+ All Categories
Home > Documents > Dynamic Recovery

Dynamic Recovery

Date post: 02-Feb-2023
Category:
Upload: stjohns
View: 0 times
Download: 0 times
Share this document with a friend
23
401 Journal of Substance Abuse Treatment, Vol. 15, No. 5, pp. 401–423, 1998 Copyright © 1998 Elsevier Science Inc. Printed in the USA. All rights reserved 0740-5472/98 $19.00 1 .00 PII S0740-5472(97)00287-0 ARTICLE Dynamic Recovery: Comparative Study of Therapeutic Communities in Homeless Shelters for Men Hilary James Liberty, phd, Bruce D. Johnson, phd, Nancy Jainchill, phd, Judith Ryder, phd, Maria Messina, phd, Stephanie Reynolds, ba, and Mokerrom Hossain, phd National Development and Research Institutes, Inc., New York, NY Abstract – The Dynamic Recovery Project examined relationships between homelessness, substance abuse, and recovery, and investigated the effectiveness of the therapeutic community (TC) treatment model in helping homeless drug users move toward stable, drug-free living. This project compared two short-term TCs that were situated within pre-existing homeless shelters with a clean and sober dormitory. In a separate condition, peer counselors and staff were provided additional training in TC philosophy and practice to reduce program dropout. Dramatic decreases in drug and alcohol use at follow-up were verified by urinalysis. Length of time in treatment rather than specific program accounted for decreased alcohol and drug use. Important decreases in posttreatment criminality for both treatment programs were documented. The comparison group, starting with low criminality, experienced smaller, nonsignificant decreases unrelated to type of program or time in treatment. Major declines in Beck Depression Scores were evident, but were unrelated to groups or time in treatment. Training had no measurable impact on client retention or outcomes and there were no significant differences between TCs and the comparison group on posttreatment drug use, criminality, or depression. This report documents that short-term ther- apeutic communities can be successfully implemented in public shelters for homeless men. © 1998 Elsevier Science Inc. Keywords – substance abuse; alcoholism; drug treatment; therapeutic communities; homelessness. INTRODUCTION Substance abuse has been identified as the number one health problem among homeless persons (McCarthy, Argeriou, & Lubran, 1991; Wright, 1991). National stud- ies on the prevalence of alcohol and other drug problems among homeless individuals since 1980 suggest that be- tween 40 and 57% of homeless adults have alcohol disor- ders and another 10 to 20% have other drug use disorders (Fischer & Breakey, 1991; Garrett, 1992; Milburn, 1990). A 1992 study conducted in New York City indicated that 48% of those living in city shelters for single adults said they had used drugs in the past year and one third re- ported using illegal drugs at least once a month (Cuomo, 1992). Notably, this same study indicated that 85% of Received March 28, 1997; Accepted November 3, 1997. This paper was supported by the National Institute on Drug Abuse (R01 DA07598-03). The opinions expressed in this paper do not repre- sent the official position of the U. S. Government, National Institute on Drug Abuse, or National Development and Research Institutes, Inc. (NDRI). The authors express their appreciation to other key project staff: Lorenzo Aponte, Peter Blasko, Phyllis Curry, Hector Guadalupe, Kelvin Murry, William Paige, Bonnie Rodriguez, Joseph Wilson. Spe- cial thanks to Jesse Raphael, Director of STAR Therapeutic Commu- nity; William Washington, Director of Project Renewal (formerly the Manhattan Bowery Project) and the staff of STAR, Project Renewal, and the New York City Department of Homeless Services (DHS) in- cluding the DHS staff who maintain the Clean and Sober Dormitory. Without the co-operation of the hard-working staff of these programs this research project would not have been possible. Requests for reprints should be addressed to Hilary James Liberty, PhD, National Development and Research Institutes, Inc., Two World Trade Center, 16th Floor, New York, NY 10048. E-mail: hilary.liberty@ ndri.org
Transcript

401

Journal of Substance Abuse Treatment, Vol. 15, No. 5, pp. 401–423, 1998Copyright © 1998 Elsevier Science Inc.

Printed in the USA. All rights reserved0740-5472/98 $19.00

1

.00

PII S0740-5472(97)00287-0

ARTICLE

Dynamic Recovery: Comparative Study of Therapeutic Communities in Homeless Shelters for Men

Hilary James Liberty, p

h

d, Bruce D. Johnson, p

h

d, Nancy Jainchill, p

h

d,Judith Ryder, p

h

d, Maria Messina, p

h

d, Stephanie Reynolds, ba,and Mokerrom Hossain, p

h

d

National Development and Research Institutes, Inc., New York, NY

Abstract –

The Dynamic Recovery Project examined relationships between homelessness, substanceabuse, and recovery, and investigated the effectiveness of the therapeutic community (TC) treatmentmodel in helping homeless drug users move toward stable, drug-free living. This project compared twoshort-term TCs

that were situated within pre-existing homeless shelters

with a clean and sober dormitory.In a separate condition, peer counselors and staff were provided additional training in TC philosophyand practice to reduce program dropout. Dramatic decreases in drug and alcohol use at follow-up wereverified by urinalysis. Length of time in treatment rather than specific program accounted for decreasedalcohol and drug use. Important decreases in posttreatment criminality for both treatment programs weredocumented. The comparison group, starting with low criminality, experienced smaller, nonsignificantdecreases unrelated to type of program or time in treatment. Major declines in Beck Depression Scoreswere evident, but were unrelated to groups or time in treatment. Training had no measurable impact onclient retention or outcomes and there were no significant differences between TCs and the comparisongroup on posttreatment drug use, criminality, or depression. This report documents that short-term ther-apeutic communities

can

be successfully implemented in public shelters for homeless men. © 1998Elsevier Science Inc.

Keywords –

substance abuse; alcoholism; drug treatment; therapeutic communities; homelessness.

INTRODUCTION

Substance abuse has

been identified as the numberone health problem among homeless persons (McCarthy,Argeriou, & Lubran, 1991; Wright, 1991). National stud-ies on the prevalence of alcohol and other drug problemsamong homeless individuals since 1980 suggest that be-tween 40 and 57% of homeless adults have alcohol disor-ders and another 10 to 20% have other drug use disorders(Fischer & Breakey, 1991; Garrett, 1992; Milburn, 1990).A 1992 study conducted in New York City indicated that48% of those living in city shelters for single adults saidthey had used drugs in the past year and one third re-ported using illegal drugs at least once a month (Cuomo,1992). Notably, this same study indicated that 85% of

Received March 28, 1997; Accepted November 3, 1997.

This paper was supported by the National Institute on Drug Abuse(R01 DA07598-03). The opinions expressed in this paper do not repre-sent the official position of the U. S. Government, National Institute onDrug Abuse, or National Development and Research Institutes, Inc.(NDRI). The authors express their appreciation to other key projectstaff: Lorenzo Aponte, Peter Blasko, Phyllis Curry, Hector Guadalupe,Kelvin Murry, William Paige, Bonnie Rodriguez, Joseph Wilson. Spe-cial thanks to Jesse Raphael, Director of STAR Therapeutic Commu-nity; William Washington, Director of Project Renewal (formerly theManhattan Bowery Project) and the staff of STAR, Project Renewal,and the New York City Department of Homeless Services (DHS) in-cluding the DHS staff who maintain the Clean and Sober Dormitory.Without the co-operation of the hard-working staff of these programsthis research project would not have been possible.

Requests for reprints should be addressed to Hilary James Liberty,PhD, National Development and Research Institutes, Inc., Two WorldTrade Center, 16th Floor, New York, NY 10048. E-mail: [email protected]

402 H.J. Liberty et al.

those reporting monthly usage

would

participate in adrug rehabilitation program, were one available. Unfor-tunately, the few substance abuse programs that areavailable have long waiting lists and generally are notprepared to treat the multiplicity of problems associatedwith the combined stigma of being homeless and a druguser (Johnson & Muffler, 1997; National Coalition forthe Homeless, 1992).

The Dynamic Recovery Project evaluates two majorefforts that New York City has taken to establish thera-peutic communities (TCs) in shelters specialized to pro-vide intensive drug treatment for homeless men exhibit-ing dependency upon drugs and alcohol, particularlycrack. These short-term TCs were transformed from andare actually situated within homeless shelters for men.

The core philosophy of the TC is that addiction is asymptom, not the essence, of the disorder. Substanceabuse is considered a disorder of the whole person. Theprinciple aim of a TC is “a global change in lifestyle,” in-cluding abstinence from drugs and alcohol, eliminationof antisocial behavior and the internalization of prosocialattitudes and values (De Leon & Ziegenfuss, 1986). The“therapist” is the community itself. Peers and staff modelsuccessful personal change and serve as rational authori-ties and guides in the recovery process (De Leon &Rosenthal, 1989). De Leon (1997) has described this pro-cess in detail, calling it “Community as Method.” Thera-peutic community residents share the tasks of running allphases of the community. By living together in a residencethey eat, sleep, and work together, and as a communitypublicly share their private, and often painful, life stories.

De Leon (1991) characterized substance abuse treat-ment in therapeutic communities as a “dynamic recoveryprocess,” in that the relationship between the client andthe program changes during treatment. This approachhighlights the need for a greater understanding of thecomplexity of clients’ problems, about the clients’ sub-stance abuse, and about the recovery process itself as ex-perienced by clients.

Homeless, substance-abusing persons are chronicallydisaffiliated and socially isolated from conventional per-sons and society. When they enter therapeutic communi-ties they are placed into a setting where they must relateto and help each other recover as a community. Affilia-tion within a self-help community is designed to facili-tate the building of trust and the establishment of mean-ingful relationships, learning responsibility, and ultimately,moving into mainstream living. Every aspect of the TC isdesigned to aid recovery through subtle (and not so sub-tle) group pressure to change, and through the constantpresence of role models at all levels for all kind of appro-priate behavior. Through intense involvement in the lifeof the therapeutic community, the homeless, substance-abusing individual is socialized or resocialized and trans-formed to conventional living patterns (De Leon, 1994).

The effectiveness of the TC approach in rehabilitatinglong-term drug users has been well documented. Stan-

dard improvements in client behaviors involve signifi-cant reductions in drug use and criminality and concomi-tant significant increases in employment (Condelli &Hubbard, 1994; De Leon, 1984; De Leon, 1985; De Leon,1995; Hubbard, Marsden, Valley, Craddock, & Ginzburg,1989; Hubbard, Valley, Craddock, & Cavanaugh, 1984;Simpson & Sells, 1982; Tims, De Leon, & Jainchill,1994; Tims & Ludford, 1984).

Recently, attention has turned to the relationship be-tween homelessness and substance abuse. Link et al.(1994) estimated that prevalence of lifetime ever home-lessness at 14% and homeless in the last 5 years at 4.6%.They suggest that since their survey method was by tele-phone, these figures probably underestimate the truepopulation rate. Joseph (1992) attributes the causes ofhomelessness to: (a)

lack of adequate low income hous-ing stock

due to diminished construction; (b)

changes inthe job market

, with reductions in unskilled and semi-skilled manufacturing jobs and increases in low-payingservice sector and temporary jobs; (c)

unusually high un-employment rates historically

, usually concentrated inthe cities, such that even well-paid blue collar workersfound themselves jobless; (d)

decreases in governmentbenefits to poor people

, unemployed workers, personswith physical and mental disabilities, and families withmarginal income dependent on supplementary income;(e)

deinstitutionalization of the mentally ill

, which beganin the 1950s, followed by hospitals employing stricter in-take criteria in the 1970s; (f)

doubling up

, where familiesof modest means share an apartment, temporarily be-coming “couch people,” many of whom eventually endup on the street; (g)

personal crises

that cause homeless-ness, particularly for women and youth, for example,family violence that causes them to leave their homes;and (h)

alcoholism and substance abuse

which result inthe stereotypical homeless (e.g., the alcoholic male thathas frequented flophouses or the street junkie.)

Takahashi (1996) talks about the stereotyping andstigmatizing of homeless persons into this last category.She notes that the general public is often willing to as-sume that the homeless person is responsible for his orher situation and the general public often perceives theseindividuals as threatening (e.g., drug addicts and crimi-nals). These perceptions lead to the “Not-In-My-Back-yard” (NIMBY) syndrome, the result of which is thatmany facilities for homeless individuals are located inpoor or industrial neighborhoods where there is less re-sistance from local residents. (Indeed, the facilities eval-uated in this paper are located in just such neighbor-hoods.) These localities increase the homeless person’ssense of isolation, but, ironically, this situation can resultin a concentration of social services in a single area pro-viding for more efficient and convenient delivery ofservices.

Miescher and Galanter (1996) examined men whowere admitted to an alcoholism treatment clinic. Theycontrasted men who came for treatment and were domi-

Dynamic Recovery 403

ciled (

N

5

55), with men in a clean and sober dormitory(

N

5

100), and men staying in other, generally unsuper-vised men’s shelter (

N

5

34). They found no significantdifferences between groups on drug and alcohol use. Theonly significant difference in retention appeared at com-pletion of 1 year of treatment; more men were retained intreatment in the domiciled group than in the other sheltergroup. Overall retention in treatment was about 44% af-ter 6 months.

