1 Demographics by Site

a

Native American, Alaska Native, Asian/Pacific Islander, and Unspecified Other.

37

38

Ronald E. Claus and René M. Dailey

Participants were 35 years old on average, with mean age ranging from 31.8 years (New Orleans) to 40.2 years (San Francisco). The multisite sample was largely comprised of individuals aged 21–29 years old (25.1%), 30–39 years old (41.5%), and 40–49 years old (22.2%). Persons 18–21 years old (5.4%) and those 50–96 years old (5.7%) made up smaller portions. A large segment of the Target Cities sample had never married (56.3% overall; 41–62% across sites). The next largest group reported their status as divorced (17.2% overall; 14.9–25.4% across sites), followed by those who were married (13.6%; 7.7–23.2% across sites), with fewer participants separated (10.9%; 4.0–13.8% across sites) or widowed (2%; 1.3–3.1% across sites) at the time of assessment. Participants from all sites had on average completed 11 years of education. More than half of the participants had received a high school diploma or its equivalent (56.1%), ranging from 46.8% (Cleveland) to 66.8% (San Francisco). Participants frequently reported living with others, namely, a spouse or significant other (27.6%), living as a single parent (7.6%), or living with other family or friends (38.7%). Others reported living alone (15.5%), in a controlled environment (3.6%), or as having no stable living situation (6.9%) at the time of assessment.

Substance Use Table 3-2 describes the prevalence of drug and alcohol use in the 30 days prior to assessment. Median days of substance use are listed for participants who reported any use of that substance. Overall, 60.4% reported using alcohol at least once, 43.4% reported using crack or cocaine, 29.7% reported using marijuana or hashish, and 17.6% reported using heroin. Of those that reported any substance use, 20.7% used alcohol only, 22.7% used drugs only, and 56.6% used both alcohol and drugs. Over 60% of the participants from most sites reported some alcohol use in the past 30 days with the exception of San Francisco (47.9%) and Portland (37.1%). The proportion reporting use of crack and powdered cocaine in the past 30 days ranged from 18.1% (Portland) to 52.7% (Cleveland). The proportion reporting marijuana use in the past 30 days ranged from 19.1% (Portland) to 34.2% (Cleveland), while the proportion reporting heroin use in the past 30 days ranged from 10.9% (St. Louis) to 30.9% (Chicago). Amphetamine use in Portland and San Francisco was relatively high compared to other sites, while Chicago and Detroit had a higher rate of heroin use.

Chicago

No Use Alcohol Heroin Methadone Other Opiates Barbiturates Crack/Cocaine Marijuana Amphetamines Hallucinogens Inhalants

Cleveland

Detroit

Portland

St. Louis

San Francisco

Overall

%

Days

%

Days

%

Days

%

Days

%

Days

%

Days

%

Days

16.6 61.6 30.9 5.8 4.2 4.5 47.6 29.3 0.3 1.2 0.5

– 7.5 29.5 8.5 4.5 7.5 7.5 3.5 4.5 2.5 7.5

20.4 67.1 11.2 0.1 3.3 4.5 52.7 34.2 0.4 0.9 –

– 15 30 2 7.5 4 17 5 9 3 –

17.5 61.0 22.5 3.3 2.9 1.2 49.4 27.9 0.2 0.1 0.1

– 15 30 25 8 13 14 5 3 17 15

41.3 37.1 11.3 2.9 2.2 0.6 18.1 19.1 11.5 0.6 0.5

– 4 7 10 8 30 7.5 3 5 1 4

23.7 61.2 10.9 0.5 1.1 0.5 37.1 32.3 1.1 0.8 0.1

– 10 28 15 10 10 12 5 6 2 10

28.0 47.9 14.4 6.6 3.8 4.3 33.6 21.8 9.3 1.4 0.5

– 12 15 30 9 10 10 6 5 2.5 4

21.1 60.4 17.6 2.6 2.5 2.0 43.4 29.7 1.4 0.7 0.2

– 11 29 21 7 7 12 5 5 2 7.5

Participants in the Target Cities Program

Table 3-2 Recent Substance Use by Site

Note. Median days of use for those who used substances in previous 30 days.

39

40

Ronald E. Claus and René M. Dailey

More than one fifth (21.1%) of the sample reported no use of alcohol or drugs in the 30 days prior to assessment. Several reasons for this high level of abstinence should be considered. First, some participants may have minimized or denied recent substance use. The criminal justice system in certain cities such as St. Louis routinely required persons on probation to be assessed regardless of evidence of recent substance use. Participants facing charges of driving while intoxicated, a population that Portland targeted, or those receiving SSI for drug- and alcohol-related disability, a group served in San Francisco, may have been directed to abstain from alcohol and drugs or mandated to treatment. While some of these participants likely had not used alcohol or drugs, other participants may have denied recent use for fear of legal consequences. Second, some participants were in controlled environments at the time of assessment and thus had limited access to alcohol or drugs. Portland, for instance, had CIUs within the criminal justice system, which may in part explain their high prevalence of recent abstinence reports (41.3%). Characteristics of alcohol and drug use in the 30 days prior to assessment are presented in Table 3-3. Medians are presented as well as means and standard deviations due to the relatively high proportion of zero responses. Participants who reported use of alcohol drank a median of 10 days and experienced 4 days of alcohol-related problems in the month prior to assessment. In addition, participants spent a median of $20 on alcohol during that period. Frequency of alcohol use varied widely by site. For those who reported any use, median days of alcohol use ranged from 4 (Portland) to 15 (Cleveland and Detroit), and median days of alcohol-related problems extended from 0 (Dallas and Portland) to 15 (Detroit). As would be expected, sites with fewer days of use also reported fewer days of problems related to alcohol, r = .685, p < .001, n=40,773. Median amount spent on alcohol ranged from $10 (Portland) to $30 (Detroit and San Francisco). These findings for Portland may simply reflect that site’s greater number of participants in a controlled environment. Almost two thirds of participants reported using at least one drug other than alcohol in the 30 days prior to assessment. Those that reported use of any primary drug used a median of 18 days and experienced 21 days of drug-related problems in the previous 30. Median days of drug use ranged from 7 (Portland) to 21 (Cleveland), and median days of drug-related problems ranged from 7 (Portland) to 30 (Cleveland). Similar to the alcohol results, days of use and days of experienced problems were strongly related, r = .740, p < .001, n =

