Content uploaded by Charles M Borduin
Author content
All content in this area was uploaded by Charles M Borduin
Content may be subject to copyright.
Long-Term Follow-Up to a Randomized Clinical Trial of Multisystemic
Therapy With Serious and Violent Juvenile Offenders
Cindy M. Schaeffer
University of Maryland Baltimore County
Charles M. Borduin
University of Missouri—Columbia
In this study, the authors examined the long-term criminal activity of 176 youths who had participated
in either multisystemic therapy (MST) or individual therapy (IT) in a randomized clinical trial (C. M.
Borduin et al., 1995). Arrest and incarceration data were obtained on average 13.7 (range ⫽ 10.2–15.9)
years later when participants were on average 28.8 years old. Results show that MST participants had
significantly lower recidivism rates at follow-up than did their counterparts who participated in IT (50%
vs. 81%, respectively). Moreover, MST participants had 54% fewer arrests and 57% fewer days of
confinement in adult detention facilities. This investigation represents the longest follow-up to date of a
MST clinical trial and suggests that MST is relatively effective in reducing criminal activity among
serious and violent juvenile offenders.
Keywords: treatment, juvenile offenders, multisystemic therapy
The treatment of serious and violent juvenile offenders has
become a pressing issue on the national social policy agenda,
largely because of the considerable social and economic costs
exacted by these offenders. Indeed, serious juvenile offenders are
at high risk for mental and physical health problems, substance
abuse, low educational and vocational achievement, and interper-
sonal difficulties (Laub & Sampson, 1994; Lyons, Baerger, Quig-
ley, Erlich, & Griffin, 2001). Likewise, the financial impact of
violent crime is substantial, including costs related to victimization
(i.e., health-related injuries, lost productivity, reduced quality of
life), law enforcement, and incarceration (Britt, 2000; Cohen &
Miller, 1998; Cohen, Miller, & Rossman, 1994; Robinson &
Keithley, 2000). Preventing or attenuating further criminal activity
in serious and violent juvenile offenders would favorably affect
their lives, families, and communities.
Historically, mental health and juvenile justice services have
had little success in ameliorating the serious antisocial behavior of
youths (Kazdin, 2000; Tate, Reppucci, & Mulvey, 1995). Re-
cently, however, an intensive family- and home-based treatment
(multisystemic therapy [MST]; Borduin & Henggeler, 1990;
Henggeler, Schoenwald, Borduin, Rowland, & Cunningham,
1998) has demonstrated significant effects on the criminal activity
of serious juvenile offenders in several randomized trials. For
example, with a sample of violent and chronic juvenile offenders
at imminent risk of incarceration, Henggeler and his colleagues
(Henggeler, Melton, & Smith, 1992; Henggeler, Melton, Smith,
Schoenwald, & Hanley, 1993) showed that MST reduced incar-
ceration by 64% at a 59-week follow-up and doubled the survival
rate (i.e., percentage of youths not rearrested) at a 2.4-year follow-
up. Similarly, with a sample of chronic juvenile offenders, Borduin
et al. (1995) found that MST produced a 63% reduction in rearrests
for violent and other serious crimes at a 4-year follow-up. These
and other randomized trials (e.g., Borduin, Henggeler, Blaske, &
Stein, 1990; Henggeler, Pickrel, & Brondino, 1999) suggest that
MST is a promising approach to the treatment of serious antisocial
behavior in adolescents.
Although clinical trials of MST have demonstrated significant
reductions in youths’ criminal activity during adolescence and
early adulthood, the longer term impact of MST on criminal
activity during adulthood has not been evaluated. It is important to
determine whether MST is effective in preventing longer term
criminal activity among serious juvenile offenders because such
youths are, by far, at the greatest risk for committing additional
serious crimes (Moffitt, 1993). Information regarding the possible
lasting benefits of empirically supported treatments such as MST
could greatly assist policymakers and program administrators in
selecting and implementing mental health programs for serious
juvenile offenders (Weisz, Hawley, Pilkonis, Woody, & Follette,
2000). However, if treatment effects similar to those observed at
shorter term follow-ups were not maintained over a longer period
of time, then such findings could suggest a need for refinements in
the treatment, such as providing posttreatment booster sessions or
ongoing support services in early adulthood (Weisz & Hawley,
1998).
Cindy M. Schaeffer, Department of Psychology, University of Maryland
Baltimore County; Charles M. Borduin, Department of Psychological
Sciences, University of Missouri—Columbia.
This research was supported by grants from the Missouri Department of
Social Services and the University of Missouri—Columbia Research
Council. Article preparation was supported by National Institute of Mental
Health Grant T32 MH18834. The data in this article were collected as part
of doctoral dissertation research for Cindy M. Schaeffer. We thank Robert
Perry, Gene Hamilton, Alan Sirinek, and Patrick Brown of the Missouri
13th Judicial Circuit Juvenile Court for their support. We also thank the
therapists and community professionals, too numerous to mention individ-
ually, who worked on this project. Finally, we extend special thanks to the
families who participated in this project.
Correspondence concerning this article should be addressed to Cindy M.
Schaeffer, Department of Psychology, University of Maryland Baltimore
County, 1000 Hilltop Circle, Baltimore, MD 21250, or to Charles M.
Borduin, Department of Psychological Sciences, University of Missouri—
Columbia, Columbia, MO 65211-2500. E-mail: cschaeff@umbc.edu or
borduinc@missouri.edu
Journal of Consulting and Clinical Psychology Copyright 2005 by the American Psychological Association
2005, Vol. 73, No. 3, 445–453 0022-006X/05/$12.00 DOI: 10.1037/0022-006X.73.3.445
445
In the current study from the Missouri Delinquency Project, we
examine long-term criminal activity for 176 serious and violent
juvenile offenders who participated in an MST clinical trial an
average of 13.7 years earlier (Borduin et al., 1995). As such, the
study extends the follow-up period for participants in the largest
clinical trial of MST to date. The study provides information
regarding the long-term efficacy of MST across a range of criminal
recidivism outcomes, including (a) number of adult arrests, (b)
number of days sentenced for confinement in the adult court
system, and (c) number of days sentenced to adult probation.