Stahler and Stimmel (1995) provides an overview of10 quantitative outcome evaluations of interventionsfunded through a joint NIAAA

/

NIDA initiative. Each ofthese interventions employed a treatment condition and ano

/

low treatment alternative condition with random as-signment of subjects to conditions. In summarizing theresults of these studies they make five important conclu-sions:

1. Treatment programs for homeless persons must notonly focus on substance abuse, but also on tangibleneeds of clients; particularly housing, income sup-port, and employment.

2. Dropout rates are particularly high for this population,no matter what type of intervention was provided.Typical completion rates are 25% to 35%.

3. Clients in both experimental and control groupsseemed to improve significantly by the end of treat-ment. However, there were rarely significant differ-ences between experimental and control groups. Hesuggests two possible reasons. First, any exposure toservices (which typically include housing) may be somuch better than no treatment that differencesbetween treatment groups are minimized by compari-son. Second,

regression to the mean

by which hemeans that many individuals are homeless at the lowpoint or extreme of their drug careers. They enteredtreatment after they had “hit bottom” in many cases.Therefore, it would be a natural progression for suchindividuals to improve regardless of treatment.Rosenblum and Magura (1996) add an additionalreason for the typical failure to find differencesbetween groups. In reviewing these studies they note,“Although some studies report having conductedurine tests, none of the studies reported pre or posttest comparisons of toxicologies, or verification ofself-reported abstinence at follow-up with a toxicol-ogy result” (p. 50).

4. Posttreatment outcomes are positive but diminishover time.

5. Certain client subgroups (e.g., those with moreeducation, less severe substance abuse, less criminality,or are less socially isolated) have a more positiveprognosis.

Stahler and Stimmel (1995) conclude “Homeless indi-viduals who enter treatment programs for their substanceabuse problems also come with a multiplicity of otherproblems . . . . Because of these concomitant problems

and characteristics, they represent one of the greatestchallenges to substance abuse treatment providers” (p.xxiii).

Overview of the Problem in New York City

When New York City began to develop TCs for home-less men in the late 1980s, they did so because the dormi-tories in armories with hundreds of residents were prov-ing extremely difficult to manage. One solution was toopen smaller shelters. In the middle of the crack epi-demic (1985–90),

1

however, many otherwise poor andlow-income men were being impoverished by their crackuse, and were either residing in shelters for long periodsof time or continually moving in and out of shelters. Anysustained effort to reduce their homelessness had to dealwith the major “cause”—their continuing abuse of crack.The need for some form of drug treatment was regardedas increasingly important.

The Greenpoint, Brooklyn complex had originallybeen built as a hospital, but it was closed earlier. ThisGreenpoint complex had three large buildings that wereconverted to dormitories for homeless men in the early1980s. In 1987–88, the New York State Division of Sub-stance Abuse Services (DSAS)

2

engaged in a carefulplanning process with the New York City Human Re-sources Administration (HRA), which then administeredthe city’s shelter system, to develop and implement a TCwithin a shelter. One building in the Greenpoint complexwas chosen to be completely devoted to this program.DSAS appointed an experienced administrator wellversed in TC principles and operations to head thisproject and also to train HRA staff in the operation andadministration of a TC. Likewise, HRA appointed an ex-perienced and committed shelter administrator to be theday-by-day administrator of the program.

The STAR (Short-Term Assessment and Referral

/

Re-habilitation) TC began operation in October 1989. Dur-ing the first 2 years, key DSAS staff implemented the TCdesign and programming, and provided training to otherHRA staff in how to maintain and improve the program-ming. This team succeeded in establishing a modifiedTC, which routinely introduced hundreds of homelessmen to the need for treatment, and referred a small pro-portion (generally less than 10% of all STAR admis-sions) to longer-term therapeutic communities through-out New York City. In 1992, DSAS staff withdrew fromSTAR House. Since then the TC has been solely admin-istered by Shelter System employees and the program’speer counselors. Administrative control for New YorkCity’s shelter programs for homeless individuals was

1

See Johnson (1991); Johnson (1990); Johnson and Muffler (1997) forextended summaries of the crack era in New York City.

2

This agency was the Division of Substance Abuse Services (1976–1993), but is currently the Office of Alcoholism and Substance AbuseServices (1993–Present).

404 H.J. Liberty et al.

transferred from HRA to the newly formed New YorkCity Department of Homeless Services in July, 1993—just 1 month before data collection for this project began.

In the early 1990s, HRA decided against further at-tempts to directly administer drug treatment to homelessmen, although STAR House continued to be adminis-tered by the Shelter system through 1996. Rather, the Di-vision of Homeless Services opted to “contract out” theprovision of drug treatment to several nonprofit organi-zations. Manhattan Bowery Corporation (MBC) ob-tained a contract to completely renovate the (Manhattan)Third Street Men’s Shelter, and reopened it as a TC forhomeless men. MBC then hired the former DSAS ad-ministrator who had established the STAR program in1989 to administer the MBC program, and he has doneso since 1993.

3

Thus, STAR House and the Third Street MBC pro-gram are very comparable programs. Both are modifiedTCs developed for homeless men, and were initially es-tablished and their key staff were trained by the sameperson. The vast majority of clients entering both pro-grams were referred by the DHS assessment centers,which served as central intake units for the shelter sys-tem. The New York City Division of Homeless Services(DHS) funded both programs during the research period.

The Clean and Sober dorm was located in one wing atthe Greenpoint complex, which contained dormitorybeds for homeless men. It was a “low intensity treat-ment” condition. Clients left the premises for treatment,and services included Narcotics Anonymous (NA) andAlcoholics Anonymous (AA) meetings. These activities,however, did not generally constitute a significant effortto engage drug users in changing their lifestyles—whichis the very purpose of the TC. Moreover, the Clean andSober dorm lacked the comprehensive treatment envi-ronment and feeling of community established at the TCsites.

METHODS

Description of Three Treatment Settings

STAR House.

STAR House is a modified TC that pro-vides addicted homeless men with a regimented, drug-free residence as well as a variety of supportive services.It is one of a few shelters in New York City with a sys-tematic program for introducing homeless individuals tointensive drug treatment. This clean, orderly, shelter-based, TC evolved from an ordinary public shelter sooverflowing with drugs, territorial assaults, and otherproblems in 1987–88 that an Emergency Medical Ser-vice (EMS) vehicle was regularly parked on the corner torush shelter residents to the hospital. Traditional TCs

have senior staff who have demonstrated long-term suc-cessful recovery. Because of the withdrawal of DSASand the transfer of management from HRA to DHSshortly before the project began, STAR lacked perma-nent staff with substantial training or experience in TCsduring the research period. Thus, the “treatment process”was truly in the hands of the homeless men, foundedupon core TC principles of self-help.

The programming in the STAR TC was “modified” inseveral key respects. First, the 3-month TC program atSTAR is much shorter than the traditional TC, wherelength of treatment typically ranges from 12 to 24months. Second, a DHS contractor provided all meals forresidents, brought from the outside. Thus, the key func-tions of planning, preparing, serving, and cleaning aftermeals were key activities designed to build a strongsense of community for many new clients entering moststandard TCs. STAR House clients had no such responsi-bilities, and were fed the same food as most residents inthe other Greenpoint shelters. Third, in most traditionalTCs, paid staff are often program graduates from previ-ous years; who are intimately incorporated into the lifeof, and often provide major leadership roles, in the com-munity. By contrast, the DHS-paid shelter staff typicallyremained in their offices and rarely directly participatedin the various programmatic activities of the clients (thiswas purposefully designed). Peer counselors and seniorpeer counselors effectively directed almost all aspects ofthe STAR TC programming. Persons in these roles wereunpaid, but had major responsibility for supervising thework and activities of all residents. The “senior peercounselors” were effectively the “directors” and “manag-ers” of the entire TC programming for 100–125 home-less men. They could consult on an as needed basis withsenior DHS-paid staff when major problems arose. Oth-erwise, the decisions and directives made by senior peercounselors effectively “ruled” the entire treatment re-gime. Much responsibility was delegated to several “peercounselors” who were responsible for various majorfunctions (housekeeping, resident movement, commu-nity meetings, etc.), and who personally supervised sev-eral newcomers and

/

or the less advanced. Because theSTAR program had a planned length of stay of only 3months, residents were often promoted to “peer counse-lor” as soon as month 2, and to “senior peer counselor”by month 3 (a promotion process that takes at least threetimes longer in standard TCs with longer planned lengthsof stay) (Messina, 1997).

The Manhattan Bowery Corporation.

The Third StreetMen’s shelter was originally planned by Dynamic Re-covery staff to be a control comparison site. However,while the original grant application, which subsequentlyfunded this evaluation, was being reviewed, the Manhat-tan Bowery Corporation (MBC) developed a modifiedTC at this site for homeless men with substance abuseproblems. This program, currently referred to as “Project

3

This program is now called Project Renewal and continues under thesame administrator.

Dynamic Recovery 405

Renewal,” was more typical of a traditional TC becausemost paid staff positions were occupied by prior gradu-ates, and its employees were not covered by city govern-ment collective bargaining procedures. Therefore, theMBC program avoided the administrative controls of abureaucratic government agency. Further, all meals andfacility maintenance were provided by the residents, andmany more “jobs” or “positions” were available in theMBC hierarchy for residents to advance rapidly.

The program provided substance abuse treatment, vo-cational, educational, and housing placement services forapproximately 200 clients. MBC’s goal was to providecomprehensive substance abuse treatment for men fromthe NYC shelter system who were identified and referredby DHS outreach teams.

As with most TCs, the community at MBC was theprimary agent of recovery; however, the clinical staffalso provided individualized counseling for each resi-dent. Case Managers prepared psychosocials, developedservice plans, and met with clients regularly to share in-formation and assess social service needs. ResidentialAides (similar to peer counselors in STAR) administeredthe therapeutic drug treatment phases of the program andmonitored client compliance with program regulations.

MBC was a 180-day treatment program divided intothree phases. During Orientation and Phase I, clients re-ceived medical evaluations and legal assessments,learned program rules, participated in group therapy andsubstance abuse seminars, and participated in a variety ofprogram job assignments. Initial vocational and educa-tional assessments occurred at this level. Level II clientswere more fully involved in program structure and as-sumed middle-level leadership roles in program func-tions. Case Managers monitored compliance with serviceplan objectives, assuring that legal, medical, and other is-sues were addressed. Clients continued with the voca-tional evaluation process, in-house educational seminarsand group therapy, and become eligible for passes onweekends. Level III and Re-Entry were the final phasesof the program, with clients concentrating on resocializa-tion, building recovery support networks, and obtainingemployment and housing. Referrals to individual psy-chotherapy were also made at this level. Re-Entry clientsworked at regular jobs and were mandated to save 75%of their net earnings. For up to 3 months after moving outof the facility, they attended After-Care groups, where is-sues such as money management and relationships wereaddressed.

Residential Aides provided most of the programming.Key paid staff at MBC, however, were more involved inthe daily operation of the TC programming than were thecomparable DHS Shelter staff at STAR House, Thus, theMBC program provided a direct comparison betweentwo TCs, both providing services to homeless menfunded by and with referrals from the same agency(DHS). MBC had somewhat longer planned length ofstay (6 months) than STAR (3 months).

Clean and Sober Dormitory.

The Clean and Sober dor-mitory (C&S) reflected conditions at a standard NewYork City shelter. The C&S dormitory was described bythe Human Resources Administration of New York Cityas a “Shelter-based support program for individuals in-volved in a community-based chemical abuse program.Counseling, peer support and discrete sleeping area areoffered to participants” (Diglio, 1992, p. 19). As such, itreflected the current realities of providing a dormitorysetting with some services for homeless persons withsubstance abuse problems.

The men in the C&S dorms were required to attend anoutside program 5 days a week, in addition to the Mon-day through Friday in-house NA meetings. Typically, themen were sent out of the dormitory during the daytime,expected to work and

/

or attend community-based ambu-latory programs. Otherwise, they were “referred out” tovarious services (e.g., employment, housing, medical,mental health, etc.). Thus, the individuals residing inC&S constituted a comparison group of homeless indi-viduals who were similar to the STAR and MBC clientsin many respects, but for whom no TC treatment or espe-cially intensive treatment was provided. The C&S dormsprimarily provided a place for homeless men to sleep andhave some meals at specific times.

C&S dorms had no planned length of stay. While afew men stayed for a long time, the vast majority leftwhen and if they found a relative or friend with whomthey could stay. Time in residence varied, since C&S cli-ents were not to be discharged until appropriate progresshad been made and an adequate alternative residence hadbeen secured. For the most part, homeless men self-selectedthemselves into the C&S wings; typically asserting intheir assessment interviews that they were clean and so-ber and

/

or were regular participants in an ambulatoryprogram. They did not want to be in the regular Shelterdorms (where many drug users and alcoholics slept).While the effective “rules” of the C&S dorm included nouse of illegal drugs or alcohol, and being expected towork at a job or to attend some outpatient program in thecommunity, considerable noncompliance was evident.Nevertheless, persons living in the C&S sections of shel-ters were considered by DHS staff to be among the mostconventional and least problematic, especially whencompared with the homeless men residing in the generalshelter population.