Participants in the Target Cities Program

41

39,868. Those who used drugs spent a median of $200 on drugs; the amount spent on drugs ranged from $50 (Portland) to $300 (Chicago and Detroit). Overall, Portland reported the least drug use, the fewest problems associated with drugs, and the lowest amount spent on drugs, again possibly because more participants were incarcerated. Chicago, Cleveland, and Detroit reported the highest days of drug use and days of drug-related problems. Many participants had previously received substance abuse treatment. All sites except Cleveland and New Orleans contributed information regarding previous alcohol or drug treatments. Over one third of the participants (35.4%) reported at least one previous alcohol treatment, with a substantial portion (11%) reporting three treatments or more. In addition, almost half (45.7%) of the participants had at least one previous drug treatment, with 13.5% reporting three or more treatments.

Psychiatric Problems Over one in four of all participants (27%) had previously received psychiatric treatment. The majority of participants, however, had received neither inpatient (82.6%) nor outpatient (83.4%) psychiatric treatment in their lifetime. Nine percent of participants had received only one inpatient treatment, and another 9% had received two or more inpatient treatments. Ten percent of participants had received only one outpatient treatment, and another 7% had received two or more outpatient treatments. Prevalence of inpatient treatments did not vary greatly by site (15–21%). In contrast, prevalence of one or more outpatient treatments ranged from 9.1% (Detroit) to 24.9% (Portland). Participants were asked about the lifetime occurrence of psychiatric symptoms that were not a direct result of their substance use (Table 3-4). Half of all participants (49.3%) reported depression at some point in their lifetime and 39% reported experiencing considerable anxiety. More than one in four (27.5%) reported a history of suicidal thoughts, and 16.3% reported a past suicide attempt. About one fourth of all participants reported trouble concentrating (25.7%) or trouble controlling violence toward others (24.6%), and fewer participants reported a past experience of hallucinations (11.5%). The prevalence of these symptoms indicates that many participants experienced psychological symptoms as well as substance abuse problems.

42

% Reporting Alcohol Use Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall a

63.4 67.1 67.9 61.0 51.3 37.1 61.2 47.1 61.3

Days of Alcohol Usea

Days Experienced Alcohol Problemsb

Median

Mean

(SD)

Median

Mean

(SD)

Median

Mean

(SD)

7 15 8 15 8 4 10 12 10

10.7 15.6 12.0 14.7 11.5 7.0 13.5 14.1 13.4

(9.5) (11.0) (10.6) (10.7) (10.3) (8.1) (10.7) (10.8) (10.7)

2 8 0 15 – 0 1 14 4

8.4 13.8 9.6 15.2 – 5.1 9.8 14.9 11.6

(11.2) (13.6) (12.9) (12.7) – (10.0) (12.4) (12.4) (12.8)

20 – 20 30 – 10 20 30 20

53.62 – 83.18 65.96 – 54.23 71.86 78.60 69.63

(120.68) – (187.88) (132.43) – (156.86) (148.29) (132.71) (147.50)

Overall Median (including those who reported no use) = 2; Overall Mean (SD) = 9.1 (10.8). Overall Median (including those who reported no use) = 0; Overall Mean (SD) = 7.6 (11.7). c Overall Median (including those who reported no use) = 2; Overall Mean (SD) = $46.81 (128.76). b

Amount Spent on Alcoholc

Ronald E. Clause and René M. Dailey

Table 3-3 Characteristics of Recent Substance Use by Site

Table 3-3 (cont.) Characteristics of Recent Substance Use by Site

Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall d

71.3 66.6 66.6 70.6 45.8 41.4 57.2 56.6 63.6

Days of Primary Drug Used

Days Experienced Drug Problemse

Median

Mean

(SD)

Median

Mean

(SD)

Median

Mean

(SD)

20 21 18 20 13 7 15 15 18

17.9 18.8 17.1 17.9 14.8 10.8 15.3 16.6 17.0

(11.2) (10.9) (11.2) (10.9) (10.7) (10.0) (11.1) (11.2) (11.1)

21 30 20 27 – 7 14 20 21

18.2 21.2 17.1 19.6 – 12.3 14.9 17.6 17.7

(12.4) (12.0) (13.3) (11.9) – (12.6) (13.0) (12.4) (12.7)

