Transcript MDFT Alternative to Residential Treatment Study

JMATE 2012
Multidimensional Family Therapy: New Settings,
New Studies, New Outcomes
Howard A. Liddle, Gayle A. Dakof
Department of Epidemiology & Public Health,
and Center for Treatment Research on Adolescent Drug Abuse
University of Miami Miller School of Medicine
Miami, Florida
Craig Henderson
Department of Psychology
Sam Houston State University,
Huntsville Texas
Am. J. Drug & Alcohol Abuse, 2009, 35, 220-2009
 Certain malleable parent and youth characteristics predict
engagement
 Parent expectations about education, and severity of
externalizing
 Youth report of higher levels of family conflict
 Used as part of the content base that informs MDFT
engagement strategies
 Differential strategies for youth and parent
Evaluations of MDFT
NIDA NOTES 2011
“MDFT treatment outcomes are among the best
there are for adolescents. Not only does it work, but
it joins the category of behavioral interventions
whose effects seem to endure after treatment ends.”
Lisa Onken, PhD, Chief of the Behavioral and Integrative Branch
National Institute on Drug Abuse
2008
”Multidimensional Family Therapy was the only probably
efficacious treatment for drug-abusing ethnic minority youth.
(p. 206)
The strong research base demonstrating the effects of MDFT in both indicated
prevention and treatment settings has led it to be recognized as a best practice by the
Office of Juvenile Justice and Delinquency Prevention, the National Institute on Drug
Abuse (1999), the U.S. Department of Health and Human Services (2002), and
SAMHSA (2005).
“The strongest empirical support has been provided for Multidimensional Family Therapy
(MDFT) and group administered Cognitive Behavioral Therapy (CBT). While MDFT and
Multisystemic Therapy (MST) have similar treatment foci and theoretical underpinnings,
MDFT has stronger empirical support, with replicated sustained results.”
Perepletchikova, Krystal, & Kaufman, J. (2008)
Is It Possible to Create an Effective, Outpatient
Alternative to Residential Treatment?
Howard A. Liddle, Gayle A. Dakof, Cindy Rowe, Craig
Henderson, Paul Greenbaum, and Linda Alberga
JMATE July 12, 2012
Center for Treatment Research on Adolescent Drug Abuse
University of Miami Miller School of Medicine
A challenge, a puzzle, a scandal… a mess!
 Adolescent substance abuse, juvenile justice involvement of youth, high risk





sexual activity, school failure, family stresses and dysfunction
Co-morbidity is the norm in clinically referred samples
5% of youth who need it get treatment
When youth do get treatment they drop out with an alarming frequency
 Kazdin’s 40-60%
 Grella et al 2001 DATOS-A 23% complete 90 days, 77% drop out before
90 days
Existing services are rarely evidence-based programs
 Standard treatment yields worse outcomes than EBPs
Family-based therapies among the most tested and transferred to community
clinics
 Still, family-based treatment is far from the norm
 Knudsen (2010) JSAT adolescent specific services, few families
 Chassin et al (Pathways to Desistance) – family involvement cases offer
better outcomes, but less than 20% of the cases get any family
involvement
 Family involvement does not equal evidence based therapy
Think of a sunset…
Start to describe it…
http://www.youtube.com/watch?v=tu-r27w6mgg
Link to
youtube video
Context
 Co-morbidity is the norm in adolescent substance abuse samples
 One of the few rigorous evaluations of an outpatient treatment,
multidimensional family therapy, tested as an alternative to the
residential treatment of substance abusing and conduct disorder
youths
 Inquiring minds want to know
 Alternative to residential treatment?
 Can youths meeting ASAM criteria for intensive interventions that
remove the youths from their home and communities be safely and
effectively treated with a family-based outpatient alternative.
 To our knowledge, this is the first randomized controlled trial of a
family-based treatment evaluated as an outpatient alternative to
residential drug abuse treatment for a substance abuse, co-morbid
sample.
Participants: Sample Characteristics
 113 adolescents (84 males [75%] and 29 females [25%]) with an average
age of 15.36 (SD = 1.07)
 Ethnically diverse:
 Hispanic (68%)
 African American (15%)
 white, non-Hispanic (12%)
 American Indian (3%)
 Haitian or Jamaican (2%)
 Socioeconomic status with a median yearly family income of $18,777
 Parents - 33% previous criminal involvement; 50% previous or current
alcohol or drug use problems
Sample Characteristics
 81% involved in the juvenile justice system at intake, either on probation