Method
Design
In the current study, we track the long-term criminal activity of 176
serious adolescent offenders who received either MST or individual ther-
apy (IT) 11.8–15.2 years (M ⫽ 13.7 years) earlier in a randomized clinical
trial (Borduin et al., 1995). The original study used a pretest–posttest
control group design, with random assignment to conditions and a 4-year
follow-up for rearrests, to compare the effectiveness of MST versus IT.
Because this sample has been described extensively elsewhere (Borduin et
al., 1995), a shorter description of the participants and therapeutic inter-
ventions is provided here.
Participants
Participants were 176 adolescent offenders (12–17 years of age) and
their families who had been referred consecutively to the Missouri Delin-
quency Project by juvenile court personnel from July 1983 to October
1986. Referrals to the project included all families in which the youth (a)
had at least two arrests, (b) was currently living with at least one parent
figure, and (c) showed no evidence of psychosis or dementia. Following
agreement to participate in the clinical trial, each family was randomly
assigned (simple randomization with a coin toss) by a court administrator
to receive either MST (n ⫽ 92) or IT (n ⫽ 84). Of the 176 families, 140
(79.5%) completed treatment (hereafter referred to as completers) and 36
(21.5%) dropped out, defined as unilaterally terminating after the first
session (with the youth or family) and before the seventh session. Of the 36
youths and their families who dropped out of treatment (hereafter referred
to as dropouts), 15 were from the MST condition, and 21 were from the IT
condition (dropout rates for MST [16.3%] and IT [25.0%] were not
significantly different). In the current study, treatment completers and
treatment dropouts were collapsed in each condition to provide a conser-
vative test of MST effects.
The arrest histories of the referred youths attest to their serious criminal
involvement. The youths averaged 3.9 previous arrests for felonies (SD ⫽
1.9), and 47.8% of the youths had at least one arrest for a violent crime
(e.g., sexual assault, assault and battery with intent to kill, aggravated
assault). Moreover, all of the youths had been incarcerated previously for
at least 4 weeks. The mean age of the youths at the time of their first arrest
was 11.7 years (SD ⫽ 1.9) and at the time of treatment was 14.5 years
(SD ⫽ 1.4); 69.3% were boys, 30.7% were girls; 76.1% were White, 22.2%
were African American, 1.1% were Asian American, and 0.9% were
Hispanic. In addition, 56.8% lived with two parental figures (biological
parents, stepparents, foster parents, grandparents). The primary caretaker
of the youths included biological mothers (89.5%); step-, foster, or adop-
tive mothers (5.5%); other female relatives (2.5%); or biological fathers
(2.5%). Families averaged 3.2 children (SD ⫽ 1.9), and 63.4% of the
families were of lower socioeconomic status (SES; Class IV or V; Holl-
ingshead, 1975). The average age of participants at the time of follow-up
was 28.8 years (SD ⫽ 1.8).
Treatment Conditions
The mean numbers of hours of treatment were 20.7 (SD ⫽ 7.4) for the
MST group and 22.5 (SD ⫽ 10.6) for the IT group. These means were not
significantly different, F(1, 175) ⫽ 1.85, p ⫽ .176.
MST. Therapeutic interventions were based on the multisystemic ap-
proach to the treatment and prevention of behavior problems in children
and adolescents (Henggeler et al., 1998). The treatment and prevention
emphases of MST fit closely with findings on the correlates and causes of
serious delinquent behavior (for a review, see Loeber & Farrington’s, 1998,
study). With interventions that are present focused and action oriented,
MST directly addresses intrapersonal (e.g., cognitive) and systemic (i.e.,
family, peer, school) factors that are known to be associated with adoles-
cent antisocial behavior. Moreover, because different combinations of
these factors are relevant for different adolescents, MST interventions are
individualized and highly flexible. Guidelines for designing and imple-
menting MST interventions with antisocial adolescents and their families
are described in detail elsewhere (Borduin & Henggeler, 1990; Henggeler
et al., 1998).
MST uses a home- and community-based model of service delivery. To
promote cooperation and enhance generalization, therapists usually held
sessions in the family’s home and in community locations (e.g., school,
recreation center) at a convenient time. In addition, services were time
limited, with an overriding goal of empowering parents with the skills and
resources needed to independently address the inevitable difficulties that
arise in raising adolescents.
IT. The therapy provided in this condition was selected to represent the
usual community treatment for juvenile offenders in this judicial district
and perhaps in many other judicial districts as well (see the following
studies: Loeber & Farrington, 1998; Office for Juvenile Justice and De-
linquency Prevention, 1993). All of the offenders in this condition received
IT that focused on personal, family, and academic issues. The therapists
offered support, feedback, and encouragement for behavior change. Their
theoretical orientations were an eclectic blend of psychodynamic (e.g.,
promoting insight and expression of feelings), client-centered (e.g., build-
ing a close relationship, providing empathy and warmth), and behavioral
(e.g., providing social approval for school attendance and other positive
behaviors) approaches. Although there were some variations in the treat-
ment strategies used by the therapists (e.g., some therapists provided less
empathy or were more directive than other therapists), the common thread
of the approaches was that the interventions focused on the individual
adolescent rather than on the systems in which the adolescent was
embedded.