300 – 250 300 – 50 200 100 200

637.01 – 802.80 585.63 – 454.30 594.10 372.12 613.38

(929.98) – (1923.01) (930.81) – (965.81) (1103.91) (713.20) (1169.47)

Overall Median (including those who reported no use) = 4; Overall Mean (SD) = 11.8 (12.1). Overall Median (including those who reported no use) = 3; Overall Mean (SD) = 11.8 (13.2). f Overall Median (including those who reported no use) = 25; Overall Mean (SD) = $440.11 (1,122.31) e

Amount Spent on Drugf

Participants in the Target Cities Program

% Reporting Drug Use

43

44

Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall

Depression (n = 40,804)

Anxiety (n = 40,790)

Hallucinations (n = 40,811)

Trouble Concentrating (n = 40,801)

56.5 46.1 48.9 45.5 – 46.1 46.1 80.9 49.3

48.1 31.9 41.5 35.4 – 46.0 33.9 72.0 39.0

10.8 26.6 9.0 5.8 – 16.7 9.6 33.4 11.5

32.4 26.4 30.8 9.7 – 33.5 29.3 55.1 25.7

Trouble Controlling Violence (n = 37,065)

Suicidal Thoughts (n = 42,251)

Suicide Attempts (n = 39,865)

42.2 – 23.0 14.4 – 26.9 25.1 36.9 24.6

33.5 34.5 31.1 18.5 19.1 27.3 26.9 48.6 27.5

21.2 – 19.7 10.4 9.0 17.7 15.8 35.0 16.3

Ronald E. Clause and René M. Dailey

Table 3-4 Psychiatric Symptoms Experienced in Lifetime (%) by Site

Participants in the Target Cities Program

45

Employment and Financial Support Employment status at the time of assessment and income obtained during the 30 days prior to assessment are portrayed in Table 3-5. Most participants (63.2%) were unemployed at the time of assessment, with smaller segments employed full-time (19.5%) and part-time (9.5%). The remaining participants (7.7%) were unable to work due to being retired, disabled, a student, a homemaker, or in a controlled environment. Employment status varied widely by site. New Orleans, Portland, and especially St. Louis reported relatively high rates of full- and part-time employed participants; Chicago, Cleveland, Detroit, and Portland reported higher rates of unemployed participants; and San Francisco and New Orleans reported the most persons unable or ineligible to work. More than one third of the participants reported employment income in the month preceding assessment. Sites with higher employment rates also reported higher rates of employment income. Overall, median income for those who reported employment income was $550 during that period, while median income for individual sites ranged from $382 (Cleveland) to $800 (Dallas). Forty percent of participants reported financial assistance in the form of unemployment, public assistance, or contributions from family or friends. Chicago reported the greatest prevalence of financial assistance (67.6%), whereas Cleveland (23.2%) and San Francisco reported the least (23.0%). The median amount for those who reported financial assistance was $147, varying from $115 (Portland) to $421 (Cleveland). Information regarding illegal income was also reported by most sites. Over 7% of participants reported receiving illegal income during the month before assessment, with Chicago reporting the highest rate (18.8%) and Cleveland reporting the lowest (1.7%). For those who reported any illegal income, the median amount was $600, with a range from $300 (Cleveland) to $771 (San Francisco).

Criminal Justice Involvement Table 3-6 describes the lifetime criminal justice involvement of participants. Over one third were on probation or parole at the time of assessment. Portland had the highest rate of participants on probation or parole (76.7%), followed by St. Louis (52.7%). Other sites had approximately 30% of participants on probation with the exception of

46

Current Employment Status Full Time Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall

Part Time

Other1

Unemployed

%

(n)

%

(n)

%

(n)

%

(n)

10.3 17.2 – 9.8 22.7 21.2 31.8 13.0 19.5

(294) (640) – (1,102) (2,77) (167) (3,994) (315) (6,789)

7.1 4.0 – 4.1 7.4 7.6 17.0 9.1 9.5

(202) (150) – (459) (90) (60) (2,141) (222) (3,324)

74.4 78.7 – 79.6 51.5 71.0 44.9 48.5 63.2

(2,118) (2,925) – (8,982) (628) (558) (5,640) (1,178) (22,029)

8.2 – – 6.6 18.4 0.1 6.3 29.4 7.7

(232) – – (739) (225) (1) (789) (713) (2,699)

Note. Other = Retired, Disabled, Student, Homemaker, Controlled Environment, Maternity Leave, or Health Reasons

Ronald E. Clause and René M. Dailey

Table 3-5 Employment Status and Income by Site

Table 3-5 (cont.) Employment Status and Income by Site Income in Last 30 Days

Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall a

Unemployment, Public Aid, and Contributionsb

% reporting income received

Median (n = 15,723)

% reporting income received

Median (n = 17,577)

% reporting income received

Median (n = 3,087)

29.5 31.7 50.4 31.1 – 36.6 48.4 9.8 36.4

480 382 800 500 – 700 520 513 550

67.6 23.2 29.0 42.8 – 52.0 46.2 23.0 40.6

200 421 188 120 – 115 120 345 147

18.8 1.7 4.9 9.3 – 10.2 4.6 3.2 7.1

700 300 600 500 – 625 600 771 600

Mean (SD) for nonzero Employment Income responses = $789.06 (996.14). Mean (SD) for nonzero Unemployment, Public Aid, and Contributions responses = $302.95 (531.63). c Mean (SD) for nonzero Illegal Income responses = $1254.68 (1627.84). b