or pending a court hearing and had an extensive history of school
problems
66% having repeated at least one grade, and 16% having repeated two
or more.
Psychiatric evaluation conducted by a single board certified child and
adolescent psychiatrist who was blind to participant’s treatment condition
assignment.
79% met criteria for cannabis dependence (4% for abuse), 16% for
alcohol dependence (14% for abuse), 15% for polysubstance
dependence, 13% for cocaine dependence (12% for abuse), and 1% for
opioid dependence (1% for abuse).
90% had initiated substance use before the age of 15, and 39% reported
substance use initiation before age 12.
Sample Characteristics
 Consistent with the study and program eligibility criteria, all youth also
met criteria for a comorbid psychiatric disorder at intake:
 78% had moderate to serious conduct disorder
 21% ADHD
 18% major depressive disorder
 8% bipolar disorder
 9% dysthymic disorder
 Youth had an average of 3.83 (SD = 3.31) total psychiatric diagnoses,
including substance use disorders.
 Seventy-nine percent of adolescents had a previous substance abuse
treatment episode (34% having two or more), with 71% having had a
previous residential treatment episode (17% two or more).
 The treatment groups did not differ significantly (p= > .05) on any of
these variables at baseline or on any demographic characteristics.
Therapists


Primary therapists in both conditions held a master’s degree in
counseling, social work, family therapy or a related field, and
had equivalent prior experience (M=2 yrs.).
In both conditions, primary therapists worked on a
multidisciplinary team, assisted by therapist assistants/case
managers (MDFT) and milieu staff (residential), and having the
same adolescent psychiatrist conduct an initial evaluation and
regular appointments to monitor medications and compliance.
Measures


Measures administered at all assessment points.
Data capture rates were high for parents and
youths, respectively:

intake 98/99%

2 month 97/99%

4 month 96/96%

12 month 95/96%

18 month 97/99%

48 month data being collected
Outcome Measures- Substance Use

The Personal Experience Inventory (PEI; Winters & Henley,
1989) is a multi-scale self-report measure assessing
substance use problem severity and psychosocial risk.

The Personal Involvement with Chemicals scale was
used in the current study and is a 29-item scale focusing
on the psychological and behavioral depth of substance
use involvement and related consequences in the
previous 30 days.

Items composing this scale address substance use to
feel calm; substance use during the whole day,
weekends, or school; and canceling plans to get high.
Widely used in applied research settings (Weinberg,
Rahdert, Colliver, & Glantz, 1998), the PEI demonstrates
excellent reliability (alpha=.84 to .97) and validity (e.g.,
scales significantly related to diagnostic ratings) across
diverse adolescent samples (Henly & Winters, 1989;
Tarter, 1990; Winters, Latimer, Stinchfield, & Egan,
2004). Coefficient alpha for the current study was .95.
Outcome Measures - Substance Use
 Timeline Follow-Back Method (TLFB) measured
youths’ substance consumption (Sobell & Sobell,
1992). The measure was adapted to measure
adolescent drug use (Leccese & Waldron, 1994).
 TLFB obtained 30-day retrospective reports of daily
substance use by employing a calendar and other
memory prompts to stimulate recall.
 Youth report on specific substances used daily for
the 30-day period just prior to the intake evaluation
and each follow-up evaluation.
 A 30-day period was selected given the potential for
recall bias for longer periods of time (Vinson,
Reidinger, & Wilcosky, 2003).
Outcome Measures - Delinquent
Behavior
 National Youth Survey Self Report Delinquency Scale
(SRD) is a well-validated instrument that has been used
extensively with African American and Hispanic
adolescents.
 This measure was administered to youth at all
measurement occasions. Part of the National Youth
Survey (Huizinga & Elliot, 1983), the SRD assesses
criminal behavior on five subscales: total delinquency,
general theft, crimes against persons, index offenses,
and drug sales.
 The SRD is well validated with clinical samples and
serious offenders (Henggeler, 1989).
Outcome Measures - Mental Health
Symptoms
 The Youth Self-Report (YSR; Achenbach, 1991a) and Child
Behavior Checklist (CBCL; Achenbach, 1991b) were used to
assess adolescent and parent reports of youth internalizing
and externalizing symptoms.
 We used the internalizing scale to assess internal distress
and the aggression and delinquency subscales to assess
these specific externalizing symptoms.
 The YSR, and the CBCL on which the YSR is based, are two
of the best validated measures of child-behavioral
functioning.
Outcome Measures- Data Analytic Approach
 MDFT and ATP treatments were compared on the following