Therapists
MST was provided by three female and three male graduate students
(ages ranged from 23 to 31 years; M ⫽ 26) in clinical psychology. Each
had approximately 1.5 years of direct clinical experience with children or
adolescents before the study. Therapist supervision was provided by
Charles M. Borduin in a 3-hr weekly group meeting and continued
throughout the course of the investigation. During these meetings, the
therapists and supervisor reviewed the goals and progress of each case,
observed and discussed selected videotaped or audiotaped therapy sessions,
and made decisions about how best to facilitate the family’s progress.
Interventions in the IT group were provided by three female and three
male therapists (ages ranged from 25 to 33 years; M ⫽ 28) at local mental
health outpatient agencies, including the treatment services branch of the
juvenile court. Each therapist had a master’s degree (or equivalent training)
in either counseling psychology, social work, or another mental health-
related field and had approximately 4 years of direct clinical experience
with adolescents. These therapists attended 2.5-hr weekly case reviews
with the treatment coordinator from the juvenile court to discuss the goals
and progress of each case.
446
SCHAEFFER AND BORDUIN
Treatment Integrity
To sustain the integrity of MST, therapists documented each therapeutic
contact by summarizing what transpired and how much progress had been
made in meeting the goals of treatment; ongoing clinical supervision and
feedback were provided throughout the investigation. To monitor the
integrity of IT, therapists were required to provide monthly reports that
summarized the nature of therapeutic contacts, who was present at the
contacts, and adolescent progress in meeting the goals of treatment. The
project director (Charles M. Borduin) met periodically with the therapists
to review selected videotapes of sessions and to ensure that the therapists
adhered to their stated treatment plans. Adherence to treatment plans also
was promoted by the juvenile court treatment coordinator, who met weekly
with the therapists in the IT condition. Although it was not possible to
include an independent assessment of the integrity of either MST or IT, the
therapists in both conditions completed a checklist for each of their cases
to indicate the systems directly addressed during the course of treatment
(i.e., individual, marital, family, peer, school). These checklists revealed
that all MST cases received interventions in two or more systems (M ⫽
3.5), whereas the vast majority (90.5%) of IT cases received interventions
in only one system (always the individual adolescent).
Research Procedures
Original outcome study. All families who were referred to the project
were initially contacted by phone or a home visit by a research assistant. It
was emphasized that the family’s participation in research was voluntary
and that refusal to participate (or exercising the right to discontinue
participation at any time) would not jeopardize the receipt of treatment
services or result in any sanctions from the court. Families also were
informed that the youth’s juvenile arrest records would be obtained from
the juvenile court through the youth’s 17th birthday and that adult arrest
records would be obtained on the youth thereafter. Youths remained under
the jurisdiction of the court regardless of their families’ decisions about
participating in the research assessments or in treatment. Family members
provided written consent or assent for the research procedures. All proce-
dures were approved by the Institutional Review Board of the University of
Missouri—Columbia.
For the original outcome study, all families completed extensive pre- and
posttreatment assessment batteries of self-report instruments, behavior
rating inventories, and observational tasks that measured individual, fam-
ily, peer, and school functioning (see Borduin et al.’s, 1995, study). Each
assessment session was scheduled at the family’s convenience, either in
their home (91% of sessions) or at a youth center in their neighborhood.
Data collected in these sessions were not included in the current study.
Only those procedures and measures relevant to the current study are
described below.
Current study. Participants’ juvenile and adult criminal records were
obtained within the state of Missouri. A broader search of criminal records
in other states was not possible because fingerprints would have been
required to conduct a national criminal records check, and these were not
obtained at the time of the original study. Nevertheless, we assumed that
arrest rates for those participants residing outside of Missouri were not
systematically different from those remaining in the state. We also assumed
that variation between treatment groups in arrest rates would be consistent
whether the participant resided within or outside of Missouri.
It was necessary to determine whether each participant had lived in
Missouri since the previous follow-up assessment (conducted approxi-
mately 10 years earlier; Borduin et al., 1995) and, thus, whether he or she
was available to have an arrest record in the state after that time. Accord-
ingly, several steps were followed to confirm state residency. First, state
arrest records were searched, and crimes committed in the state after the
previous follow-up with this sample were noted. Next, for those individuals
whose names did not appear in the state criminal registry, a search of state
driver’s license records was conducted. An individual was considered to
have resided in the state during the follow-up period if he or she held a
Missouri driver’s license. Finally, we used original phone numbers and
addresses of parents to confirm residence in the state of several additional
youths for whom there were no arrest records or driver’s license records.
Overall, 93.8% (n ⫽ 165) of the sample was located and determined to
have lived in the state during the follow-up period. The number and
percentage of youths found in each group were as follows: MST com-
pleters (n ⫽ 73, or 94.8%), MST dropouts (n ⫽ 14, or 93.3%), IT
completers (n ⫽ 59, or 93.7%), and IT dropouts (n ⫽ 19, or 90.5%). Those
youths (n ⫽ 11) who could not be verified to live in the state were
considered lost to long-term follow-up (see Figure 1).
Juvenile and adult arrest data that had been collected for this sample
previously (Borduin et al., 1995) were included in survival analyses,
resulting in at least partial data for those youths lost to long-term follow-up.
Thus, each youth’s follow-up period was anchored by the point of release
from probation (i.e., within 2 weeks of treatment termination for com-
pleters and an average of 6 months from the time of referral for dropouts)
and was considered to run through the latest date for which the youth could
be confirmed to live in the state.