Illegal Incomec

Participants in the Target Cities Program

Employment Incomea

47

48

Currently on Parole or Probation (n = 41,207) Site Chicago Cleveland Dallas Detroit New Orleans Portland St. Louis San Francisco Overall

DWI Charges in Lifetime (%) (n = 38,168)

Drug Charges in Lifetime (%) (n = 38,845)

%

None

One Charge

Two or More

None

One Charge

Two or More

18.8 34.6 28.7 33.6 – 76.7 52.7 30.7 37.7

85.0 – 70.1 88.2 83.3 59.6 79.4 73.7 80.7

9.3 – 17.7 7.4 10.1 23.2 11.2 14.3 11.2

5.7 – 12.2 4.5 6.7 17.1 9.4 12.0 8.1

63.4 – 70.9 71.7 69.0 38.1 57.7 58.8 64.4

20.0 – 21.0 18.5 20.5 28.3 27.3 16.8 22.1

16.6 – 8.1 9.8 10.5 33.6 14.9 24.4 13.5

Ronald E. Clause and René M. Dailey

Table 3-6 Criminal Justice Involvement by Site

Participants in the Target Cities Program

49

Chicago, which reported less than 19%. Almost one in five participants had been charged with Driving While Intoxicated (DWI) in their lifetime: 11% had one charge and another 8% had two or more charges. Even more participants had received a drug charge in their lifetime: over 22% reported one charge and over 13% had two or more charges. Portland reported the highest prevalence of both DWI (40.4%) and drug charges (61.9%). Dallas and San Francisco reported relatively high rates of DWI charges (29.9% and 26.3%, respectively); St. Louis (42.3%), and San Francisco (41.2%) reported relatively high rates of drug charges. Dallas had similar numbers of drug and DWI charges, whereas Chicago, New Orleans, St. Louis, and San Francisco reported substantially greater numbers of drug charges than DWI charges. These striking site differences reflect, in part, differences in participant inclusion criteria and the criminal justice focus at particular Target Cities sites such as Portland.

Psychosocial Problem Severity Individuals with substance abuse problems commonly experience difficulties in multiple life areas. The Addiction Severity Index (ASI); (McLellan et al., 1992), used at all sites except Cleveland and New Orleans, explores problems in seven domains: medical, employment, alcohol, drug, legal, family/social, and psychiatric. Composite scores were developed for measuring treatment outcomes in each domain (McLellan et al.). These summary measures, which range from 0 (no problem) to 1 (severe problem), reflect a participant’s status during the 30 days preceding assessment. Several notable findings emerge from an examination of problem severity represented by the composite scores at each site (Table 3-7). First, the pattern of problem severity in the seven domains varied across sites. Second, not only did the Target Cities projects serve different groups of participants, but the variability of problem severity at each site was large. Finally, the proportion of participants with composite scores equal to floor (0 on medical, legal, alcohol, drug, family/social, or psychiatric) or ceiling values (1 on employment) varied by domain and across sites. To better interpret this set of composite scores, differences in the samples at each site were considered. Differences in psychosocial problem severity between sites may be due to demographic differences or may reflect other dissimilarities. Gender and race differences have previously been observed in the psychosocial functioning of treatment seekers (see,

Table 3-7 ASI Composite Scores by Site

Chicago Dallas Detroit Portland St. Louis San Francisco Overall

Medical (n = 37,281)

Employment (n = 37,284)

Alcohol (n = 36,637)

Drugs (n = 35,621)

% Mean (SD)

% Mean (SD)

% Mean (SD)

% Mean (SD)

56.9 70.2 61.1 64.7 68.4 25.8 62.1

.203 (.305) .153 (.290) .213 (.323) .171 (.306) .175 (.305) .322 (.312) .195 (.311)

55.3 18.3 48.6 53.0 32.6 56.2 40.2

.831 (.235) .586 (.279) .830 (.223) .822 (.240) .684 (.296) .869 (.209) .746 (.277)

27.1 13.9 25.4 52.4 36.5 27.4 28.3

.248 (.284) .262 (.269) .311 (.296) .111 (.208) .247 (.303) .273 (.281) .267 (.293)

7.7 26.9 3.9 37.4 39.5 20.9 21.5

.205 (.148) .156 (.151) .187 (.118) .096 (.118) .128 (.140) .132 (.116) .158 (.138)

Legal (n = 35,160)

Family/Social (n = 36,776)

Psychiatric (n = 36,993)

% Mean (SD)

% Mean (SD)

% Mean (SD)

46.9 63.5 55.9 52.9 50.6 45.7 53.4

.150 (.217) .093 (.168) .103 (.165) .145 (.212) .129 (.176) .097 (.163) .117 (.178)

22.1 17.1 42.9 40.3 34.0 20.6 31.2

.240 (.227) .235 (.224) .167 (.200) .215 (.164) .210 (.225) .202 (.188) .205 (.216)

38.5 43.2 60.6 47.1 44.5 11.7 46.2

.219 (.234) .213 (.251) .122 (.192) .170 (.219) .195 (.238) .346 (.249) .187 (.233)