SPCD Training Innovation: Implementing the Staff

/

Peer-Counselor Development Model of Intervention at STAR House

A hypothesis of the current project was that by providingan experienced trainer who can powerfully introduce themany major principles about TC life in a language andmanner that can be comprehended by active crack andheroin users at the early stages of recovery in the TC, cli-ent retention and outcomes could be affected. The pur-

406 H.J. Liberty et al.

pose of this technique was to more effectively articulateTC principles so that drug abusers entering treatmentcould comprehend

why

and

how important

it was to re-main in treatment. This intervention was based on the“senior professor” model (De Leon, 1984), which wasoriginally developed and implemented at Phoenix Housein the mid-1980s. The senior professor technique wasmodified in that the principles were introduced to per-sons in key roles, especially peer counselors (as definedmore carefully above), particularly junior and senior peercounselors, and staff rather than directly to crack andheroin users. Therefore, the intervention was called aStaff

/

Peer-Counselor Development Model (SPCD)Training Innovation. Such training of peer counselorswas hypothesized to significantly improve the retentionrate(s) above that occurring during the baseline studies.Moreover, such training was hypothesized to impactupon subsequent measures of drug use, criminality, psy-chological stability, and housing. Three sessions wereconducted for Peer Counselors and two sessions wereconducted for staff.

Research Design

The purpose of this research was to: (a) compare the ef-fectiveness of these three different programs, and (b) as-sess the effectiveness of a treatment intervention calledthe Staff

/

Peer-Counselor Development Model (SPCD).The design of the research was such that client domainstracked at baseline entry into these programs (drug use,criminality, depression, homelessness, and unemploy-ment) were reassessed at follow-up interview (generallyconducted 3–9 months after departure from treatment)and significant decreases in rates of drug use, criminal-ity, etc., could be viewed as programmatic successes.Significantly greater rates of success in these domains ofone program over another can be viewed as indicators ofthat program’s superior effectiveness in treating these so-cial problems. If the STAR sample collected

after

theSPCD Training Innovation showed significant decreasescompared to the STAR Main sample (collected

before

the training innovation) in rates of retention and in thedomains noted above, this finding would indicate that theSPCD training innovation was successful.

Data Collection

Dynamic Recovery research staff sought to contact newclients within 48 hours of admission to the three treat-ment programs; however, this proved extremely difficultfor three interviewers, as clients tended to come in andleave the facility rapidly, often within hours of signingin. The average amount of time it took to contact, “re-cruit,” and complete baseline interviews with new clientswas 5 days. To assess the effectiveness of the SPCDTraining Innovation sessions, data from STAR were col-

lected in two waves, one pre- and one postintervention.The baseline interviewing time frames for the three siteswere staggered: The preintervention STAR clients wereinterviewed August 30, 1993 to April 15, 1994, (

N

5

299),

4

SPCD Training Innovation Sessions were heldAugust 24, 1994 to September 28, 1994, and the postin-tervention group was interviewed October 25, 1994, toJanuary 6, 1995 (

N

5

79); MBC clients were inter-viewed January 20, 1994 to June 1, 1994, (

N

5

150);Clean and Sober clients were interviewed May 15, 1994,to August 1, 1994 (

N

5

75). The time frames at each sitewas staggered to effectively utilize the interviewer stafftime. Midpoint interviews were conducted at 3 months or6 weeks, depending on program length, but will not bediscussed in this article. Follow-up interviews were plannedto occur 3 to 6 months after the client left treatment.

Instrumentations

All of the instruments were administered by NDRI re-search staff. Completion of the entire baseline batterytook approximately 3 hours and the follow-up interviewtook 1 to 2 hours. Questions asked at baseline were usu-ally asked again at follow-up to assess client change. In-struments consisted of: The

Center for Therapeutic Com-munity Research (CTCR) Baseline Interview

, whichincluded sections on drug and alcohol use, criminal andarrest history, HIV risk assessment, education and train-ing, employment history, family background, and healthstatus;

Residency Questionnaire

, which collected infor-mation on where the subject was living prior to treat-ment, history and pattern of homelessness; and psycho-logical measures—the

Beck Depression Inventory

(BDI)(Beck & Steer, 1988); the

Beck Hopelessness Scale

(BHS) (Beck & Steer, 1987); and the

Circumstances,Motivation, Readiness, and Suitability Scale

(CMRS; DeLeon, 1991). These are brief, paper and pencil instru-ments that were self-administered. The BDI assessed theintensity of cognitive, affective, somatic, and behavioralsymptoms of depression. The BHS instrument evaluatedthe extent of hopelessness that reflected loss of motiva-tion, feelings about the future, and expectations. The

CMRS

(De Leon, 1984) instrument assesses dynamicfactors that contributed to a subject’s entering and com-pleting treatment. The

CTCR Locator Form,

which wasalso administered, included pretreatment address andother information to enable interviewers to find the sub-ject at follow-up. The

Follow-up Questionnaire

was sim-ilar to the Baseline and inquired about a client’s druguse, illegal activities, treatment experiences and physicaland psychological health since separating from the treat-ment program. At follow-up, clients also completed a

Residency Questionnaire

that inquired about their livingarrangements, employment status, and social affiliations

4

Partial interviews are excluded from subsequent analyses.

Dynamic Recovery 407

since they separated from the treatment program. Clientswere also asked to complete the BHS and BDI. Finally,clients were asked to provide a urine specimen for thepurpose of drug testing. Additional details about the in-strumentation are included in Appendix A.

RESULTS

The results of this evaluation will be presented in fourparts: First, demographic characteristics and types ofdrugs taken will describe the sample entering treatment.Second, retention and temporal pattern of drop-out fromthe treatment will be examined to determine if clientsstayed in treatment. Third, representativeness of the fol-low-up sample to the original baseline sample will be as-sessed. Finally, the follow-up status of clients will beevaluated in terms of drug use, criminality, employmentand psychological status. Both retention and follow-upstatus will be used to determine the effectiveness of theSPCD Training Innovation and also to compare the ef-fectiveness of the STAR and MBC TCs with the Cleanand Sober dorm.

Demographic Characteristics of the Sample

Subjects in the study were exclusively male, since theseTCs and the C&S shelter location in the study acceptedonly males. Forty percent of the sample was over 35years of age, while approximately a quarter was between31 and 34 and another quarter between 26 and 30 (seeTable 1). The sample was mostly African American(73.2%), with the second largest group being Hispanic(19.0%). The remaining 7% percent of the sample wasmostly White. Half (49.5%) completed high school,while 43.3% did not, and 7.2% had at least some college.More than three quarters (76.9%) were single. These ageand racial distributions were quite typical of New YorkCity’s general assistance population (for single adults)and drug treatment populations.

In the 30 days preceding treatment (not shown in Ta-ble 1), nearly three quarters (72.6%) were unemployed,while one fifth (20.6%) claimed a full-time job. Table 1shows that 56.2% of the sample were unemployed for the6 months preceding treatment and only 34.1% had held afull-time job at any time during this 6 month period. Al-most three quarters (67.4%) earned less than $15,000 last

TABLE 1Demographic Characteristics of Study Sample at Baseline

STARMain Sample

(

n

5

299)%

STARPost-Training

(

n

5

79)%

ManhattanBowery Corp.

(

n

5

152)%

Clean and Sober(

n

5

75)%

Total(

N

5

605)%

p

Age (

N

5

606)18 to 25 10.7 12.0 11.0 11.8 11.126 to 30 23.3 25.3 31.0 18.4 24.931 to 34 24.7 24.0 23.2 21.1 23.8Over 35 41.3 38.7 34.8 48.7 40.3 .64

Ethnicity (

N

5

597)African American 73.9 68.4 71.4 78.7 73.2Hispanic 18.7 23.7 23.1 14.7 19.0White 7.4 7.9 5.4 6.7 6.9 .70

Education (

N

5

598)Did not complete high school 41.6 48.1 46.7 37.8 43.3High school diploma/GED 49.8 46.8 46.7 56.8 49.5Any college 8.5 5.1 6.6 5.4 7.2 .63

Employment status (

N

5

572) (6 monthspretreatment)

Full-time 41.3 15.2 34.9 24.3 34.1Part-time/irregular 6.8 8.9 12.5 16.2 9.7Unemployed 51.9 75.9 52.6 59.5 56.2 .61

Marital status (

N

5

579)Single 75.8 78.5 73.7 85.3 76.9Married 9.5 7.6 6.6 2.7 7.6Other (Divorced, widowed or separated) 14.7 13.9 19.7 12.0 15.5 .27

Income (

N

5

598) (past year)Less than $15,000 58.0 81.0 69.7 85.1 67.4$15,000 to $24,000 19.5 10.1 17.1 8.1 16.2$25,000 or above 22.5 8.9 13.2 6.8 16.4 .001*

Residency (

N

5

440)

a

(last 30 days)Domiciled 51.4 45.6 55.9 32.0 49.1 .006*Homeless 48.6 50.6 43.4 68.0 50.9

a

Sample is smaller because the Residency Questionnaire was implemented 6 months after study had begun.*

p

,

.01.

408 H.J. Liberty et al.

year, indicating that the typical client did not have suffi-cient income to afford a residence.

When we compared the four groups (STAR Pre-Training Innovation, STAR Post-Training Innovation,MBC, and C&S) on these characteristics obtained fromthe baseline questionnaire, only income in Table 1showed a significant difference. C&S had more individu-als in the lowest income category while the STAR prein-tervention sample had fewer people in this category andmore individuals reported earning $25,000 or above.

The

Residency Questionnaire

was completed by 440subjects or 72.7% of the sample. The interviewer, withthe client, completed a grid that listed 18 possible typesof residences. Clients responded, for each of these cate-gories, whether or not they stayed in that type of resi-dence; and if they did, for how many days in the past 30.The primary residence was where the client had lived formost of the 30 days prior to entering treatment. Clientswere considered domiciled if they had lived in: an apart-ment, a house, a condominium, a mobile home, a room-ing

/

boarding house, a hotel, a motel and an SRO. Clientswere considered homeless if they lived in a shelter, anoutdoor public place (e.g., abandoned building, car, park,sidewalk, street, train

/

bus station or tunnel, or on publictransportation), jail, prison, detention center, other treat-ment program, or in the hospital. As shown in Table 1,49.1% of those completing residency questionnaires weredomiciled and 50.9% were homeless. A significantly higherproportion of clients at C&S (68.0%) were homeless.

The upper portion of Table 2 shows drug use for thelast 30 days. Two thirds (66.4%) of the sample usedcrack in the last 30 days, compared to half (51.7%) re-porting use of alcohol, and more than a third (36.5%)who used marijuana. In this sample, snorting cocaine(17.5%) and heroin (12.2%) was more common than in-jection of these substances (3.6%, 5.6%). The low levelof injection was consistent with most comparable dataand reflects trends toward alternative means of ingestionrather than injection. Two possible reasons for this trendshould be noted. First, others have noted (Office of Na-tional Drug Control Policy, 1996) the increasing strengthof street heroin and cocaine meant that a “high” can beobtained without the need for injection. Second, the om-nipresent fear of AIDS and HIV infection.

C&S subjects reported the lowest prevalence rates formost drugs, while MBC subjects reported crack usenearly 10% higher than the STAR treatment groups. Thetwo STAR samples had nearly equivalent drug use pat-terns at their baseline interviews. Also of note, 15.7% re-ported using more than one drug at a time at least onceduring the last 30 days.

The data in Table 2 indicate that C&S subjects weresignificantly less likely to be current users of crack andalcohol to intoxication. C&S subjects also had the lowestproportions reporting use in the past 30 days of alcohol,marijuana, heroin, and cocaine, although these differ-ences were not statistically significant. This is not too

surprising. C&S dorms were for homeless men who werelargely self-selected and reported avoiding drugs. Bycontrast, STAR and MBC were specifically designed forthose exhibiting and seeking treatment for their depen-dency upon crack and other drugs.

Few differences in the criminality during the past 30days were evident among STAR and MBC treatmentgroups; current criminality was much lower among C&Sclients for every offense category. (These data are illus-trated in the middle of Table 2.) Use

/

possession of drugsoffenses were predictably highest (74.4%) in the past 30days, alcohol offenses were second (30.2%). Patterns ofillegal activity differed significantly between sites.STAR clients were highest in alcohol-related offenses(40.5%), MBC highest in drug-related offenses (posses-sion: 88.8%; sale or manufacture: 22.4%), and fencing(19.7%). A single variable was constructed aggregatingfencing, burglary, auto theft, armed robbery, and mug-ging in the past 30 days. Each of these crimes had incommon the potential for monetary gain. The purpose ofcombining these crimes was to examine whether clientsmight be committing them to obtain money for drugs.Fifteen percent of the population reported such seriousproperty crimes. Serious property crimes were highest inMBC (19.7%), intermediate in the STAR and STAR(Post Treatment) samples (15.1%, 13.9%), and lowest inC&S (1.3%).

The lower portion of Table 2 indicates that 4.5% ofthe population self-reported being HIV-positive with nosignificant differences across sites. Most (84%) of thesample indicated that they had been taught how to mini-mize the chances of getting the HIV

/

AIDS virus, withsignificant differences between samples. STAR (PostTreatment) had the lowest percent reporting (79.7%) thatthey had received instruction while MBC had the highest(91.4%). Needle injection behavior was low (5.3%) andwas consistent with the results of the reported drug usepatterns noted above. Thus, recent needle-related risk be-havior for HIV was low in these treatment groups.