primary outcomes:
 (1) substance use
 (2) externalizing symptoms of aggression and delinquency
 3) internalizing symptoms
 (4) frequency of delinquent behaviors
Individual client change for the primary outcomes was analyzed
using latent growth curve (LGC) modeling (Curran & Hussong,
2003).
Individual differences are captured in random variances for the
growth parameters, providing estimates of individual variation
around the average group intercept and slope estimates.
Consistent with our hypotheses, we modeled growth trajectories as
a discontinuous change process (i.e., a piecewise model) using
two distinct trajectories.
The first trajectory represented change during early treatment
(intake through the 2 month follow-up) and the second trajectory
represented change during follow-up (4 month to the 18 month
follow-up).
Outcome Measures - Data Analytic
Approach
 In addition to self- and parent-report data,
we also obtained official court records
regarding youth arrests and charges, along
with school outcomes using records
obtained from the public school’s database
for:
(a) grades
(b) absences
(c) suspensions
Outcome Measures- Data Analytic
Approach
 LGC models controlled for adolescent age, gender, time in treatment, and




initial severity of the outcome variable by entering these variables as
covariates and included all randomized participants in the analyses
regardless of the number of therapy sessions they received (i.e., intent to
treat analyses).
Growth curve modeling was done using Mplus software (Version 5.1; Muthén
& Muthén 1998–2012).
Robust maximum likelihood estimation was used to minimize bias due to
nonnormal outcome variables (Satorra & Bentler, 1994).
In addition, natural log transformation was used to improve the normality of
frequency of delinquent activity and school suspensions.
Missing data were handled using full information maximum likelihood (FIML)
estimation under the missing at random (MAR) assumption (i.e., after
conditioning on observed variables, any remaining missingness is completely
at random; Graham, 2009; Little & Rubin, 1987).
Outcome Measures - Data Analytic Approach
 Due to the severity of substance abuse symptoms and delinquency, the number of psychiatric







diagnoses, the number of previous substance abuse treatment placements, and the extent to which
participants were involved in the justice system at study entry, they were at high risk for being
placed in a long-term juvenile justice or substance abuse treatment facility at some time during the
study follow-up period.
As noted by McCaffery et al. (2007), behavioral frequency data such as TLFB-assessed substance
use and number of delinquent acts committed (e.g., NYS assessment results) are subject to
selection and suppression effects when placement in a post-treatment controlled environment is not
taken into account (i.e., the placement environment may artifiically reduce or eliminate the frequency
of the outcome).
Therefore, we treated TLFB and NYS outcomes differently than our other outcomes that were less
susceptible to such biases.
For these measures, a latent class pattern mixture analysis (LCPMM; Morgan-Lopez & FalsStewart, 2007) was conducted to control for potentially biased reports of substance use and
delinquency.
LCPMM is a variant of Growth Mixture Modeling (GMM) that can take into account participants’
different longitudinal patterns present in data such as therapy attendance (Morgan-Lopez & FalsStewart, 2007), missingness (Linehan et al., 2006), or controlled versus nonrestrictive posttreatment placements.
GMM identifies subgroups or latent classes of individuals with similar growth trajectories; individuals
within each latent class share the same average intercept and slope.
Accounting for bias due to controlled environment placements, LCPMM extends GMM by taking into
account the probability of placement at each month of the 18 month follow-up period.
LCPMM forms latent classes of participants with similar placement probabilities and
outcome trajectories, and treatment effects are examined within each latent class, allowing
treatment comparisons to be made between clients with approximately equivalent placement
patterns.
Study Implementation
 Missing data due to missed assessments at each follow-up
assessment was
 1% at the 2-month follow-up
 4% at the 4-month follow-up
 5% at the 12-month follow-up
 2% at the 18-month follow-up.
 The presence of missing data did not differ by treatment
condition ([(2 (1, N = 113) = 1.83, p = .18).
Fidelity
 We conducted a rigorous treatment fidelity evaluation of both treatments based on
adherence procedures developed in previous MDFT trials (Hogue et al., 1998, 2004)
and methods adapted from evaluation research in residential care settings (Holland,
1986) to specify and measure the components and therapeutic processes of the
residential treatment (Faw et al., 2005).
 In order to demonstrate that therapists adhered to the basic parameters of the
treatments (i.e., session frequency and duration, domains targeted), therapists in
MDFT completed therapeutic contact logs for every contact with clients.
 Residential treatment program daily logs were completed by all ATP staff members
who provided services to the adolescent during a routine program day, including
basic living services (e.g., meals, school, hygiene), therapeutic services (e.g.,
therapy sessions, milieu groups, psychological and psychiatric consultations), and
recreational services.
 Daily logs were routinely completed at the ATP prior to this study; that is, they were
not introduced as a feature of the randomized clinical trial. ATP staff members
logged the amount of time spent in each contact, the general goal of the contact, the
identity of the staff member involved, and any pertinent notes or clinical
observations gathered in the contact.
Fidelity
 Evaluation of treatment contacts revealed that both interventions were delivered