Participants in the Target Cities Program

51

for instance, Brown, Alterman, Rutherford, Cacciola, & Zaballero, 1993). To better assess patterns of problems among the Target Cities participants, a series of 2
3
6 (Gender
Race
Site) ANOVA analyses on the composite scores were conducted. Because of the small proportion of participants identified as Hispanic, Native American, Alaskan, Asian/Pacific Islander, or other ethnic descent (6.8% of those with composite scores), these five groups were combined. In addition to F-ratio significance, effect sizes were used to evaluate practical significance due to the large size of the CIU sample. For analysis of variance, Cohen’s f was used to denote effect size, with values of 0.10, 0.25, and 0.40 considered small, medium, and large effects (Cohen, 1988). No medium or large effects were observed for any of the independent variables or their interactions. Also, no small effects were observed for gender or race/ethnicity. Small effects were found for Site in the domains of Employment (F = 67.907, p < .001, f = 0.212) and Psychiatric Status (F = 120.125, p < .001, f = .128). For Employment, a posthoc Tukey test revealed that, after controlling for gender and race, Dallas (.586) reported the least severe employment problems; St. Louis (.684) had slightly more problems; Portland (.822), Detroit (.830), and Chicago (.831) reported relatively higher severity; and San Francisco (.869) reported the highest severity. For Psychiatric status, Detroit (.122) reported the lowest severity of psychiatric problems, followed by Portland (.170) and St. Louis (.195); Dallas (.213) and Chicago (.219) reported relatively severe psychiatric problems; and San Francisco (.346) reported the highest severity. These findings indicate that the Target Cities sites served different populations of participants, but further show that, after controlling for race and gender, differences between the sites were small. Overall, participants reported similar problem severity in many life areas.

Discussion Target Cities participants presented a broad range of problems that encompassed substance abuse, psychological, employment, medical, family, and legal issues. The similarities and differences that were noted among Target Cities sites reflect the common problems and the diversity among these individuals, as well as system differences between sites. A number of factors should be considered in relation to the sample differences between cities. Regional differences in substance abuse trends are associated with the characteristics of

52

Ronald E. Claus and René M. Dailey

vulnerable populations and the psychosocial consequences specific to substance use (U.S. Department of Health and Human Services [DHHS], 1998a). Further, each Target Cities site was unique, with the centralized intake concept implemented in a variety of ways that differed around the system factors associated with local treatment delivery (Scott, Muck, & Foss, 2000). The ASI is considered the gold standard of assessment tools for substance abuse treatment, and the composite scores constructed from its seven domains are widely used in research (Leonhard, Mulvey, Gastfriend, & Shwartz, 2000). Among the Target Cities participants, problem severity based upon ASI composite scores showed small differences attributed to site rather than gender or race/ethnicity. The broad variability across the Target Cities sample in composite score severity and the inclusion of participants without recent substance abuse may contribute to these findings. A number of explanations have been proposed for gender and race differences in substance abuse–related problems, focusing on factors such as relative socioeconomic disadvantage, discrimination, and the values and traditions of specific racial/ethnic groups (Anderson & Armstead, 1995; Herd, 1997; DHHS, 1998b). The Target Cities participants shared relatively low socioeconomic status, as evidenced by seeking assessment for publicly funded treatment, a factor that may have minimized gender and race differences. Alternatively, a focus on specific clusters of psychosocial problems that occur in substance abuse populations (Foss, Barron, & Arfken, chap. 4 in this volume) may provide a practical resource for exploring and addressing this issue. Because of the large sample of participants, the Target Cities multisite data on composite scores may be useful as a benchmark for the represented cities. The degree and wide range of problem severity among the Target Cities participants as indicated by ASI composite scores are within the range reported for a variety of substance abuse treatment samples (e.g., Argeriou, McCarty, Mulvey, & Daley, 1994; Brochu, Guyon, & Desjardins, 1999; Brown et al., 1993). The proportion of participants reporting no problem in each ASI domain was identified as part of this data collection effort. The non-normal distribution of composite scores is not frequently discussed in the literature (see, however, Brochu et al.), yet potential floor and ceiling effects should be considered when using composite scores for longitudinal analyses. The volume, scope, and depth of the data made available through the Target Cities Multisite evaluation has provided a platform

Participants in the Target Cities Program

53

for the investigation of several aspects critical to improving substance abuse treatment in the country. From an examination of the assessment and treatment information gathered, researchers involved in the Target Cities Program began to explore several issues. Perhaps the most basic was the examination of the needs of substance abuse treatment participants and the ways in which ascertaining those needs may inform program planning.

This page intentionally left blank.

CHAPTER

4

Identifying Service Needs among Substance Abuse Treatment Participants Mark A. Foss, Nancy Barron, and Cynthia L. Arfken

Each Target Cities site created a centralized intake and assessment function in the community. Centralization was designed to provide rapid access to an array of services, both inside and outside of the substance abuse treatment system, to address the many co-occurring needs of the individuals seeking treatment. Such accessibility was seen as central to maximizing participants’ entry and retention in treatment, as well as perhaps alleviating or eliminating other problems that might have a bearing on their substance abuse. But what were the other issues and concerns faced by participants, and what array of services might best serve them? People seeking substance abuse treatment often present with an array of problems (Claus & Dailey, chap. 3 in this book). Some of these problems (e.g., medical or psychiatric) may mask symptoms or reduce the effectiveness of substance abuse treatment. Other problems such as unemployment, legal problems, or family and social conflict may affect retention and relapse potential. Long-term success of substance abuse treatment depends in part on addressing through ancillary or specially designed services the problems that commonly co-occur with substance abuse (e.g., McLellan, Arndt, Metzger, Woody, & O’Brien, 1993). Identifying co-occurring problems was an integral part of the Target Cities Program. As such, each site was required to address the range and diversity of service needs among those presenting for substance abuse treatment. These activities included a comprehensive, standardized assessment of participants’ problems with procedures