A substantial proportion of the sample engaged inrisky sexual behavior, placing them at risk for HIV infec-tion at least once in the past 30 days (e.g., had sex with atotal stranger, 26.9%; or had unprotected sex, 53.1%).Six percent of clients reported having sex with anothermale at least once in the last 30 days. For all sites, clientsreported a mean of two sexual partners in the last 30 days(not shown in Table 2) with no significant differencesbetween groups. Even at this level the need for more ef-fective training or counseling in safe sex practices wasapparent, since the majority of subjects had received thistraining or counseling and, yet, did not follow safe sexpractices.

Table 3 displays total and subscale mean scores on thethree psychological measures administered to clients: theBDI, the BHS, and the CMRS. The mean total score forthe entire sample on the BDI was 16.38. In a “normal”population this score would indicate a moderate level of

Dynamic Recovery 409

depression, and was higher than scores reported in theliterature for alcoholics (13.9) and heroin users (13.2)(Beck & Steer, 1988). The mean score for MBC of 19.2was nearly half a standard deviation above the other means,suggesting the MBC clients were more depressed. How-ever, significant differences in mean scores between groupsshould be interpreted with caution since the large samplesize (N 5 552) makes significance tests very sensitive.

The overall mean score for the Beck HopelessnessScale was 4.23. This is consistent with the HopelessnessScale Manual, which lists mean scores for alcoholics of4.86 and heroin addicts of 3.89. These scores are consid-

erably less than scores of dysthymics (9.03), suicide ide-ators (9.28), and major depressives (single episode:10.10, multiple episode: 10.47). These data lend supportto the notion that the study sample was not particularlypessimistic or hopeless and is discrepant with the find-ings for depression. There were no significant differ-ences between treatment groups in BHS scores.

CMRS scores were uniformly low. Mean CMRS To-tal Score (41 items) was 142 compared to mean scoresreported by De Leon, Melnick, Kressel, and Jainchill(1994) of 165 to 171 in three samples of clients at regularTCs. The mean from the present study of 142 is more than

TABLE 2Drug Use Patterns, Illegal Activity, HIV Status, and HIV-Related Behavior

STARMain Sample

(n 5 299)%

STARPost-Training

(n 5 79)%

ManhattanBowery Corp.

(n 5 152)%

Clean and Sober(n 5 75)

%

Total(N 5 605)

% p

Drug use patterns used during last30 days

Crack 67.9 60.8 77.0 45.3 66.4 .001**Alcohol 51.2 57.0 55.9 40.0 51.7 .11Alcohol to intoxication 43.8 51.9 53.9 33.3 46.1 .01*Marijuana 40.1 35.4 33.6 29.3 36.5 .26Cocaine (snorted) 20.4 17.7 17.8 5.3 17.5 .02*Heroin (snorted) 13.7 8.9 14.5 5.3 12.2 .14Cocaine (injected) 3.7 0.0 5.3 4.0 3.6 .25Heroin (injected) 6.4 2.5 5.9 5.3 5.6 .62More than one drug 15.7 5.1 22.4 8.0 15.0 .001**

Illegal activity in the past 30 daysAlcohol offenses 40.5 29.1 17.8 16.0 30.2 .001**Use/possession of illegal drugs 75.9 58.2 88.8 56.0 74.4 .001**Sale or manufacture of drugs 12.7 17.7 22.4 1.3 14.4 .001Weapon offenses 11.4 8.8 17.8 6.7 12.1 .06Gambling/numbers or bookmaking 8.7 0.0 6.6 4.0 6.4 .03*Shoplifting 9.7 11.4 13.8 8.0 10.7 .48Serious property crimes (includes

fencing, burglary, auto theft,armed robbery, mugging) 15.1 13.9 23.0 1.3 15.2 .001**

Fencing stolen property 9.4 3.8 19.7 0.0 10.1 .001**Burglary 5.7 3.8 9.9 0.0 5.8 .02*Auto theft 3.0 5.1 2.0 1.3 2.8 .47Armed robbery or mugging 7.0 5.1 7.2 0.0 6.0 .12Assault 8.7 11.4 3.3 1.3 6.8 .01**

HIV1 rate and HIV at-risk behaviorHIV positive (n 5 597) 3.8 6.3 5.9 2.7 4.5 .19Taught how to minimize chances of

getting HIV/AIDS 82.3 79.7 91.4 81.3 84.1 .03*Needle behavior

Injected (n 5 605) 5.7 2.5 5.9 5.3 5.3 .70Shared needles (n 5 605) 1.7 1.3 3.3 0.0 1.8 .33Used dirty needles (n 5 605) 1.3 0.0 1.3 1.3 1.2 .79

Sexual behaviorHad sex with a total stranger

(n 5 605) 24.1 25.3 32.2 29.3 26.9 .29Had unprotected sex (n 5 605) 49.2 45.6 64.5 53.3 53.1 .009**Had unprotected sex w/persons

with multiple partners (n 5 605) 37.8 34.2 52.6 34.7 40.7 .005**Sex with another male (n 5 605) 4.3 3.8 9.2 8.0 6.0 .14

*p , .05.**p , .01.

410 H.J. Liberty et al.

one standard deviation (19.09) below the expected mean.The CMRS version employed by Dynamic Recovery useda 7-item Readiness Scale rather than the standard 8-itemversion, and thus a 41-item Total Score rather than thestandard 42-item scale. However, one item is not suffi-cient to account for the uniformly low scores.

Clients at C&S were less psychologically prepared fortreatment. They scored significantly lower than othergroups on the CMRS Motivation and Suitability sub-scales. This finding indicated that the residents of C&Swere less motivated to change and perceived the TCtreatment modality as less appropriate or suitable forthem. The Circumstances and Readiness scales were alsolower for C&S residents, although not significantlylower, suggesting again that these subjects were not goodcandidates for treatment.

Reliabilities (not shown in Table 3) of the Beck De-pression Scale (N 5 501) were for the Cognitive Affec-tive Scale .84, Somatic Performance Scale .77, and TotalScore .88. Reliability for the Beck Hopelessness Scale(N 5 561) was .85. Finally, reliabilities for the CMRSScales (N 5 336) were for the Circumstances Scale .33,Motivation Scale .71, Readiness Scale .68, and SuitabilityScale .71. The Beck scales were all in the acceptablerange. The reliability of the CMRS scales were lower withCircumstances clearly unacceptable at .33. The CMRSscale values are based on a lower sample size because theywere added to the study after data collection had begun.

In summary, the men in the C&S dorm, when com-pared to the STAR and MBC men, were relatively lowon a variety of measures of problem behaviors includingvirtually all drug use but especially crack and alcohol; aswell as most types of criminality especially Serious Prop-erty Crimes. C&S men had the least depression and the leastmotivation for drug treatment. They tended to be older (al-most half were over 35 years), proportionally more AfricanAmericans, single and never married, unemployed or mar-ginally employed, and had incomes of under $15,000.

Retention and Temporal Pattern of Drop-OutFrom Treatment

One of the major problems confronting all drug treat-ment programs is the difficulty in keeping clients in theprogram long enough for them to benefit from treatment.Dropout remains the rule. More clients leave any giventreatment program without completion than finish theprogram and graduate from it. Individuals may entertreatment for very mixed reasons: to obtain food, shelter,reduce their drug habits, or have a brief respite from theturmoil of the streets—all without accepting the level ofcommitment required to successfully complete the pro-gram. However, the benefits of treatment in terms of pos-itive outcomes accrue in direct proportion to the amountof treatment completed. Therefore, this evaluation beginswith an analysis of how long clients stayed in each ofthese treatment programs.

TA

BL

E 3

Bas

elin

e P

sych

olo

gic

al M

easu

res

ST

AR

Mai

n S

ampl

eS

TA

RP

ost-

Tra

inin

gM

anha

ttan

Bow

ery

Cor

p.C

lean

and

Sob

erT

otal

MS

DM

SD

MS

DM

SD

MS

Dp

Sig

nific

antly

Diff

eren

t Pai

rs

Bec

k D

epre

ssio

n In

vent

ory

(n 5

288

)(n

5 7

3)(n

512

0)(n

5 7

1)(n

5 5

52)

Cog

nitiv

e A

ffect

ive

11.1

77.

1010

.72

6.73

13.3

46.

969.

857.

9011

.41

7.2

.005

**M

BC

-ST

AR

MA

IN, C

&S

Som

atic

Per

form

ance

4.72

3.81

4.55

3.87

5.82

4.14

4.96

4.51

4.97

4.00

.06

—T

otal

15.9

09.

915

.27

9.31

19.1

69.

6014

.80

11.7

016

.38

10.1

0.0

06**

MB

C-S

TA

RM

AIN

, ST

AR

PO

ST, C

&S

Bec

k H

opel

essn

ess

Sca

le( n

5 2

98)

(n 5

78)

(n 5

152

)(n

5 7

5)(n

5 6

03)

4.15

3.73

3.74

3.24

4.60

3.93

4.34

3.88

4.23

3.74

.38

—C

MR

S( n

5 2

83)

(n 5

71)

(n 5

136)

(n 5

60)

(n 5

550

)C

ircum

stan

ces

21.1

44.

4021

.49

4.86

21.6

74.

9120

.29

4.90

21.2

24.

65.2

6—

Mot

ivat

ion

44.9

25.

8844

.42

5.43

45.8

04.

4142

.46

6.25

44.8

05.

60.0

01**

C&

S-S

TA

RM

AIN

, ST

AR

PO

ST, M

BC

Rea

dine

ss27

.25

4.19

27.0

64.

0927

.77

3.74

26.1

83.

9727

.24

4.06

.09

—S

uita

bilit

y53

.44

6.64

52.9

16.

0454

.08

5.35

51.1

05.

9353

.27

6.64

.02*

C&

S-S

TA

RM

AIN

, MB

CT

otal

142.

3615

.63

141.

5013

.56

144.

7911

.77

135.

8315

.69

142.

1014

.68

.001

**C

&S

-ST

AR

MA

IN, S

TA

RP

OS

T, M

BC

Not

e. I

f cl

ient

s fa

iled

to r

espo

nd t

o fo

ur o

r fe

wer

item

s, t

he m

ean

of t

he n

onm

issi

ng it

ems

for

that

clie

nt w

as in

sert

ed in

pla

ce o

f th

e m

issi

ng v

alue

. If

clie

nt f

aile

d to

res

pond

to

mor

e th

an f

our

item

s, h

is to

tal s

core

was

set

to m

issi

ng. 1

7.2%

of B

DI c

lient

s ha

d at

leas

t one

item

mis

sing

. 7.3

% o

f BH

S c

lient

s an

d 44

.5%

of C

MR

S c

lient

s fa

iled

to r

espo

nd to

at l

east

one

item

. The

hig

h pr

o-po

rtio

n of

the

clie

nts

skip

ping

at l

east

one

val

ue o

f the

CM

RS

is p

roba

bly

due

to th

e gr

eate

r sc

ale

leng

th (

41 it

ems)

.*p

, .0

5. *

*p ,

.01.

Dynamic Recovery 411

Program Retention. Table 4 shows the proportion ofsubjects dropping out of treatment at sequential time pe-riods. These intervals are of increasing length, since theproblem of programmatic drop-out (as in most programs)was most severe early in treatment and diminished thelonger the individual stayed in treatment. Nearly a quar-ter of the interviewed sample in each of the programsdropped out in the first 2 weeks of treatment. Anothersixth dropped out by the end of the first month. Approxi-mately half had left within 2 months. About a third of thesamples completed 90 days, the planned length of stay(except for MBC, whose planned length was 6 months)and a fifth remained for over 3 months.

In therapeutic communities, an individual’s progressthrough the program’s stages or levels is affected by hisor her personal behavior. Thus, considerable variabilityexisted in how long individuals stayed. Comparison ofthe proportion of subjects staying 90 days or longer inthe program provides a retention rate. STAR wasplanned to be 90-day duration of treatment; the MBCwas 180 days; and the C&S dormitory was approxi-mately 90 days, but without a clearly defined length ofstay. The overall average length of time in treatment was66 days (SD 5 57). Mean number of days the samplesstayed in treatment was STAR (Main) 60.9 days, STAR(Post-Training Innovation) 60.6 days, C&S dormitory66.6, and MBC 83.2 days. The 90-day retention rates forMBC and C&S samples were nearly identical to theSTAR samples. Thus, retention rates in these programswere influenced far more by the propensity of homelessmen to leave a treatment or a residential setting (such asC&S), than by the varied intensity of the treatment or bythe planned lengths-of-stay.

Program Drop-Out. The numbers presented here under-estimate the true drop-out rate, since interviewers oftenwere unable to interrupt the program’s intake process to

interview the subject for 2 or 3 days. Frequently, subjectswould drop out of treatment before interviewers couldconverse with them. Since the Dynamic Recoveryproject had only three research assistants to interviewsubjects admitted to the project sites, staff attempted tointerview only a portion of subjects entering treatment.Not surprisingly, those subjects least likely to be inter-viewed would be those who dropped out most quickly.Examination of the administrative records for STARsuggest that as high as 45% of subjects dropped out oftreatment in the first 5 days. These numbers included in-dividuals who stay only a few hours and in some casesadministrative staff were lucky even to have obtained avalid client name. Nonetheless, comparisons of drop-outand retention rates among the various samples are appro-priate since the Dynamic Recovery baseline interviewprocedures were the same for all conditions.