Measures of Criminal Activity
Both juvenile and adult criminal records were obtained for this study. A
condition of the original clinical trial was that the juvenile court would
provide access to juvenile arrest records through consenting participants’
18th birthdays. Youths’ criminal arrest data were obtained yearly from
juvenile office records by research assistants who were uninformed as to
each participant’s treatment condition. Adult criminal arrest data, which in
the state of Missouri are available to the public, were obtained from a
computerized database by a state police employee (also uninformed as to
treatment condition) who conducted a search by participant names; these
records were searched at the time of the original outcome study and again
for the current study. Dates of all juvenile and adult arrests were recorded
to ensure that arrest information was not duplicated. For both juvenile and
adult arrest records, only substantiated arrests for index offenses were
included in the data set (i.e., charges that were dismissed at trial were
excluded). The average length of the follow-up period for those youths
confirmed to have been available for arrest was 13.7 years (range ⫽
11.8–15.2 years; SD ⫽ 1.2 years).
Each arrest was coded as having taken place during the follow-up period
if it occurred after the date of the posttreatment assessment (or, if a
posttreatment assessment was not completed, then after the date of termi-
nation from MST or IT). In addition, each arrest was classified as either a
nonviolent (e.g., breaking and entering, theft), violent (e.g., assault, rape),
drug related (e.g., possession of marijuana, distribution of cocaine), or
other (e.g., violation of probation, failure to pay child support) offense;
Figure 1. Flow diagram of participants from randomization to follow-up.
447
LONG-TERM FOLLOW-UP OF MST
these categories were mutually exclusive. Juvenile and adult arrest data
were combined in analyses to provide a complete record of all arrests (i.e.,
number and type) during the follow-up period.
In addition to obtaining information pertaining to the date and reason for
each juvenile and adult offense, information about punitive sentencing also
was obtained for each adult offense. The total number of days that a
participant convicted for an adult crime was sentenced to serve in an adult
correctional facility composed the adult confinement variable. Similarly,
the total number of days that a participant convicted of an adult crime was
sentenced to a term of probation composed the adult probation variable. A
sentence of probation was given as an alternative to incarceration for less
serious crimes and required the individual to adhere to a number of
stipulations, such as abstaining from criminal activity, holding a regular
job, and meeting frequently with court personnel; a longer probation
sentence reflected a relatively more severe crime than did a shorter sen-
tence. These variables reflected days sentenced to either confinement or
probation, not days served; participants in the current study may have
served fewer days than actually sentenced, given that most convicted
persons do not serve the entire length of their original sentences (Hughes,
James-Wilson, & Beck, 2001). In addition, because sentencing was done
prospectively, the length of some participants’ sentences exceeded the
length of the follow-up period for the current study. Differences in the
length of participants’ follow-up periods were controlled for in all analyses.
Results
Likelihood and Relative Odds for Rearrest
As noted, treatment completers and treatment dropouts were
collapsed in each condition.
1
At 13.7 years of follow-up, the
overall recidivism rate for the MST group (50%) was significantly
lower than the overall rate for the IT group (81.0%),
2
(1, N ⫽
176) ⫽ 18.45, p ⬍ .0001. Between-groups differences in recidi-
vism rates for various types of offenses are described in Table 1.
To describe the relative risk of arrest in the IT group versus the
MST group, we conducted binary logistic regressions. Youths in
the IT condition were 4.25 times more likely than youths in the
MST condition to be rearrested (95% confidence interval [CI] ⫽
2.15–8.40, p ⬍ .001) during the follow-up period. More specifi-
cally, youths in the IT condition were 2.57 times more likely to
have an arrest for a violent offense (95% CI ⫽ 1.22–5.45, p ⬍ .01),
2.63 times more likely to have an arrest for a nonviolent offense
(95% CI ⫽ 1.43–4.84, p ⬍ .01), and 3.33 times more likely to
have an arrest for a drug offense (95% CI ⫽ 1.56–7.11, p ⬍ .01).
Survival Functions
Survival analysis (based on the SURVIVAL procedure; SPSS
Version 11.0.1 for Windows; SPSS, 2001) was used to obtain the
cumulative survival functions (or survival curves) for participants
who were randomly assigned to the MST (n ⫽ 92) or IT groups
(n ⫽ 84), whose average follow-up periods were 4,898.7 and
5,102.3 days, respectively. The cumulative survival function rep-
resents the proportion of participants who survived any type of
arrest (i.e., were not arrested) in each group by the length of time
(in days) from release from treatment (or juvenile probation for
treatment dropouts). A log-rank test (with the Kaplan–Meier esti-
mator; Kaplan & Meier, 1958) revealed that the survival functions
for the two groups were significantly different,
2
(1, N ⫽ 176) ⫽
7.92, p ⬍ .01. As depicted in Figure 2, MST participants were at
lower risk of rearrest (i.e., more likely to survive) during follow-up
than were IT participants. By the end of 13.7 years (5,007.2 days),
81.0% of the participants in the IT group had been arrested at least
once, compared with 50.0% of the participants in the MST group.
To determine an effect size for this survival function, we per-
formed a Cox proportional hazards regression (Cox, 1972). The
hazards ratio for treatment condition (MST or IT; p ⫽ .002) was
.576, suggesting a medium effect size for the lower risk of rearrest
observed for MST participants.
Another set of survival analyses was conducted to examine
between-groups differences on time to first arrest for various types
of offenses. As depicted in Figures 3, 4, and 5, respectively,
participants in the MST group were at lower risk of arrest for
violent offenses,
2
(1, N ⫽ 176) ⫽ 4.88, p ⬍ .05; nonviolent
offenses,
2
(1, N ⫽ 176) ⫽ 6.27, p ⬍ .05; and drug offenses,
2
(1,
N ⫽ 176) ⫽ 7.97, p ⬍ .01, during follow-up than were participants
in the IT group. Cox proportional hazards ratio tests of these
survival functions suggested medium to large effect sizes for MST
versus IT (violent offenses, p ⫽ .011,

⫽ .844; nonviolent
offenses, p ⫽ .009,

⫽ .572; and drug offenses, p ⫽ .010,

⫽
.864).