55

56

Mark A. Foss et al.

for matching participants to treatment programs; linkage and integration among existing substance abuse treatment and other health and human service agencies; and case management services to assure participant access and utilization of available services (see Arfken, Klein, Agius, & di Menza, chap. 7 in this volume). The challenge of developing an integrated system is not only obtaining the resources to address various service needs but also to develop programs and facilities that can efficiently address the needs of individuals with multiple, interrelated problems. The research conducted within the Target Cities Program has demonstrated that it is not efficient to conduct a needs assessment to estimate the number of individuals with problems that indicate specific areas of service need (medical, job training, child care, etc.) without also analyzing the co-occurrence of problems within individuals. The approach presented here uses cluster analyses to define common profiles of problems. These profiles can then be used to construct solutions in system and clinical program design that address service need complexes rather than those which target each need separately. Leff, Lieberman, and Wise (1995) made a similar argument to that presented here for the usefulness of cluster analysis in system planning. However, they conducted a retrospective investigation of mental health service utilization data rather than conducting a needs assessment. Other researchers have classified people involved in substance abuse treatment using various techniques. Most studies have focused on dual diagnosis participants with co-occurring conditions of substance abuse and mental illness. For example, Luke, Mowbray, Klump, Herman, and BootsMiller (1996) reported results from cluster analyses of the Addiction Severity Index (ASI) Severity Scores among psychiatric patients with substance abuse problems that found “extreme heterogeneity,” indicating the need for “individualized treatment programs [to] match the particular needs of patients” (p. 298). Lehman, Myers, Dixon, and Johnson (1994) found that alcohol, drug, family, and psychiatric ASI composite scores differed in expected ways among substance abuse participants with differing psychiatric problems, and they suggested different service strategies for each group. Other researchers have made service recommendations based on profiles of participants, but examined only a limited set of co-occurring conditions (e.g., Barry, Fleming, Greenley, Kropp, & Widlak, 1996; Brown, Huba, & Melchior, 1995; Brunette & Drake, 1997; Carey, 1996; Drake, Osher, & Wallach, 1989; Lambert, Griffith, & Hendrickse, 1996; Kaskutas, 1997; Reed & Mowbray, 1999; Singer, Kennedy, & Kola, 1998). Other researchers have argued for an integrated treatment sys-

Identifying Service Needs

57

tem as the best approach to the multiple problems of dually diagnosed clients, thus asserting the usefulness of combinations of participant problems for system planning (e.g., Carey, 1996; Drake, Bartels, Teague, Noordsy & Clark, 1993; Hoff & Rosenheck, 1999; Lehman, Myers, Johnson, & Dixon, 1995; Mueser, Bellack & Blanchard, 1992; Osher, 1996; Singer et al., 1998). System planning based on combinations of participant problems has yielded some positive results. Ridgely, Lambert, Goodman, Chichester, and Ralph (1998) reported a 1-year increase in interagency referrals, joint assessments of clients, and jointly sponsored client and training services in a Maine program designed to develop communities of providers working in concert to offer coordinated mental health and substance abuse treatment. Thacker and Tremaine (1989) reported enhancement of services to mentally ill substance abusers based on state system response to identified barriers. These studies, although providing valuable information, are of limited value for general system planning. They typically focus on a specific subpopulation (e.g., dual diagnosis clients) or a selected set of problems, and do not address the full range of potential problems and service needs in the entire population of those seeking publicly funded substance abuse treatment. A method for conducting a systemwide, comprehensive needs assessment using a standard, validated, widely used instrument is presented here. That approach was then extended to needs assessment by using cluster analysis to uncover profiles of participant problems among the multiple treatment systems of the Target Cities Program. The differences among Target Cities sites, noted in previous chapters, are highlighted to emphasize the importance of local needs assessment and program planning.

Method Participants A sample of 1,200 records was selected from the Target Cities Multisite Centralized Intake Unit Sample. Materials Six sites (Chicago, Dallas, Detroit, Portland, St. Louis, and San Francisco) used the Addiction Severity Index (ASI), fifth edition (McLellan et al., 1992) as part of their evaluation (see Leahy, Stephens,

58

Mark A. Foss et al.