Comparison of Temporal Pattern of Drop Out and Re-tention Rate for STAR (Main Sample) with STAR (Post-Training Innovation). A major research question waswhether the Training Innovation was effective in reduc-ing drop-out. Visual examination of Table 4 shows thatthe intervention did not reduce drop-out. That is, thedrop-out rate for the Post-Training Innovation is margin-ally higher at first, and the overall (90 day) retention rate(30.5%) is virtually indistinguishable from the MainSample (31.4%). A chi-square analysis of the data for thetwo STAR conditions shown in the table was not signifi-cant (x2 5 6.9, df 5 6, p 5 .33) indicating no significantdifferences in either the temporal pattern of drop out orthe overall retention rate.

Comparison of Temporal Pattern of Drop Out and Re-tention for STAR (Main Sample) with Clean and SoberDormitory. Comparison of STAR (Main Sample) withthe C&S sample is important because these are two pro-

TABLE 4Proportion of Subjects Who Drop Out of Treatment by Number of Days in Program

STARa

Main Sample(n 5 299)

STARa,b

Post-Training(n 5 79)

Manhattan Bowery Corp.c

(n 5 152)Clean and Sober

(n 5 75)

Days in Treatment

%Drop

Cumulative %

% Drop

Cumulative%

%Drop

Cumulative %

% Drop

Cumulative%

5 or less 7.7 7.7 11.4 11.4 9.9 9.9 6.7 6.76 – 14 15.7 23.4 17.7 29.1 16.4 26.3 10.7 17.415 – 30 11.0 34.4 7.6 36.7 12.5 38.8 16.0 33.431 – 60 11.0 45.4 13.9 50.6 11.8 50.6 20.0 53.461 – 89 23.1 68.5 19.0 69.6 11.2 61.8 8.0 61.490 7.0 75.5 1.3 70.9 0.7 62.5 28.0 89.491 or More 24.4 99.9 20.3 91.2 37.5 100.0 10.7 100.1

Retention rates 31.4 30.5 38.2 38.7

aRetention rates based on 90 days for modal program completion.bSeven clients or 8.9% of this sample were still in residence at treatment site when data collection was completed.c180 days was modal program length. Retention rate based on proportion of subjects staying 90 days or more.

412 H.J. Liberty et al.

grams of roughly comparable length, although STARwas a therapeutic community and C&S was a shelter-based residential program enhanced with some substanceabuse treatment programming such as AA and NA meet-ings. While the overall chi-square was significant (x2 541.63, df 5 6, p 5 .0001), individual, one degree of free-dom, chi-squares show no differences between STARand C&S for the first three time periods. Then at 31–60days STAR had a significantly lower drop-out rate(11.0%) than C&S (20.0%) (x2 5 11.17, df 5 1, p 5.0008). This pattern reversed at 61–89 days with STARhaving a significantly higher drop-out rate (23.1%) thanC&S (8.0%) (x2 5 26.75, df 5 1, p 5 .0001). (These sig-nificance tests are not shown in Table 4).

Turning from the perspective of those who left tothose who stayed, the overall retention rate (those thatdid not drop-out) after 90 days was 31.4% for STAR, and7% higher for the C&S comparison group (38.7%). MBChad an equivalent 90-day retention rate (38%). This fur-ther declined to 20% at planned completion (180 days).

In summary, the Training Innovation did not signifi-cantly reduce program dropout patterns to any meaning-ful extent at the STAR facility. Retention in all these pro-grams was uniformly low. However, these retention ratescould be viewed more optimistically within the contextof TCs established in high-volume, homeless sheltersproviding services to a severely distressed inner-citypopulation of homeless substance-abusing men whosechief means of dealing with situations that they foundunpleasant was usually to leave. Under these circum-stances these retention rates (about a third remain for 3months) leaves room for cautious optimism. Clearlynone of these programs have discovered or been able toimplement procedures that are effective at interveningwith homeless men’s patterns of “moving on” and avoid-ing ongoing commitments to their self-improvement andreducing their drug abuse. Whether or not these brief in-terventions had a measurable impact upon changes in be-haviors remains to be documented below.

Follow-Up Interviews

This section presents findings regarding the posttreat-ment status of men who were interviewed 3 to 6 monthsafter their separation from treatment. Project staff con-ducted follow-up interviews with 211 (34.6%) of the 610subjects interviewed at baseline. This proportion of cli-ents contacted is well below the standard typically ex-pected in treatment outcome studies. Other studies ofhomeless persons have achieved follow-up interviewcompletion rates of about 75% or better (Stahler & Stim-mel, 1995). The procedures for finding subjects and pos-sible reasons why the follow-up rate was low are pro-vided below.

Outreach procedures were as follows. At the comple-tion of the baseline interview, each subject was asked forinformation that would help locate him at a later time

on a Locator Form. In addition to his pretreatment ad-dress, the client was asked for the name, address, andphone number of each parent, all siblings; any friendswho might be able to locate him; counselors from previ-ous medical or drug/alcohol treatment; and, finally, thelocations where he liked to “hang out” when on thestreet. When the follow-up interview for a particular in-dividual was to be scheduled, the DHS Shelter residencedatabase was searched to determine whether a particularsubject was currently registered as living in another shel-ter. Each client was mailed a letter at his last knownmailing address indicating that he would be contacted bya follow-up interviewer. The letter included an 800phone number for the project at NDRI (a free phone callfor the subject) and encouraged subjects to call to ar-range an appointment. If no one responded to the mail-ing, phone outreach was then attempted using phonenumbers provided in the Locator Form. If a subject re-sponded to the mailing by calling, the client was immedi-ately scheduled for an interview on that day, or the nextday. Whenever telephone contact with the subject wasmade, an interview was scheduled as soon as possible,since these subjects represent a highly mobile popula-tion. Delays in scheduling could result in lost contact.Such procedures have proven effective in samples drawnfrom populations with stable living arrangements andregular mailing addresses. Ideally, subjects in this projectwere to be interviewed about 6 months after they left theprogram, although the actual average was about 7 monthsafter program departure (M 5 220 days; SD 5 91 days).Many subjects at baseline interview could not give phonenumbers of friends/relatives, either because they hadlong ago lost contact with such persons, did not knowtheir current number, or were unwilling to give numbersof persons whom they did not see. Even when subjectsprovided phone numbers, these phones were often dis-connected during the intervening 6–12 months. Manysubjects were homeless due to the severity of their drugproblems. Many probably slept in different places overseveral nights, were fearful of authority, and avoided re-contact with their program. Even when project inter-viewers went to a subject’s reported “hang-out” and druglocations, most persons there, including local drug deal-ers, had little motivation to assist interviewers seekingsubjects.

Representativeness of the Follow-Up Sample to the Original Baseline Sample

If the sample of subjects interviewed at follow-up dif-fered in substantial ways from the original baseline inter-view sample, this differential follow-up could bias thefindings substantially at follow-up. For example, if theclients with highest incidence of substance abuse orcriminality were not interviewed at follow-up, this wouldlower the proportion of those engaging in these behav-

Dynamic Recovery 413

iors in the follow-up sample, increasing the appearancethat the treatment worked. Similarly, if the clients whowere most depressed were hardest to find, depressionscores would drop at follow-up. However, depressionscores in the follow-up sample were nearly identical withoriginal baseline scores for the total sample (not shown).

The effectiveness of the follow-up sample can be ex-amined in Table 5. While the overall follow-up rate wasmodest at best (a 34.8% reinterview rate—not shown inTable 5), the outreach workers and interviewers wereclearly somewhat more successful at reinterviewing theSTAR Main Sample at 41.1% and C&S at 36.0%, andless successful with the STAR Post-Training (27.8%)and MBC (25.7%) samples. STAR and C&S were bothlocated in the same Greenpoint shelter site while MBCwas at a distant Manhattan location. The STAR Post-Training sample was collected late in the project, so staffhad a much shorter period of time in which to do follow-up tracking.

Table 6 shows that in spite of these differences in fol-low-up rates, the follow-up samples for each grouplooked very similar to the total sample on baseline char-acteristics of age, proportion using key drugs in the last30 days, and proportion committing serious propertycrimes in the last 30 days. Hence, the follow-up sampleappeared quite representative of those persons inter-

viewed as part of the baseline sample. Since the findingsabout changes from Time 1 (baseline) to Time 2 (follow-up) are limited only to persons who completed both in-terviews, and the follow-up sample at baseline had fewdetectable differences from the total baseline sample, themain findings about decreased drug use, criminality, anddepression, probably were not dramatically affected bythe low reinterview rate.

In the following analysis of the follow-up data, an im-portant caveat must be raised. Because the follow-up ratewas so low, the sample may be biased in other unknownways. For example, the improvements in client outcomedocumented below might be misleading because ourstaff were able to complete follow-up interviews withperhaps the subjects who did “best” or who “bettered”themselves. Those who could not be contacted or inter-viewed at follow-up may have relapsed into drug use,crime, and continued homelessness. However, our analy-sis (shown above) reveal no particularly important differ-ences between those followed-up and the entire sampleat their initial interviews.

Follow-Up Status of Clients

The following analyses compare changes reportedamong persons who had both initial (baseline) interviewsand follow-up interviews, and who answered comparablequestions in both interviews. Thus, this outcome analysisdocuments behavior changes that occurred among thesubjects with both baseline and follow-up interviews.

Residency. During the follow-up interview, subjectswere shown questions with types of residential locations,similar to those used for the residency portion of thebaseline interview. The purpose of asking residencyquestions at follow-up was to determine if the proportionof subjects in the sample who remained homeless wassmaller than the proportion of homeless clients at base-line. Fifty-four percent of the sample were domiciled at

TABLE 6Comparison of Baseline Characteristics in Follow-up Sample Versus Entire Sample

STAR Main Sample STAR Post-TrainingManhattan Bowery

Corp. Clean and Sober

Follow-up(n 5 113)

Entire Sample

(n 5 299)Follow-up(n 5 21)

Entire Sample(n 5 79)

Follow-up(n 5 38)

EntireSample

(n 5 152)Follow-up(n 5 28)

EntireSample(n 5 75)

Age (mean) 33.9 34.6 33.2 33.9 31.9 31.4 34.5 34.6Last 30 days

% Crack use 67.9 67.3 60.8 52.4 70.0 76.3 45.3 48.1% Alcohol use 51.2 51.3 57.0 52.4 55.9 60.5 40.0 48.1% Marijuana use 40.1 40.7 35.4 19.0 33.6 31.6 29.3 33.3% Committing serious

property crimes 15.1 13.7 13.9 19.0 23.0 26.3 1.3 3.6

TABLE 5Comparison of Follow-up Rates

Total SampleSize

Follow-up Ratea

no. (%)

STAR Main Sample 299 123 (41.1)STAR Post-Training 79 22 (27.8)Manhattan Bowery Corp. 152 39 (25.7)Clean and Sober 75 25 (36.0)

ax2 p , .006.

414 H.J. Liberty et al.

some time during the follow-up period, as shown in Ta-ble 7, while the remainder (46.0%) were homeless. Thisrate of homelessness is only a slight, and not statisticallysignificant, decrease in homelessness from the 51.4% ratenoted in the respondent portion of the sample at baseline(McNemar Test x2 , 1, p 5 NS). The majority of thosecategorized as homeless (31.5%) lived in public shelters.

Not shown in Table 7 is that 25.4% of those respond-ing to the question “Were you doubled up?” (N 5 142),or about half of those reporting some housing, said“Yes.” This statistic suggests that the true proportion ofhomeless might be higher among those with a follow-upinterview. No significant variation in residency or home-lessness by treatment program was evident (data not pre-sented). In conclusion, these analyses show no change inthe proportion of subjects who were homeless posttreat-ment compared to pretreatment, and no significant dif-ferences in the rates of homelessness by program eitherbefore treatment or after.

Employment. Table 7 also documents that no significantchanges in employment status occurred after treatment,nor in the proportion claiming full-time employment.Nor did employment income among the workers change(data not presented). The drug treatment programs evalu-ated in this project did not have a substantial job trainingcomponent nor job placement efforts, which were espe-cially targeted for these homeless men. For many men,their substance abuse problems and homelessness, weredue to and contributors to their lacking job skills neces-sary to obtain full-time employment. Even when theyhad good skills in some field (e.g., carpentry, or playinga musical instrument), they rarely had the conventionalsocial skills necessary to obtain positions when compet-ing with nondrug users, or to maintain those jobs. Futureprograms targeted for homeless, substance-abusing pop-ulations need to implement a job skills training compo-nent, a job readiness component teaching “world ofwork” skills, and a job placement component that couldmore effectively help individuals find stable employmentpositions and maintain steady employment.

Drug/Alcohol Use. Significant decreases occurred in il-licit drug use or alcohol consumption following treat-ment. Comparisons are provided for self-reported druguse in the last 30 days as recorded at baseline with self-reported drug use during the follow-up period or havinga positive toxicology from the urine sample collected atfollow-up.5 Figure 1 illustrates these substantial de-creases. For STAR (Main Sample) the use of any drugdropped by 33%, from 88% to 55% while for STARPost-Training sample any drug use dropped by 31%,

5Of 211 subjects with self-reported drug use/nonuse, 167 could be veri-fied by urinalysis (79.2%).