Number of Posttreatment Arrests and Days Sentenced to
Incarceration and Probation
Additional analyses examined the number of posttreatment ar-
rests, adult incarceration days, and adult probation days during the
follow-up period among youths in the MST and IT groups. Be-
cause these recidivism variables are continuous and nonnormal,
they can be considered censored-dependent variables (Greene,
1993), containing both a qualitative (e.g., arrested vs. not arrested)
and a quantitative (e.g., number of arrests among recidivists)
component. Accordingly, Tobit regression analyses (based on the
STATA 6.0 statistical package; StataCorp, 1999), which use
maximum-likelihood estimation procedures and a corresponding
chi-square test statistic, were used to examine differences between
1
In the current study, we collapsed across completers and dropouts for
each treatment group. However, it is informative to note that the recidivism
rates for MST completers and MST dropouts were 45.5% and 73.3%,
respectively, and for IT completers and dropouts were 79.4% and 85.7%,
respectively.
Table 1
Likelihood of Rearrest for Different Types of Crimes by Therapy
Condition
Type of posttreatment offense %
2
(1)
p
Any offense 18.44 .001
IT 81.0
MST 50.0
Any violent offense 6.34 .012
IT 29.8
MST 14.1
Any nonviolent offense 9.86 .002
IT 59.5
MST 35.9
Any drug offense 10.29 .001
IT 33.3
MST 13.0
Note. Sample sizes for therapy conditions are as follows: individual
therapy (IT; n ⫽ 84); multisystemic therapy (MST; n ⫽ 92).
448
SCHAEFFER AND BORDUIN
groups on these measures.
2
For these analyses, only those individ
-
uals who were available for long-term follow-up were included. In
addition, variation in the length of the follow-up period was
controlled for in all analyses by entering the number of follow-up
days into the regression equations.
As shown in Table 2, MST participants had half as many
rearrests and were sentenced to less than half as many days of
incarceration as their IT counterparts. MST participants also had
significantly fewer arrests for violent, nonviolent, or drug-related
offenses. Although MST participants were sentenced to 43% fewer
days of adult probation than were IT participants, this difference
only approached statistical significance (p ⫽ .08).
Potential Moderators of Criminal Recidivism in MST
Tobit regression analyses were used to evaluate the effects of
potential moderators (age, race, SES, gender, pretreatment arrest
for a violent crime, and number of pretreatment arrests) of MST
effectiveness. The dependent variables were number of posttreat-
ment arrests, confinement days, and probation days. For each of
the regression analyses, a dummy variable that represented treat-
ment group (which collapsed across treatment completers and
dropouts in each group), the moderating variable, and the cross-
product term of the treatment group and the moderating variable
(which controlled for the length of the follow-up period) were
entered simultaneously. Moderator variables that were continuous
(age, SES, and pretreatment arrests) were centered around their
means in each cross-product term. A significant coefficient for the
cross-product term indicated whether MST was differentially ef-
fective with youths and families from different backgrounds. In no
case was the cross-product term significant. Although these results
generally suggest that MST was equally effective with youths of
different backgrounds, it should be noted that some of the variables
that were tested for moderation included relatively few participants
in certain subgroups. For example, there were small numbers of
girls and minority youths in each treatment condition. Thus, any
conclusions about moderators of MST effectiveness should be
considered tentative.
3
Discussion
The current study represents the longest follow-up to date of a
MST clinical trial with serious juvenile offenders. The results
indicated that MST participants were significantly less likely to be
arrested than were than IT participants (50% vs. 81%, respec-
2
The use of ordinary least squares regression is inappropriate for
censored-dependent variables because the relationship between an inde-
pendent variable and a censored-dependent variable is inherently nonlinear
(Greene, 1993; Kinsey, 1981; Tobit, 1958). Common data transformation
techniques also are inappropriate with such variables. Making a log trans-
formation of the dependent variable does not eliminate the large degree of
clustering at the lower limit and does not adequately capture the qualitative
difference between a zero and a nonzero value. Another possible transfor-
mation, dichotomizing the dependent variable and using logistic or probit
regression analysis, also is inefficient because it ignores important infor-
mation on the extent of criminal activity. To avoid these transformation
problems, researchers have applied Tobit regression to many types of
nonnormal variables, including measures of substance abuse (e.g., Frone,
Cooper, & Russell, 1994), life events (e.g., Fraser, Jenson, Kiefer, &
Popuang, 1994), and health status (e.g., Grootendorst, 2000).
3
A table of the main effects of these moderators (i.e., independent of
treatment condition) on outcome variables is available from Cindy M.
Schaeffer on request.
Figure 2. Survival functions for multisystemic therapy (MST) and individual therapy (IT) groups on time to
any first arrest following treatment. Completers and dropouts are combined in each group.
449
LONG-TERM FOLLOW-UP OF MST
tively) within 13.7 years after treatment termination. In addition,
the odds of rearrest for drug, violent, or nonviolent offenses were
2–4 times lower for MST participants than for IT participants.
Moreover, MST participants were sentenced to 61% fewer days of
confinement in adult detention facilities and to 37% fewer days of
probation as adults than were comparison counterparts.
The results extend those of a previous follow-up study with this
sample in which only 26.1% of the MST participants had been
Figure 3. Survival functions for multisystemic therapy (MST) and individual therapy (IT) groups on time to
first violent arrest following treatment. Completers and dropouts are combined in each group.
Figure 4. Survival functions for multisystemic therapy (MST) and individual therapy (IT) groups on time to
first nonviolent arrest following treatment. Completers and dropouts are combined in each group.