Huff, & Kaye, chap. 2 in this volume, for a description of the ASI). A Composite Score can be constructed to represent the current level of participant functioning in each of the seven ASI domains (drug, alcohol, medical, legal, psychological, family/social, and employment). The composite scores range from 0 to 1, with higher scores indicating greater severity of problems—except for employment, where the score measures lack of employment-related assets (McGahan, Griffith, & McLellan, 1986; McLellan et al., 1992). Analytic Procedures A sample of 200 records was randomly selected from each of the six sites that used the ASI for a total sample size of 1,200. Samples of equal size allowed each site to contribute equally to the analysis. A cluster analysis was conducted based on the seven ASI composite scores. Cluster analysis requires either a rating of similarity or the computation of a similarity measure based on a set of variables. When computing similarity on a set of variables it is necessary that the variables be interval scales. The ASI composite scores are constructed by counting the number of problem indicators selected from the preexisting instrument items and then transforming to a 0 to 1 scale (see McGahan et al., 1986). Item analyses supporting an interval scale interpretation have not been reported. Three of the scales have five or fewer items, and examination of item content would suggest the scales are more ordinal than interval; that is, while the scores for each domain are ordered, there is no inherent meaning to the magnitude of difference among the scores. Therefore, the scores were transformed into percentiles (with the average percentile assigned to tied scores). These percentile scores are on interval scales (a difference between two scores is the percent of the sample who have a score in between these two scores). Furthermore, because the raw composite scores from the different domains, although nominally on a 0 to 1 scale, have markedly different distributions with significant floor or ceiling effects (and therefore so do the percentile scores), it was decided to further transform the scores to T-scores (mean of 50 and standard deviation of 10). Although not removing the floor/ceiling effect, the transformed scores have equal means and variances that have two advantages: (a) each ASI domain contributes equally to the estimated similarity scores and (b) comparison of the scores across domains is easier. All analyses were conducted using SPSS® 9.0. A Ward’s cluster analysis (Ward, 1963) was conducted using the squared Euclidean

Identifying Service Needs

59

distance among the transformed composite scores as an estimate of similarity among individuals. The number of clusters was selected based on the interpretability of the clusters, distinctiveness of the clusters as they joined together, and the rate of correct classification of individuals back into their own cluster by a discriminant analysis with the transformed composite scores as the predictor variables in the discriminant functions. Two random samples (N=1,200) were drawn from the six sites’ data in the multisite data set (N=37,401). The cluster solutions from the two samples were compared based on the presence of clusters with similar profiles. Cluster analyses were conducted on each site’s data and compared to the multisite cluster solutions. The solutions from the various samples were found to yield similar, although not identical, results. The goal of these analyses on multiple samples was to ensure that no significant cluster present in any of the cluster solutions, from one of the site samples or one of the global samples, failed to be represented in the reported cluster solution. The reported cluster solution met this criterion. The clusters in the final solution were then further validated using other participant information not in the composite scores but contained in the Target Cities Multisite CIU Sample. External variables not included in the cluster analysis that discriminate among the cluster, in addition to adding to the interpretability, support the cluster solution’s reliability and validity. Variables examined were, (a) gender, (b) age at intake, (c) race, (d) chronic medical problem, (e) employment status, (f) prompting of admission by the criminal justice system, (g) currently on probation/parole, (h) marital status, (i) current living arrangement, (j) how one spent one’s free time, (k) living with someone who has a drug/alcohol problem, (l) emotionally/physically/sexually abused in the past 30 days, and (m) received outpatient psychiatric services lifetime. Reported for each cluster are those variables where the average for the cluster was higher/lower than the average of the other clusters. The selection criterion for reporting a variable as a distinguishing characteristic of the cluster was based on a comparison of the clusters using a Chi-square test or ANOVA (p < 0.05).

Results A seven-cluster solution provided distinctive and interpretable clusters. For each cluster, (a) the cluster profile (b) the participant characteristics where the cluster members differed from the members

60

Mark A. Foss et al.

of the other clusters, (c) cluster membership by site (d) the percent of the sample represented by the cluster, and (e) for each significantly distinctive variable, the percent in the cluster and the percent in the total sample is reported. The clusters were labeled based on the domains where members of the cluster had an average T-score one-half a standard deviation (5 points) above the total average (50) or, if no domain was above average, the most prominent problem areas. T-scores for each domain for each cluster are shown in the accompanying figures. Care should be taken in interpreting the profiles, because these relative comparisons are among participants seeking publicly funded substance abuse treatment, and in many of the domains (especially substance use and employment), even a low T-score can indicate a need for services. Cluster 1. Medical, Alcohol, and Psychiatric Problems The members of first cluster exhibited a set of problems often found among people whose primary substance of abuse is alcohol (cf. Type B Alcoholics in Babor et al., 1992). In addition to their higher than average problems with alcohol, cluster members also reported greater medical and psychiatric problems than the rest of the sample, and were above the sample average on all problem areas except for Legal Problems (Figure 4-1). The individuals in this cluster appear similar to a previously identified group of dually diagnosed clients (cf. with Group 2 in Lehman, et al., 1994). Compared to the sample as a whole, members of this cluster were older (38.8 years of age vs. 35.2 years of age for the sample as a whole), more likely to have one or more chronic medical conditions that interfered with their life (63.2% vs. 32.6%), not have been prompted to enter treatment by a judge, probation, or parole officer (76.6% vs. 58.5%), not have been on probation or parole (80.3% vs. 61.1%), have had sometime in their life received psychiatric treatment (43.8% vs. 22.8%), have been living with someone who also had an alcohol or drug problem (21.8% vs. 12.5%), have spent most of their free time alone (46.9% vs. 33.8%), and were more likely to have been a male living alone with his children (24.2% vs. 9.8% of the males in the sample). Cluster 2. Psychiatric and Family/Social Problems Members of this cluster had much higher levels of psychiatric and family/social problems and tended to have more problems with both alcohol and drug use than the sample as a whole (see Figure 4-2).