TA

BL

E 7

Pro

po

rtio

n o

f S

ub

ject

s H

om

eles

s an

d P

rop

ort

ion

Em

plo

yed

in t

he

Las

t 6

Mo

nth

s B

efo

re T

reat

men

t an

d a

t 6-

Mo

nth

Fo

llow

-Up

ST

AR

Mai

n S

ampl

e( n

5 1

20)

ST

AR

Pos

t-T

rain

ing

(n 5

21)

Man

hatta

n B

ower

y C

orp.

(n 5

39)

Cle

an a

nd S

ober

(n 5

25)

Pre

trea

tmen

tB

asel

ine

(%)

Pos

ttrea

tmen

t F

ollo

w-u

p(%

)C

hang

e1(%

)p

Pre

trea

tmen

tB

asel

ine

(%)

Pos

ttrea

tmen

t F

ollo

w-u

p(%

)C

hang

e1(%

)p

Pre

trea

tmen

t B

asel

ine

(%)

Pos

ttrea

tmen

t F

ollo

w-u

p (%

)C

hang

e1

(%)

p

Pre

trea

tmen

t B

asel

ine

(%)

Pos

ttrea

tmen

tF

ollo

w-u

p(%

)C

hang

e1(%

)p

Dom

icile

d59

.257

.51.

747

.565

.72

18.2

61.5

46.2

15.3

28.5

41.7

213

.2H

omel

ess

48.8

42.5

6.3

0.79

52.5

72.7

220

.2.3

438

.553

.82

15.3

.29

72.5

58.3

14.2

.45

(n 5

120

)(n

5 2

2)(n

5 3

8)(n

5 2

7)U

nem

ploy

ed45

.037

.52

7.5

68.2

63.6

24.

644

.760

.515

.859

.366

.77.

4E

mpl

oyed

55.0

62.5

7.5

.23

31.8

36.3

4.5

.99

55.3

39.5

215

.8.2

340

.733

.32

7.4

.72

Ful

l-tim

e49

.255

.86.

618

.222

.74.

542

.131

.62

10.5

25.9

18.5

27.

4P

art-

time

orirr

egul

ar5.

86.

70.

913

.613

.60.

013

.27.

92

5.3

14.8

14.8

0.0

Sig

nific

ance

test

s ar

e M

cNem

ar T

for

corr

elat

ed s

ampl

es (

n .

100

) or

Bin

omia

l Tes

t (n

, 1

00).

Dynamic Recovery 415

from 77% to 46%; both were statistically significant de-clines in illicit drug use. Due to the small number of sub-jects with follow-up interviews, the following declines indrug use were not statistically significant: Among C&Ssubjects, drug use dropped 26%, from 74% to 48%.MBC subjects showed the smallest decline in any druguse of 11%, from 88% to 76%.

Detailed comparisons are given in Table 8 for majordrugs consumed. The upper third of Table 8 comparesthe percentages of self-reported use in the last 30 days atthe baseline interview (Time 1) for individual drugscombined into constructed variables for major classes ofdrugs (alcohol, cocaine, heroin, or any drug) with self-reported use of these substances at any time during thefollow-up period and verified by urinalysis. These fol-low-up data indicate the proportion of those positive fordrugs or alcohol by either any self-reported use of a drugclass or by a toxicology report that a drug was detectedby urinalysis. The percent decrease was the absolute per-centage decrease (e.g., Time 1 - Time 2). The detaileddata in Table 8 reveal that the STAR main sample expe-rienced the most significant decreases in drug consump-tion—due partly to the larger sample size (N 5 113),which provides greater sensitivity for the significancetests. Moreover, substantial decrements in usage weredocumented for all four treatment samples in crack, alco-hol, marijuana, and polydrug usage. The absolute per-centage decline in crack use was about 35% in STAR(Main), MBC, and C&S dorms; the STAR (Post-Train-ing) Sample also had a 24% decline in crack use, but thisdid not reach statistically significance. Likewise, signifi-

cant declines of 25–40% or more occurred in alcohol useand alcohol to intoxication in all four treatment samples.The STAR (Main) sample also showed significant de-creases in cocaine powder, marijuana, and heroin, al-though baseline use of cocaine and heroin use was quitelow. Comparison of the self-report data only (not shown)with the self-report augmented by urine test resultsshowed either similar results or estimates that were only5% to 10% higher. This finding suggests that the truelevel of drug usage in the follow-up period was indeedmuch lower than was self-reported at the baseline inter-view for the past 30 days.

Time in treatment (measured in days) did differslightly across treatment programs, but the differenceswere not statistically significant when using an analysisof variance (ANOVA), F(3, 206) 5 1.3, p 5 .27; STAR(Main)—M 5 73, SD 5 44; STAR (Post-Training)—M 5 64, SD 5 51; MBC—M 5 88, SD 5 73; andC&S—M 5 71, SD 5 52, F(3, 206) 5 1.3, p 5 .27.However, homogeneity of variance assumptions of thisstatistical test were not met. The variances of the fourgroups were significantly unequal (Levene statistic 513.4, p , .001). Therefore, the logarithm of the numberof days was used and the ANOVA was recomputed. Itremained not significant, F(3, 206) , 1.0, p 5 .73, indi-cating no significant differences in the number of days intreatment across programs. Table 8 shows the results oflogistic regressions comparing decreases in drug use acrosstreatment programs while controlling for number of daysin treatment. These data are difficult to interpret. Signifi-cance patterns between treatment conditions cannot be

FIGURE 1. Decreases in alcohol consumption or any illegal drug use following treatment.

416 H.J. Liberty et al.

TA

BL

E 8

Dru

g U

se a

nd

Ille

gal

Act

ivit

y (L

ast

30 D

ays)

at

Bas

elin

e V

ersu

s D

rug

Use

Sin

ce L

eft

Pro

gra

m

ST

AR

Mai

n S

ampl

e( n

5 1

13)

ST

AR

Pos

t-T

rain

ing

(n 5

21)

Man

hatta

n B

ower

y C

orpo

ratio

n(n

5 3

8)C

lean

and

Sob

er(n

5 2

8)

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Dru

g us

e se

lf-re

port

ed

Any

alc

ohol

a56

.829

.227

.6.0

01**

54.5

23.8

30.7

.11

64.1

44.7

19.4

.12

48.1

18.5

29.6

.02*

Any

coc

aine

(sn

orte

d,

inje

cted

or

crac

k)70

.833

.637

.2.0

01**

57.1

28.6

28.5

.11

81.6

42.1

39.5

.001

**55

.614

.840

.8.0

07**

Any

her

oin

(sno

rted

or in

ject

ed)

15.9

7.0

8.8

.03*

4.8

0.0

4.8

.99

21.1

15.8

5.3

.73

14.8

0.0

14.8

.13

Any

dru

g (a

lcoh

ol, a

lcoh

ol

to in

toxi

catio

n, a

nd

coca

ine

or h

eroi

n [s

nort

ed o

r in

ject

ed]

or c

rack

)87

.644

.243

.4.0

01**

76.2

33.3

42.9

.01*

86.8

63.2

23.6

.04*

74.1

22.2

51.9

.001

**D

rug

use

self-

repo

rted

or u

rine

posi

tivea

Any

alc

ohol

(N

5 2

00)

56.8

29.8

27.0

.01*

*54

.528

.625

.9.1

864

.144

.719

.4.1

948

.118

.529

.6.0

2*M

ariju

ana

( N 5

201

)40

.713

.926

.8.0

1**

19.0

9.5

9.5

.63

31.6

28.9

2.7

.99

33.3

22.2

11.1

.51

Any

coc

aine

(sn

orte

d,

inje

cted

or

crac

k)70

.844

.726

.1.0

01**

57.1

36.4

20.7

.18

81.6

57.9

23.7

.04*

55.6

33.3

22.3

.18

Any

her

oin

(sno

rted

or

inje

cted

[ N 5

201

])15

.913

.02.

9.3

64.

80.

04.

8.8

921

.115

.85.

3.7

314

.83.

711

.1.2

5A

ny d

rug

( N 5

205

)87

.655

.132

.5.0

01**

76.2

45.5

30.7

.04*

86.8

76.3

10.5

.39

74.1

48.1

26.0

.09

Illeg

al a

ctiv

ity in

the

past

30

days

Alc

ohol

offe

nses

42.7

3.4

39.3

.000

1**

19.0

9.5

9.5

.62

10.5

5.3

5.2

.68

2510

.714

.3.2

2U

se/p

osse

ssio

n of

ill

egal

dru

gs80

.312

.368

.0.0

001*

*61

.914

.347

.6.0

06*

86.8

15.8

71.0

.000

1**

67.9

10.7

57.2

.000

1**

Sal

e or

man

ufac

ture

of

dru

gs12

.04.

37.

7.0

49*

14.3

4.8

9.5

.63

23.7

5.3

18.4

.039

**3.

63.

60.

0.9

9W

eapo

n of

fens

es9.

40.

09.

4.0

01**

9.5

0.0

9.5

.50

26.3

026

.3.0

02**

7.1

0.0

7.1

.50

Gam

blin

g, n

umbe

rs,

book

mak

ing

8.5

1.0

7.5

.004

**0.

00.

00.

0—

b5.

30.

05.

3.5

04.

00.

04.

0.5

0S

hopl

iftin

g5.

11.

73.

4.2

84.

80.

04.

8.9

97.

92.

65.

3.6

310

.70.

010

.7.2

5

cont

inue

d

Dynamic Recovery 417

TA

BL

E 8

Co

nti

nu

ed

ST

AR

Mai

n S

ampl

e( n

5 1

13)

ST

AR

Pos

t-T

rain

ing

(n 5

21)

Man

hatta

n B

ower

y C

orpo

ratio

n(n

5 3

8)C

lean

and

Sob

er(n

5 2

8)

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Tim

e 1

(%)

Tim

e 2

(%)

Dec

reas

e(%

)p

Ser

ious

pro

pert

y cr

imes

(in

clud

es fe

ncin

g,

burg

lary

, aut

o th

eft,

arm

ed r

obbe

ry,

mug

ging

)13

.73.

410

.3.0

04**

190.

019

.0.1

226

.35.

321

.0.0

39**

3.6

0.0

3.6

.99

Fen

cing

sto

len

prop

erty

7.7

1.7

6.0

.06

4.8

0.0

4.8

.99

21.1

0.0

21.1

.008

**0.

00.

00.

0—

b

Bur

glar

y2.

60.

91.

7.6

24.

80.

04.

8.9

97.

92.

65.

3.6

30.

00.

00.

0—

b

Aut

o th

eft

3.4

0.9

2.5

.37

9.5

0.0

9.5

.50

0.0

2.6

22.

6.9

93.

60.

03.

6.9

9A

rmed

rob

bery

or

mug

ging

6.0

0.9

5.1

.03*

4.8

0.0

4.8

.99

0.0

2.6

22.

6.9

90.

00.

00.

0—

b

Ass

ault

7.7

0.9

6.8

.008

**14

.30.

014

.3.2

510

.50.

010

.5.1

30.

00.

00.

0—

b

a16

7 ur

ine

sam

ples

wer

e co

llect

ed fr

om th

e fo

llow

-up

sam

ple

of 2

11. R

epor

ted

drug

usa

ge fo

r th

e fo

llow

ing

five

varia

bles

was

con

side

red

posi

tive

if ei

ther

sel

f-re

port

or

urin

anal

ysis

was

pos

itive

.bIn

suffi

cien

t var

ianc

e fo

r si

gnifi

canc

e te

st.

In a

few

cas

es th

ere

wer

e po

sitiv

e ur

inal

yses

whe

re s

elf-

repo

rt d

ata

was

mis

sing

.*p

# .0

5. *

*p #

.01.

Sig

nific

ance

test

s ar

e M

cNem

ar c

hi-s

quar

es fo

r co

rrel

ated

dat

a.

418 H.J. Liberty et al.

readily compared to Table 9, since they are statisticallyadjusted for each subject’s number of days in treatment.It was important to include number of days in treatmentin these regression models because of the significantlydifferent variances between treatment programs notedabove. However, Table 8 documents that length of timein treatment is a better predictor of decreased drug andalcohol usage than the specific treatment program at-tended. This result substantiated the usual findings oftreatment outcome research by De Leon and others(Brook & Whitehead, 1980; De Leon, Wexler, & Jain-chill, 1982) that length of time in treatment is a good pre-dictor of posttreatment declines in drug use. However,this finding is typical in TCs with planned lengths of stayof 1 year or longer. This report documents this finding intreatment programs of shorter duration (3 months). Addi-

tionally, although MBC has significantly higher drug useat follow-up than C&S on four out of five drug-use indi-cators controlling for number of days in treatment. Themeaning of this is unclear since MBC had a longerplanned length of stay (6 months). It should be recalledthat statistical comparisons are pairwise with C&S, thenontherapeutic community condition.