450
SCHAEFFER AND BORDUIN
rearrested 4 years following treatment (Borduin et al., 1995). The
longer follow-up in the current study showed that an additional
23.9% of the MST participants had been arrested since the previ-
ous follow-up, for an overall recidivism rate of 50.0%. Thus, for
some individuals, the short-term preventive effects of MST on
criminal activity did not persist into adulthood. The findings also
highlight the need for longer term follow-ups of treatment out-
comes for serious juvenile offenders, even for a treatment as
successful as MST. We hope to conduct a more comprehensive
follow-up with the current sample to determine why some youths
desisted from offending, whereas others did not.
Although the childhood behavioral histories of the serious ju-
venile offenders in the current study are not known, it is possible
that this sample of youths is representative of those life-course-
persistent offenders (see Moffitt’s, 1993, study) about whom poli-
cymakers and researchers are most concerned. Indeed, the youths
in the current study were quite young at the time of their first arrest
(M ⫽ 11.7 years of age) and had been arrested an average of three
more times by the age of 14 years. Moreover, the vast majority of
youths (81%) in the IT condition were rearrested at some point
during the follow-up period, which lasted well into adulthood
(participants’ average age at follow-up was 28.8 years). Other
research has suggested that youths who are involved in serious
antisocial behavior throughout the life course, although estimated
to represent only 15%–25% of all juvenile offenders, account for
more than half of the total volume of youth crime in a community
(Farrington, Ohlin, & Wilson, 1986; Loeber & Farrington, 1998).
To our knowledge, the current study is the first to demonstrate the
efficacy of MST in altering the criminal trajectories of serious
juvenile offenders beyond early adulthood.
The findings also have implications for the cost savings and
fiscal viability of MST. For more than 13 years of follow-up, MST
reduced the rate of incarceration by an average of 62.4 days per
year. A recent study that examined cost effectiveness with this
sample indicated a cost savings of at least $50,000 for each youth
receiving MST (Merbler, Borduin, & Schaeffer, 2004). These
savings compare favorably with MST program costs of $4,000–
$6,000 per youth (Henggeler, Schoenwald, Rowland, & Cunning-
ham, 2002). To conclude with greater confidence that the MST
program in the current study was less costly than IT, however, we
need a comprehensive examination of service use across service
sectors (e.g., social welfare, mental health, juvenile justice, pri-
mary care) to more fully explicate the types of services received by
the youths and to explore the possibility of cost shifting. Never-
theless, a cost analysis of MST with substance abusing or depen-
dent delinquents showed no evidence of cost shifting (Schoenwald,
Ward, Henggeler, Pickrel, & Patel, 1996).
The relative efficacy of MST in reducing criminal activity in
high-risk youths may be due in part to the match between MST
intervention foci and empirically identified determinants of crim-
inality and violence in youths, including behavior problems, pa-
rental disturbance, problematic family relations, association with
deviant peers, and poor school performance (Borduin, 1999). In-
deed, a recent study that examined mechanisms of change in MST
with serious juvenile offenders found that improvements in family
functioning predicted decreases in delinquent peer affiliations that,
in turn, predicted decreases in delinquency at a 1.7-year follow-up
(Huey, Henggeler, Brondino, & Pickrel, 2000). Perhaps the longer
term effects of MST observed in the current study also are linked
with treatment-related changes in family and peer relations. More
specifically, improved family support and decreased deviant peer
involvement may have allowed MST participants to experience
increased success in accomplishing educational, occupational, and
other important developmental tasks (e.g., the formation of healthy
Figure 5. Survival functions for multisystemic therapy (MST) and individual therapy (IT) groups on time to
first drug-related arrest following treatment. Completers and dropouts are combined in each group.
451
LONG-TERM FOLLOW-UP OF MST
romantic relationships) during late adolescence and early adult-
hood. Additional research is needed to evaluate the mechanisms of
long-term change in MST.
Another important aspect of MST pertains to the accessibility
and ecological validity of services. Traditionally, mental health
services for juvenile offenders either have been inaccessible (i.e.,
office based) or have provided interventions that have little bearing
on the natural ecology of youths (e.g., residential treatment cen-
ters, incarceration). In contrast, by having the family preservation
model of service delivery, MST is provided in natural community
contexts (e.g., home, school, recreation center). The delivery of
services in the natural ecology of youths has several advantages,
including the promotion of family cooperation and the acquisition
of more accurate data regarding the assessment of identified prob-
lems and the results of treatment interventions (Borduin &
Henggeler, 1990). Indeed, there is a growing consensus that pro-
viders of children’s mental health services should recognize the
natural ecology of the child and diminish barriers to service access
(e.g., Burns & Hoagwood, 2002; Cauce et al., 2002; Henggeler,
1994; Roberts, 1994; Stroul, 1996).
As described earlier, we are contending that the favorable results
of this study were largely due to two crucial aspects of MST: its
comprehensive nature and ecologically valid delivery. However, it
must be noted that this study still has several methodological
limitations. First, we assessed criminal activity during the
follow-up period using arrest records, which are believed to un-
derestimate the actual number of crimes committed by serious
offenders (Loeber & Farrington, 1998). Nevertheless, arrest
records are one useful index of criminal involvement and likely
provided an accurate estimate of the relative effectiveness of MST
versus IT in reducing serious criminal activity. Second, we were
unable to confirm that youths maintained continuous residence in
Missouri throughout the follow-up period and cannot rule out the
possibility that a portion of youths may have committed crimes in
other states. However, it seems unlikely that length of residency in
the state would vary systematically across treatment conditions.
Moreover, at least partial criminal recidivism data were available
for the entire sample, and complete follow-up data were available
for the vast majority (91.5%) of the sample.