Figure 4-1. Medical, Alcohol, and Psychiatric Problems

Figure 4-2. Psychiatric and Family/Social Problems

62

Mark A. Foss et al.

They had fewer legal, medical, and employment problems than others in the sample. Compared to the sample as a whole, members of this cluster were more likely to be widowed, separated, or divorced (41.1% vs. 33.8%), and to live alone if male (24.4% vs. 14.0% of males in the sample). They were less likely to have been prompted to enter treatment by a judge, probation, or parole officer (24.4% vs. 41.5%), and more likely to have been on probation or parole (83.7% vs. 61.1%). The individuals in this cluster are similar to a group of dually diagnosed clients identified by Lehman et al. (1994, cf. with Group 3). Cluster 3. Medical and Employment Problems The members of this cluster reported higher medical and employment problems, above average psychiatric problems, and lower alcohol and legal problems compared to the sample averages (see Figure 4-3). Members tended to be older (36.7 years vs. 35.2 years); to be female (39.8% vs. 30.8%); to have one or more chronic medical conditions that interfered with their life (59.1% vs. 32.6%); to have spent their free time alone (51.8% vs. 33.8%); to have been living alone with their

Figure 4-3. Medical and Employment Problems

Identifying Service Needs

63

child(ren) (25.5% vs. 12.4%); and to have received psychiatric services sometime in their life (39.4% vs. 22.8%). Although cluster members had a somewhat higher than average level of psychiatric problems and prevalence of previous psychiatric treatment, the difference was not as distinctive as in Clusters 1 and 2. Members in this cluster with comorbidity had distinct ancillary problems compared to the previous two clusters and form a third distinct dual diagnosis group. Cluster 4. Legal and Drug Problems The members of this cluster had higher than average legal and drug problems, slightly above average family/social, psychiatric, and employment problems, and below average medical problems (see Figure 4-4). Compared to the sample as a whole, members in this cluster tended to be younger (32.6 years vs. 35.2 years). They described themselves as black (59.4% vs. 49.9%) rather than white (31.3% vs. 38.4%); unemployed (80.5% vs. 61.5%); currently on probation or parole (47.8% vs. 38.9%); and as having been emotionally, physically, or sexually abused in the past 30 days (23.9% vs. 13.6%).

Figure 4-4. Legal and Drug Problems

64

Mark A. Foss et al.

Cluster 5. Employment, Drug, and Alcohol Problems The members of this cluster have somewhat higher employment problems and slightly above average drug and alcohol problems. What distinguishes the cluster members from the sample as a whole is their low level of medical, psychiatric, legal, and family/social problems (see Figure 4-5). In addition, members of this cluster are more likely to describe themselves as black (57.6% vs. 49.9%) rather than white (29.9% vs. 38.4%); as unemployed (72.7% vs. 61.5%); as not having had a chronic medical condition that interfered with their life (75.4% vs. 67.3%); as not having been prompted to enter treatment by a judge, probation, or parole officer (69.4% vs. 58.5%); and as not having been on probation or parole (70.1% vs. 61.1%). Cluster 6. Legal Problems Members of this cluster reported higher than average legal problems. They also had lower than average drug, alcohol, medical, family/social, and psychiatric problems and average employment problems (see Figure 4-6). Comparing the sample as a whole, the

Figure 4-5. Substance Use and Employment Problems

Identifying Service Needs

65

.

Figure 4-6. Legal Problems

cluster members were younger (32.2 years vs. 35.2 years) and more likely to be male (78.8% vs. 69.2%), not to have had a chronic medical condition that interfered with their life (76.0% vs. 67.3%), have been full-time employed (27.2% vs. 17.5%), and have spent their free time with their family (46.5% vs. 37.0%). Their treatment admission was more likely to have been prompted by a judge, probation, or parole officer (69.7% vs. 41.5%), and they were more likely to have been on probation or parole (67.3% vs. 38.9%). Unlike Cluster 4, members of this cluster reported low substance abuse problems, and their involvement in treatment may have as much or more to do with their criminal justice system involvement than with their substance use. Cluster 7. Low Level of Problems The members of this cluster had lower levels of problems in all seven domains (see Figure 4-7). Members of this cluster were more likely to describe themselves as white (50.7% vs. 38.4%) rather than black (39.4% vs. 49.9%); male (78.3% vs. 69.2%); employed full-time (48.1% vs. 17.5%); married (23.6% vs. 14.4%); spending their free time

66

Mark A. Foss et al.

Figure 4-7. Low Level of Problems

with their family (55.3% vs. 37.0%); and currently on probation or parole (60.3% vs. 38.9%). Their treatment admission was more likely to have been prompted by a judge, probation officer, or parole officer than the sample as a whole (72.1% vs. 41.5%). Although many cluster members reported criminal justice involvement, their average ASI Legal Composite score was lower than average. Some examples of individuals who fell into this cluster were those convicted of driving while under the influence who did not have a history of substance abuse, those being released from prison or jail and entering treatment as a condition of release, and those mandated to treatment by child and family services. Case Mix by Site If the six Target Cities projects examined in this study served the same clientele, then one would expect, except for random error, the same distribution of participants classified into each of the cluster types. As can be seen in Figure 4-8, the proportion of participants in each cluster differed significantly across the Target Cities projects

Figure 4-8. Case Mix by Site

Identifying Service Needs

67

68

Mark A. Foss et al.

(X2(1)=274.8, p