Criminality. Illegal activity was assessed at baseline asoffenses committed in the last 30 days (Time 1) and dur-ing the follow-up period (Time 2). The follow-up periodspanned a 3- to 8-month period, compared to the last 30days assessed at baseline. In spite of the longer time pe-riod, the proportion of subjects who reported committingcrimes was generally substantially lower at follow-up in-terview that at baseline (Table 8). Serious property

TABLE 9Logistic Regression of Drug Use and Illegal Activity at Follow-Up on Baseline Drug Use, Treatment Condition, and

Number of Days in Treatment

Drug Use at Follow-Up

BaselineDrug Score

Significance of Number of Days in

Treatment

b p Treatment Outcomes b p

Composite self-report variables (N 5 199)Any alcohol 20.5 .004** MBC.C&S, p 5 .02 20.0109 .002**Any cocaine 20.272 .148 MBC . C&S, p 5 .02 20.0105 .002**Any heroin 20.898 .003** 20.0059 .320Any drug 20.227 .329 STAR (Main) . C&S, p 5 .04 20.012 .0004**

MBC . C&S, p 5 .005Self-report or urine positivea

Any alcohol (N 5 200) 20.531 .002** MBC . C&S, p 5 .02 20.0104 .0032**Marijuana (N 5 201) 21.036 .000** 20.0159 .0020**Any cocaine (N 5 204) 20.388 .024* MBC . C&S, p 5 .05 20.0103 .0017**Any heroin (N 5 201) 20.975 .000** 0.2724 .0001**Any drug (N 5 205) 20.2 .350 MBC . C&S, p 5 .007 20.0116 .0008**

Illegal activity at follow-upb

Alcohol offenses 20.45 .17 20.001 .82Use/possession of illegal drugs 20.07 .81 20.002 .62Sale or manufacture of drugs 20.34 .42 20.002 .74Weapon offenses —c —c —c —c

Gambling, numbers, bookmaking —c —c —c —c

Shoplifting 3.9 .95 20.01 .45Serious property crimes

(includes fencing, burglary, auto theft, armed robbery, mugging) 20.45 .33 20.01 .28

Fencing stolen property 4.09 .96 20.02 .26Burglary 4.05 .96 20.01 .49Auto theft 3.4 .96 20.002 .84Armed robbery or mugging 25.61 .86 20.003 .79Assault 26.37 .93 20.02 .50

Note. Total follow-up sample consisted of 211 individuals and occured 3 to 6 months after leaving the program.a167 urine samples were collected from the follow-up sample of 211. Reported drug usage for the following five variables is if eitherself-report or urinanalysis report was positive. In a few cases there were positive urinalyses where self-report data was missing.Comparisons are with baseline data from previous category “Composite Self-Report Variables” except Marijuana which is compared tothe primary “Self-Report” category.bN 5 203.cInsufficient variance in dependent variable or logistic regression procedure failed to converge on a solution.*p # .05. **p # .01.

Dynamic Recovery 419

crimes (which included fencing, burglary, auto theft,armed robbery, and mugging) exhibited significant de-clines for all programs except for C&S (which was verylow at baseline). For STAR (Main Sample), seriousproperty crimes declined by 10.3%, from 13.7% to 3.4%.The STAR (Post-Training) sample declined from 19.0%to 0. MBC subjects showed a decrement of 21.0% from26.3% to 5.3%. All four treatment samples significantlyreduced their self-reported involvement in illegal drugpossession offenses (Table 9). For STAR (Main Sample)use/possession offending was reduced by 68.0%, from80.3% to 12.3%. For STAR (Post-Training) a 47.6% re-duction occurred, from 61.9% to 14.3%. For MBC, a71.0% decrease, from 86.8% to 15.8%. Finally, C&S hada 57.2% drop, from 67.9% to 10.7%. All these decreasesin both drug possession crimes and serious propertycrimes were statistically significant and very substantialin magnitude. Only the reduction of 3.6% in seriousproperty crimes for C&S, the comparison condition, wasnot significant, mainly because C&S subjects report sofew property crimes at their baseline interview.

Logistic regressions were run using baseline drug us-age, dummy codes for program (with C&S as the com-parison condition) and the number of days in treatmentas independent variables (Table 9). The dependent vari-ables for these regressions were the types of crimes that aclient may or may not have committed during the follow-up period. None of these independent variables were sig-nificant. Therefore, neither baseline illegal activity nortime in treatment was important in predicting illegal ac-tivity during the follow-up period.

In summary, individuals interviewed at follow-upwere very successful in significantly reducing their self-reported criminal behavior following treatment. How-ever, consistency across variables within subjects wasnot evident. Predicted patterns of change did not occur;statistical controls and efforts to build logistic regressionmodels to predict follow-up criminality from baselinebehavior and/or time in treatment, or type of programwere not successful. The reasons for these dramatic re-ductions in drug possession and serious property crimeswere not evident in the data.

Psychological Measures. The reader will recall that sub-jects at baseline were not overly depressed since themean for the Beck Depression Inventory was initially15.9, 15.3, 19.2, and 14.8. (see Table 3). Baseline BDITotal Scores for the entire follow-up sample (N 5 184)were 16.4, 16.4, 20.2, and 14.8. These scores are verysimilar to the original baseline scores. Hence, the sub-jects interviewed at follow-up were neither more nor lessdepressed than the original baseline sample on their orig-inal baseline scores. Comparing these baseline scores totheir respective follow-up scores, significant decreases indepression occurred for each group (STAR Main 16.41–8.62, p , .001; STAR Post-Training 16.37–10.05, p ,.004; MBC 20.18–11.68, p , .001; C&S 14.77–7.50,

p , .014). These data showed that all groups experi-enced major, significant decreases in BDI scores with amagnitude of slightly more than a standard deviation, orabout eight scale points, indicating substantially less de-pression following treatment. Moreover, BDI scores ofthese homeless men at follow-up were very similar topopulations of college students or other “normal” popu-lations (Beck & Steer, 1988). The Beck HopelessnessScale also decreased in each setting, although only forthe STAR Main sample was the decline significant(4.32–3.48, p , .05). Clearly, the entire sample was lessdepressed at follow-up.

Regressions were run to determine if differences be-tween treatment groups affected the magnitude of de-pression scores when controlling for initial, pretreatmentdepression score, and the number of days in treatment foreach of the BDI Total (R2 5 .08, p 5 .02), Congnitive-Affective (R2 5 .07, p 5 .04), and Somatic-PerformanceScores (R2 5 .08, p 5 .01). Examination of these data re-vealed that the baseline score was a good predictor offollow-up score; that is, subjects who were most de-pressed before treatment were more likely to be de-pressed after treatment. In each case, there was a small,marginally significant negative relationship betweennumber of days in treatment and score at follow-up. Thisrelationship, however, was too small to be meaningful.

The magnitude of the proportion of variance ac-counted for by these regressions is not sufficient to ex-plain why subjects were less depressed following treat-ment. However, the entire sample was clearly lessdepressed following treatment. The baseline depressionscores, while normal for drug-using populations, wouldbe high and considered depressed in a normal popula-tion. In summary, these analyses show: subjects in eachtreatment group showed significant decreases in depres-sion between baseline and follow-up interviews, how-ever, time in treatment and treatment groups were notsignificant predictors of the decreases in depression doc-umented by these data.

DISCUSSION

This research project has documented several majorthemes.

Therapeutic Communities Were Effectively Implemented Within the Homeless Shelter System

Short-term TCs can be successfully established in publicshelters for homeless men and administratively directedby staff of large, urban, shelter agencies. Thus, the STARHouse program has been maintained by the New YorkCity Department of Homeless Services as a viable drugtreatment facility for nearly 7 years. On the other hand,shelter administrators can also contract with nonprofitorganizations having expertise and experience in manag-ing a TC to implement TCs for homeless men, which can

420 H.J. Liberty et al.

also accept referrals from the shelter system. After re-ceiving such a contract, the MBC literally transformedan ugly, trouble-ridden men’s shelter into a completelyrenovated facility in which clients could live comfortably.

Both STAR and MBC TCs were effective becausethey relied primarily on the self-help efforts of “addictstreating addicts” with supervision provided by other ad-dicts and rehabilitating drug abusers working together ina community. Both implemented a consistent therapeuticcommunity philosophy, program structure, hierarchywith much mobility, encounter sessions, and many otheraspects of therapeutic communities. While a majority ofentrants to a therapeutic community leave in the firstmonth or 3 months, new leadership was constantly beingidentified and trained. These homeless men were placedinto positions, job functions, and management-like re-sponsibilities that they had rarely achieved in conven-tional society. By assuming the important peer counselorroles, they performed at much higher levels of compe-tence than most of these individuals had previously func-tioned. Not only were the peer counselors not usingdrugs and committing crimes, they were engaged in ac-tivities that restored their self-esteem, confidence, andprovided a purpose for their life.

Clean and Sober Dorms Apparently Attracted Socially Isolated Homeless Men With Fewer Problematic Behaviors

One unanticipated surprise emerging from the project’sresearch design and empirical findings involved thehomeless men who were recruited at entry to the C&SDorm. When compared with the homeless men in STARand MBC, the C&S subjects were generally low on a va-riety of measures of problem behaviors, including virtu-ally all illicit drug use, and most forms of criminality.They also had the least depression. Likewise, the C&Smen also appeared to be more socially isolated. Relativeto STAR and MBC subjects (Table 1), C&S men wereolder (almost half were over 35 years), African Ameri-can (79%), single, never married (85%), unemployed(60%) or, if employed, had irregular jobs (16%), and hadincomes of under $15,000 (85%). Although not revealedin the statistical data, men entering C&S dorms often re-ported keeping to themselves and having few friends.Many reported wanting to get out of the general shelterswhere so many drug abusers and sellers were present andviolence/threats were commonplace. At the same time,most also resisted pressure from staff counselors to enterTC or other more intensive programs designed to changetheir behavior patterns. Overall, they appeared to projectinterviewers to be relatively isolated socially from theother men, not routinely involved in heavy alcohol anddrug abuse, and to be low-functioning normals. Thesedifferences may account for lack of significant differ-ences on outcome measures between the TC treatmentgroups and the C&S subjects.

It is quite doubtful that placing chronic crack users(many such persons were in STAR and MBC) in theC&S dorms, with virtually no intensive treatment pro-gramming, would result in similar declines in drug use.Rather than being a true treatment modality, the C&Sdorms appeared to be a relatively “safe” place for so-cially isolated, but otherwise relatively normal, homelessmen to sleep and eat—and to avoid the threats of othermore aggressive homeless men whom they routinely en-countered in the streets or other general shelters.

Declines in Drug and Alcohol Abuse

One major finding that this research documents is largedecreases among those who completed follow-up inter-views in drug and alcohol use verified by urinalysis. Theabsolute decreases ranged from a 33% decline in theSTAR Main sample, a 31% reduction in the STAR Post-Training sample, and 26% in the C&S dormitory sample,to 11% in the MBC sample. The statistical analysis, how-ever, showed that the length of time in treatment was themost important factor associated with respondent’s de-creased alcohol and drug use, rather than the specificprogram attended.

Declines in Criminality

Important decreases in posttreatment criminality werealso documented. Subjects from all programs were suc-cessful in significantly reducing their involvement in se-rious property crimes (fencing, burglary, auto theft,armed robbery, and mugging), except the C&S dormi-tory subjects, who began with very low involvement. Se-rious property crimes decreased from 10% for STAR(Main Sample), 19% for STAR (Post-Training Innova-tion), and 21% for MBC. Drug use/possession crimeswent down for all programs, with absolute decreasesranging from 48% to 68%. Our analyses, however, didnot show that this reduction in property and drug posses-sion crimes was due to these programs, nor to length oftime in treatment. The reason for the dramatic drop incriminality remains unknown.

Decreased Depression

Project findings also documented that follow-up depres-sion scores dropped by nearly a full standard deviation,indicating that subjects were much less depressed whenassessed at follow-up than at baseline. Indeed, depres-sion scores among the follow-up subjects were very sim-ilar to scores for “normal,” non-drug–abusing populations.

No Change in Employment Status

The proportion of the sample with any employment orfull-time employment did not increase significantly fol-lowing participation in any of these treatment programs.

Dynamic Recovery 421

Therefore, not surprisingly, subject income did not rise.Of course, none of these programs were specifically in-tended to train persons for jobs, nor did they have sup-port to provide intensive placement services for home-less men. Even among those who did obtain employment,most did not earn enough to support a residence.

No Change in Homelessness

The proportion of men who were homeless or living inshelters did not change following participation in any ofthe treatments. Again, none of these programs had strongaccess to low-income housing, nor ways of subsidizinghousing costs for graduates. Clearly subjects participat-ing in these treatment programs needed additional sup-ports and training to develop better job skills and to lo-cate jobs paying an adequate wage. Additional incomewould be necessary for most of these individuals to beable to locate and maintain permanent housing. Futureresearch must be designed to more carefully address howto gain housing.

Treatment Intervention Had NoMeasurable Impact

The treatment intervention of a Staff/Peer-Counselor De-velopment Model (SPCD) was implemented. Trainingsessions were delivered to peer counselors and DHSstaff, but this program had no measurable impact uponclient retention nor outcomes. The number of sessions or“dosage” of such a treatment intervention was clearly notsufficient to effect the rate of program drop-out.

No Significant Differences Between Treatment Conditions on Outcome Variables

At follow-up the three treatment groups did not differsignificantly on drug and alcohol use, criminality or anyother variables which were assessed.

IMPLICATIONS FOR TREATMENT AND FOR FUTURE RESEARCH

This project has documented that homeless male subjectsparticipating in these short-term TCs were able to sub-stantially decrease their alcohol and drug abuse, propertyand drug possession offending, and depression. Suchfindings have rarely been reported previously in the pro-fessional scientific literature, based upon short-term TCresearch. Typically, significant decreases in posttreat-ment drug use behaviors require 12 months or more oftreatment (Wexler, Lipton, & Johnson, 1988). Unfortu-nately, this research was not able to identify the mecha-nisms and factors associated with the dramatic declinesamong homeless men in their posttreatment criminality(e.g., substantially less serious property crime and drugpossession crime), or decreased depression. These find-

ings suggest the need for a replication study and furtherresearch to document various factors bringing about suchimportant declines in drug abuse, criminality, and de-pression among homeless men.