In summary, the current findings further support the efficacy
and applicability of MST for a broad range of serious antisocial
behaviors. Over the longest follow-up period ever examined in a
MST clinical trial, criminal recidivism among MST participants
was half that of their counterparts receiving IT. In addition to
improved life outcomes for youths and families receiving MST,
societal benefits of MST include reduced incidents of crime in
communities and enormous public savings in incarceration and
probation costs (Borduin, Schaeffer, & Ronis, 2003). The cost
effectiveness of MST, as well as its clinical effectiveness, should
be considered by policymakers and the public at large in the
selection of interventions for serious juvenile offenders.
References
Borduin, C. M. (1999). Multisystemic treatment of criminality and vio-
lence in adolescents. Journal of the American Academy of Child and
Adolescent Psychiatry, 38, 242–249.
Borduin, C. M., & Henggeler, S. W. (1990). A multisystemic approach to
the treatment of serious delinquent behavior. In R. J. McMahon & R. D.
Peters (Eds.), Behavior disorders of adolescence: Research, interven-
tion, and policy in clinical and school settings (pp. 63–80). New York:
Plenum Press.
Borduin, C. M., Henggeler, S. W., Blaske, D. M., & Stein, R. (1990).
Multisystemic treatment of adolescent sexual offenders. International
Journal of Offender Therapy and Comparative Criminology, 34, 105–
113.
Borduin, C. M., Mann, B. J., Cone, L. T., Henggeler, S. W., Fucci, B. R.,
Blaske, D. M., & Williams, R. A. (1995). Multisystemic treatment of
serious juvenile offenders: Long-term prevention of criminality and
violence. Journal of Consulting and Clinical Psychology, 63, 569–578.
Borduin, C. M., Schaeffer, C. M., & Ronis, S. T. (2003). Multisystemic
treatment of serious antisocial behavior in adolescents. In C. A. Essau
(Ed.), Conduct and oppositional disorders in children and adolescents:
Epidemiology, risk factors, and treatment (pp. 299–318). Mahwah, NJ:
Erlbaum.
Britt, C. L. (2000). Health consequences of criminal victimization. Inter-
national Review of Victimology, 8, 63–73.
Burns, B. J., & Hoagwood, K. (Eds.). (2002). Community treatment for
youth: Evidenced-based interventions for severe emotional and behav-
ioral disorders. New York: Oxford University Press.
Cauce, A. M., Domenech-Rodriguez, M., Paradise, M., Cochran, B. N.,
Shea, J. M., Srebnik, D., & Baydar, N. (2002). Cultural and contextual
influences in mental help health seeking: A focus on ethnic minority
youth. Journal of Consulting and Clinical Psychology, 70, 44–55.
Cohen, M. A., & Miller, T. R. (1998). The cost of mental health care for
victims of crime. Journal of Interpersonal Violence, 13, 93–110.
Cohen, M. A., Miller, T. R., & Rossman, S. B. (1994). The costs and
Table 2
Main Effects of Treatment Group on Criminal Recidivism
Outcomes at 13.7-Year Follow-Up
Outcome
Group
2
(1)
Multisystemic
therapy
Individual
therapy
No. of all posttreatment
offenses
17.83***
M 1.82 3.96
SD 3.24 4.30
No. of posttreatment
violent offenses
169.92***
M 0.21 0.51
SD 0.57 1.08
No. of posttreatment
nonviolent offenses
11.17***
M 0.89 2.00
SD 2.12 2.71
No. of posttreatment drug-
related offenses
11.16**
M 0.20 0.55
SD 0.71 1.12
Days sentenced to adult
confinement
8.97**
M 582.25 1,356.53
SD 2,843.84 3,120.10
Days sentenced to adult
probation
3.02†
M 420.97 738.71
SD 945.72 1,353.19
Note. For all analyses, only those youths with complete 13.7-year
follow-up data were included (i.e., individual therapy, n ⫽ 78; multisys-
temic therapy, n ⫽ 87), and completers and dropouts within each treatment
condition were combined. Chi-square test statistics reflect Tobit regression
analyses. All analyses controlled for each individual’s length of the
follow-up period.
† p ⫽ .08. ** p ⬍ .01. *** p ⬍ .001.
452
SCHAEFFER AND BORDUIN
consequences of violent behavior in the United States. In A. J. Reiss, Jr.
& J. A. Roth (Eds.), Understanding and preventing violence: Conse-
quences and control (Vol. 4, pp. 67–166). Washington, DC: National
Academy Press.
Cox, D. R. (1972). Regression models and life tables. Journal of the Royal
Statistical Society, 34, 187–220.
Farrington, D. P., Ohlin, L., & Wilson, J. Q. (1986). Understanding and
controlling crime. New York: Springer-Verlag.
Fraser, M. W., Jenson, J. M., Kiefer, D., & Popuang, C. (1994). Statistical
methods for the analysis of critical life events. Social Work Research,
18, 163–177.
Frone, M. R., Cooper, M. L., & Russell, M. (1994). Stressful life events,
gender, and substance use: An application of Tobit regression. Psychol-
ogy of Addictive Behaviors, 8, 59–69.
Greene, W. H. (1993). Econometric analysis (2nd ed.). New York: Mac-
millan.
Grootendorst, P. (2000). Censoring in statistical models of health status:
What happens when one can do better than “1.” Quality of Life Re-
search: An International Journal of Quality of Life Aspects of Treatment,
Care, and Rehabilitation, 9, 911–914.
Henggeler, S. W. (1994). A consensus: Introduction to the APA Task Force
report on innovative models of treatment and service delivery for chil-
dren, adolescents, and their families. Journal of Clinical Child Psychol-
ogy, 23, 3–6.