Additionally, the lack of significant findings betweenthe TC treatment conditions and the C&S comparisongroup is typical of findings with homeless, substance-abusing populations (e.g., Miescher & Galanter, 1996;Stahler & Stimmel, 1995.) As noted above, this findingresults from the need to provide a broad range of servicessimultaneously to this needy population, and probably aregression to the mean effect in that some homeless cli-ents have truly “hit bottom” and would have improvedwithout treatment. When these last clients fall in the non-TC comparison group, they mask true differences.

There were two findings for treatment: (a) short-termTCs can be effectively established within communityhomeless shelters, and (b) subjects in the follow-up studysignificantly reduced their drug and alcohol abuse andcriminality. The limits of the changes reflect the limita-tions of the treatment that is provided during such an ab-breviated time frame. While drop-out rates were high,and only one third of the subjects completed these short-term treatments, low retention rates remain a standardproblem for all drug treatment programs. The TCs in thisproject were evaluated within a few years of their incep-tion. When TCs have been in operation for several years,staff become more experienced, and drop-out rates typi-cally decrease. Nonetheless, new innovations are stillneeded to encourage individuals to stay in treatment forlonger periods.

From a research perspective, improved methodologiesfor remaining in contact with homeless men and ensuringtheir participation in follow-up interviews are clearlyneeded. It would be useful to reassess these programsseveral years later to see if drop-out rates are lower. Inaddition, future research is needed to document how sub-jects in such short-term programs were able to make sub-stantial gains.

REFERENCES

Beck, A.T., & Steer, R.A. (1987). Beck Depression inventory manual.New York: The Psychological Corporation/Harcourt Brace Jo-vanovich, Inc.

Beck, A.T., & Steer, R.A. (1988). Beck Homelessness Scale manual.New York: The Psychological Corporation/Harcourt Brace Jo-vanovich, Inc.

Brook, R., & Whitehead, P. (1980). Drug-free therapeutic communi-ties. New York: Human Sciences Press.

Condelli, W.S., & Hubbard, R.L. (1994). Client outcomes from thera-peutic communities. In F.M. Timms, G. De Leon, & N. Jainchill(Eds.), Therapeutic community: Advances in research and applica-tion (pp. 117–127). Rockville, MD: National Institute on DrugAbuse Research Monograph 144.

Cuomo, A. (1992). The way home. A new direction in social policy.New York: Report of the New York City Commission on theHomeless.

De Leon, G. (1984). The therapeutic community: Study of effectiveness.

422 H.J. Liberty et al.

Rockville, MD: National Institute on Drug Abuse, Research Mono-graph (no monograph number).

De Leon, G. (1985). The therapeutic community: Status and evolution.The International Journal of the Addictions, 20, 823–844.

De Leon, G. (1991). Center for Therapeutic Community Research,NDRI. Application to NIDA.

De Leon, G. (1994). The therapeutic community: Toward a general the-ory and model. In F.M. Timms, G. De Leon, & N. Jainchill (Eds.),Therapeutic community: Advances in research and application (pp.16–53). Rockville, MD: National Institute on Drug Abuse.

De Leon, G. (1995). Residential therapeutic communities in the main-stream: Diversity and issues. Journal of Psychoactive Drugs, 27, 3–15.

De Leon, G. (1997). Community as method: Therapeutic communitiesin special populations and special settings. Westport, CT: Praeger.

De Leon, G., Melnick, G., Kressel, D., & Jainchill, N. (1994). Circum-stances, Motivation, Readiness, and Suitability (the CMRS scales):Predicting retention in therapeutic community treatment. AmericanJournal Drug Alcohol Abuse, 20, 495–515.

De Leon, G. & Rosenthal, M.D. (1989). Treatment in residential thera-peutic communities. In T.B. Karasu (Ed.), Treatment of psychiatricdisorders Vol. II. Washington, DC: American Psychiatric Press.

De Leon, G., Wexler, H., & Jainchill, N. (1982). The therapeutic com-munity: Success and improvement rates five years after treatment.The International Journal of the Addictions, 17, 703–747.

De Leon, G., & Ziegenfuss, J. (1986). Therapeutic communities. NewYork: Charles C Thomas.

Diglio, S. (1992). Special Services For Adults Quarterly Report July–September 1992. New York: The City of New York, Human Re-sources Administration.

Fischer, P.J., & Breakey, W.R. (1991). The epidemiology of alcohol,drug, and mental disorders among homeless persons. AmericanPsychologist, 46, 1115–1128.

Garrett, G.R. (1992). Homelessness, alcohol, and other drug abuse. InP. O’Malley (Ed.), Homelessness: New England and beyond (pp.353–369). Amherst, MA: John W. McCormack Institute of PublicAffairs.

Golub, A., & Johnson, B.D. (1996). The crack epidemic: Empiricalfindings support a hypothesized diffusion of innovation process.Socio-Economic Planning Sciences, 30, 221–231.

Hubbard, R.L., Marsden, M.E., Valley, R.J., Craddock, S.G., & Ginz-burg, H.M. (1989). Drug abuse treatment: A national study of ef-fectiveness. Chapel Hill, NC: University of North Carolina Press.

Hubbard, R.L., Valley, R.J., Craddock, S.G., & Cavanaugh, E.R.(1984). Treatment outcome prospective study (TOPS): Client char-acteristics and behavior before, during, and after treatment. In F.M.Timms, & J.P. Ludford (Eds.), Drug abuse treatment evaluation:Progress and prospects (pp. 42–68). Rockville, MD: National Insti-tute on Drug Abuse, Research Monograph 51.

Johnson, B.D. (1991). Crack in New York City. Addiction and Recov-ery, May/June, 24–27.

Johnson, B.D. (1990). An interpretation of British and U.S. policies to-wards opiates and AIDS. International Working Group on AIDSand Drug Use, 5, 4–11.

Johnson, B.D. & Muffler, J.P. (1997). Sociocultural aspects of drugabuse in the 1990s. In J.H. Lowinson, P. Ruiz, R.B. Milliman, &J.G. Langrod (Ed.), Substance abuse: A comprehensive textbook(3rd ed., pp. 107–117). Baltimore: Williams & Wilkins.

Joseph, H. (1992). Sustance abuse and homelessness within the innercities. In J.H. Lowinson, P. Ruiz, R.B. Milliman, & J.G. Langrod(Ed.), Substance abuse: A comprehensive textbook (2nd ed., pp.875–889). Baltimore: Williams & Wilkins.

Link, B.G., Susser, E., Strueve, A., Phelan, J., Moore, R.E., & Struening,E.L. (1994). Lifetime and five-year prevalence of homelessness in theUnited States. American Journal of Public Health, 84, 1907–1912.

McCarthy, D., Argeriou, R.B., & Lubran, B. (1991). Alcoholism, drugabuse, and the homeless. American Psychologist, 46, 1139–1148.

Messina, M. (1997). Zenith House: A therapeutic home for the home-less. In G. DeLeon (Ed.), Community as method: Therapeutic com-

munities in special populations and special settings (pp. 37–52).Westport, CT: Praeger.

Miescher, A., & Galanter, M. (1996). Shelter-based treatment of thehomeless alcoholic. Journal of Substance Abuse Treatment, 13,135–140.

Milburn, N. (1990). Drug abuse among the homeless. In J. Momeni(Ed.), Homeless in the United States, Volume II (pp. 61–79). West-port, CT: Greenwood Press.

National Coalition for the Homeless. (1992). Addiction on the streets:Substance abuse and homelessness in America.

Office of National Drug Control Policy. (Spring, 1996). Pulse Check:National Trends in Drug Abuse.

Rosenblum, A., & Magura, S. (1996). Proposal for medical van out-reach to homeless drug users at HIV risk. Subsequently fundedgrant R01 DA 10431-01A1.

Simpson, D.D., & Sells, S.B. (1982). Effectiveness of treatment of drugabuse: An overview of the DARP research program. Advances inAlcohol and Substance Abuse, 2, 7–29.

Stahler, G.J., & Stimmel, B. (1995). The effectiveness of social inter-ventions for homeless substance abusers. New York: HaworthPress, Inc.

Takahashi, L.M. (1996). A decade of understanding homelessness inthe USA: From characterization to representation. Progress in Hu-man Geography, 20, 291–310.

Tims, F.M., De Leon, G., & Jainchill, N. (1994). Therapeutic commu-nity: Advances in research and application. Rockville, MD: Na-tional Institute on Drug Abuse, Research Monograph 144.

Tims, F.M., & Ludford, J.P. (1984). Drug abuse treatment evaluation:Strategies, progress, and prospects. Rockville, MD: National Insti-tute on Drug Abuse Research Monograph #51.

Wexler, H., Lipton, D.S., & Johnson, B.D. (1988). A Criminal-JusticeStrategy for Treating Cacaine-Heroin Abusing Offenders in Cus-tody. Issues and Practices. Washington, DC: National Institute ofJustice.

Wright, J. (1991). Correlates and consequences of alcohol abuse in the“Health Care for the Homeless” client population. Final Report to theNational Institute on Alcohol Abuse and Alcoholism, Rockville, MD.

APPENDIX A: INSTRUMENTS

Center for Therapeutic Community Research (CTCR)Locator Form—which included pretreatment address;the name, address, and phone number of each parent andall siblings; the name, address and phone number of anyfriends who might be able to locate them; previous medi-cal or drug/alcohol treatment they may have received—and if so, the names of key counselors they might haveknown there; and their preferred locations where theyliked to “hang out” when on the street. They were alsoasked an open-ended question regarding whether theycould think of any additional information that might behelpful in locating them in the future.

The Baseline Interview instrument was a slightlymodified version of the Center for Therapeutic Commu-nity Research Baseline Instrument, which had been usedto study a variety of populations in therapeutic commu-nity research. It inquired about clients’ drug use historyand other illegal activities, treatment experiences and re-ferral sources, physical health, family background, edu-cation, employment, and friendships. In January 1994,the main baseline instrument was computerized and in-terviewers administered it to clients reading questionsfrom and entering answers into laptop computers. The

Dynamic Recovery 423

quality of answers was equivalent to paper and pencil in-terviews; the actual questions could be asked much fasterbecause interviewers did not need to follow complicatedskip patterns. Moreover, staff did not need to edit ques-tionnaires, enter and verify data, nor engage in heavycleaning.

The Residency Questionnaire was developed by Dy-namic Recovery project staff to obtain information on therange of clients’ residential patterns and social networks.The instrument was introduced in December 1993, afterbaseline interviewing had begun at STAR House, andseveral revisions were made to the instrument in the fol-lowing months. The final version of the document wasintroduced on April 26, 1994. Therefore, 47.2% of theSTAR pretraining innovation residents, 144 subjects, re-ceived this questionnaire, compared to nearly all subjectsin the other three settings. The instrument included ques-tions concerning the stability of individuals’ lives priorto age 18, in terms of parental figures and housing. Theinstrument attempted to classify adult residency patternsin terms of responsibility for own housing versus doublingup with others, chronic versus episodic “homelessness,”and the type of housing and degree of social and emo-tional support clients believe to be available to them. Theterm homeless was not included in any of the questionsbecause of the possible stigma attached to the term. In-stead, a person’s status was coded as homeless based onthe respondent’s specific answer to questions about wherethey lived during the 30 days prior to entering treatment(e.g., in a park, subway station, or abandoned building).

Psychological questionnaires included the Beck De-pression Inventory (BDI); Beck & Steer, 1988; the BeckHopelessness Scale (BHS); Beck & Steer, 1987; and theCircumstances, Motivation, Readiness and Suitability

Scale (CMRS); De Leon, 1991). These were brief, paperand pencil instruments that were self-administered. TheBDI assessed the intensity of cognitive, affective, so-matic, and behavioral symptoms of depression. The BHSinstrument evaluated the extent of hopelessness that re-flected loss of motivation, feelings about the future andexpectations. The CMRS (De Leon, 1984) instrumentwas developed based on interviews with TC staff whodiscussed the dynamic factors that contributed to enter-ing and completing treatment. After many revisions, the51-item instrument included four scales designed to mea-sure: (1) circumstances: external conditions that drive aperson to seek treatment; (2) motivation: a person’s innerreasons for wanting to change; (3) readiness: a person’sself-perceived need for treatment (compared to non-treatment alternatives); and (4) suitability: the perceivedappropriateness of the match between the person and thetreatment modality of a residential TC. Subsequent todata collection, the 51-item CMRS was replaced with a42-item version. 41 of these items existed in the 51-itemversion, which the project had used. Therefore, all analy-ses are of the 41-item version to be comparable withmore recent research.

The Follow-up Questionnaire was similar to the base-line and inquired about a client’s drug use, illegal activi-ties, treatment experiences, and physical and psychologi-cal health since separating from the treatment program.At follow-up, clients also completed a Residency Ques-tionnaire that inquired about their living arrangements,employment status, and social affiliations since they sep-arated from the treatment program. Clients were alsoasked to complete the BHS and BDI. Finally, clientswere asked to provide a urine specimen for the purposeof drug testing.


Recommended