Henggeler, S. W., Melton, G. B., & Smith, L. A. (1992). Family preser-
vation using multisystemic therapy: An effective alternative to incarcer-
ating serious juvenile offenders. Journal of Consulting and Clinical
Psychology, 60, 821–833.
Henggeler, S. W., Melton, G. B., Smith, L. A., Schoenwald, S. K., &
Hanley, J. H. (1993). Family preservation using multisystemic treat-
ment: Long-term follow-up to a clinical trial with serious juvenile
offenders. Journal of Child and Family Studies, 2, 283–293.
Henggeler, S. W., Pickrel, S. G., & Brondino, M. J. (1999). Multisystemic
treatment of substance abusing and dependent delinquents: Outcomes,
treatment fidelity, and transportability. Mental Health Services Re-
search, 1, 171–184.
Henggeler, S. W., Schoenwald, S. K., Borduin, C. M., Rowland, M. D., &
Cunningham, P. B. (1998). Multisystemic treatment of antisocial behav-
ior in children and adolescents. New York: Guilford Press.
Henggeler, S. W., Schoenwald, S. K., Rowland, M. D., & Cunningham,
P. B. (2002). Serious emotional disturbance in children and adolescents:
Multisystemic therapy. New York: Guilford Press.
Hollingshead, A. B. (1975). The four-factor index of social status. Unpub-
lished manuscript, Yale University, New Haven, CT.
Huey, S. J., Henggeler, S. W., Brondino, M. J., & Pickrel, S. G. (2000).
Mechanisms of change in multisystemic therapy: Reducing delinquent
behavior through therapist adherence and improved family and peer
functioning. Journal of Consulting and Clinical Psychology, 68, 451–
467.
Hughes, T. A., James-Wilson, D., & Beck, A. J. (2001). Trends in state
parole, 1990–2000. Washington, DC: U.S. Department of Justice, Bu-
reau of Justice Statistics.
Kaplan, E. L., & Meier, P. (1958). Nonparametric estimation from incom-
plete observations. Journal of the American Statistical Association, 53,
457–481.
Kazdin, A. E. (2000). Treatments for aggressive and antisocial children.
Child and Adolescent Psychiatric Clinics of North America, 9, 841–858.
Kinsey, J. (1981). Determinants of credit card accounts: An application of
Tobit analysis. Journal of Consumer Research, 8, 172–182.
Laub, J. H., & Sampson, R. J. (1994). Unemployment, marital discord, and
deviant behavior: The long-term correlates of childhood misbehavior. In
T. Hirschi & M. R. Gottfredson (Eds.), The generality of deviance (pp.
235–252). New Brunswick, NJ: Transaction Publishers.
Loeber, R., & Farrington, D. P. (1998). Serious and violent juvenile
offenders: Risk factors and successful interventions. Thousand Oaks,
CA: Sage.
Lyons, J. S., Baerger, D. R., Quigley, P., Erlich, J., & Griffin, E. (2001).
Mental health service needs of juvenile offenders: A comparison of
detention, incarceration, and treatment settings. Children’s Services:
Social Policy, Research, and Practice, 4, 69–85.
Merbler, S. J., Borduin, C. M., & Schaeffer, C. M. (2004). Cost-benefit
analysis of multisystemic therapy with serious and violent juvenile
offenders. Unpublished manuscript.
Moffitt, T. E. (1993). Adolescent-limited and life-course persistent antiso-
cial behavior: A developmental taxonomy. Psychological Review, 100,
674–701.
Office of Juvenile Justice and Delinquency Prevention. (1993). Guide for
implementing the comprehensive strategy for serious, violent, and
chronic juvenile offenders. Washington, DC: U.S. Department of Justice.
Roberts, M. C. (1994). Models for service delivery in children’s mental
health: Common characteristics. Journal of Clinical Child Psychology,
23, 212–219.
Robinson, F., & Keithley, J. (2000). The impacts of crime on health and
health services: A literature review. Health, Risk, and Society, 2, 253–
266.
Schoenwald, S. K., Ward, D. M., Henggeler, S. W., Pickrel, S. G., & Patel,
H. (1996). MST treatment of substance abusing or dependent adolescent
offenders: Costs of reducing incarceration, inpatient, and residential
placement. Journal of Child and Family Studies, 4, 431–444.
SPSS. (2001). SPSS Base 11.0 users guide. Chicago: Author.
StataCorp. (1999). Stata statistical software: Release 6.0. College Station,
TX: Author.
Stroul, B. A. (Ed.). (1996). Children’s mental health: Creating systems of
care in a changing society. Baltimore: Brookes Publishing.
Tate, D. C., Reppucci, N. D., & Mulvey, E. P. (1995). Violent juvenile
delinquency: Treatment effectiveness and implications for future action.
American Psychologist, 50, 777–781.
Tobit, J. (1958). Estimation of relationships for limited dependent vari-
ables. Econometrica, 26, 24–36.
Weisz, J. R., & Hawley, K. M. (1998). Finding, evaluating, refining, and
applying empirically supported treatments for children and adolescents.
Journal of Clinical Child Psychology, 27, 206–216.
Weisz, J. R., Hawley, K. M., Pilkonis, P. A., Woody, S. R., & Follette,
W. C. (2000). Stressing the (other) three Rs in the search for empirically
supported treatments: Review procedures, research quality, and rele-
vance to practice and the public interest. Clinical Psychology: Science
and Practice, 7, 243–258.
Received January 5, 2004
Revision received November 11, 2004
Accepted November 12, 2004 䡲
453
LONG-TERM FOLLOW-UP OF MST
A preview of this full-text is provided by American Psychological Association.
Content available from Journal of Consulting and Clinical Psychology
This content is subject to copyright. Terms and conditions apply.