in accordance to their prescribed treatment parameters.
In the residential program, on average, adolescents completed 61% of the weekly
prescribed amount of treatment services, 47% of the prescribed amount of time in
functional activities, 63% of the weekly prescribed productive activities, 60% of
the prescribed number of re-entry activities, and 15% of the prescribed number of
hours of interpersonal/recreational activities (Faw et al., 2005).
Adolescents rated the therapeutic milieu as being highly therapeutic (Faw et al.,
2005).
MDFT cases averaged 3.28 hours per week (SD = 1.74) of family and individual
sessions, as prescribed in MDFT for this level of intervention.
Consistent with MDFT parameters, on average (median) participants received the
following amount to treatment in each of the four types of MDFT sessions:
 (1) adolescent alone (24.7 hours)
 (2) parent(s) alone (8.4 hours)
 (3) parents and adolescent together (37.8 hours)
 (4) extrafamilial contact with or without youth and family members (11.5 hours)
Fidelity
 Observational ratings of therapy sessions were also used to document adherence to






both treatments and differentiate the interventions delivered in individual and family
sessions.
Videotapes of individual and family sessions were selected for rating using the
Therapist Behavior Rating Scale (TBRS), an observational adherence coding system
used in previous MDFT studies (Hogue et al., 1996, 1998).
A total of 31 (27%) MDFT and residential cases were randomly selected for adherence
ratings.
For each of these cases, one session from the middle stage of therapy was randomly
selected to be rated with the TBRS.
The raters were two female doctoral-level clinical researchers trained extensively by
TBRS developers.
They rated the therapy sessions on the extensiveness with which the therapists
adhered to core MDFT and drug treatment interventions.
Raters demonstrated good interrater reliability (ICC(1,2)=.86) using a subset of 5 MDFT
sessions coded by both raters before coding study tapes.
Fidelity
 Equivalence testing procedures (Tryon, 2001) were used to compare the mean MDFT




adherence score obtained in the current study to the mean MDFT adherence score
reported in a previous MDFT fidelity study establishing the validity of the TBRS (Hogue,
Liddle, Dauber, & Samuolis, 2004). Following Fals-Stewart and Birchler’s (2002)
procedures, we used an equivalence interval (EI) of +/- 10% around the mean MDFT
adherence score obtained by Hogue et al. (2004; i.e., the reference group mean).
The reference group mean was 31.09 (SD=8.37) and the EI was +/-3.10. A 90%
confidence interval (CI) was calculated around the mean MDFT adherence score
obtained in the current study (i.e., the test group mean).
The obtained test group mean was 31.18 (SD=8.06), making the 90% CI 28.06 to
34.30.
Though the 90% CI for the test group mean fell slightly outside of the pre-established
EI around the reference group mean, it was because therapists in the current study
obtained higher scores on the TBRS than the reference group.
Thus we concluded that the therapists delivered MDFT with high fidelity.
2005
 Logic model containing two main components was measured.
 Program structure (adherence to the intended framework of service delivery) was
measured using data from daily activity logs completed by program staff.
 Treatment process, conceptualized as therapeutic milieu, was measured using an
adapted version of a scale used to measure implementation in therapeutic
communities.
 Milieu rated by the adolescents as highly therapeutic.
 Preliminary psychometrics suggest therapeutic milieu can be measured reliably in
adolescents.
 These two main variables were implemented with consistency across adolescents.
Results
Treatment Retention
 The acceptability and feasibility of outpatient MDFT with this
severely impaired, referred for residential population was
explored by comparing treatment retention rates in the two
conditions.
 Further, it was important to consider early treatment retention
due to the differing restrictiveness of the two treatments.
 Outpatient MDFT 6.5 months / Residential treatment 3.7
months
 Youth receiving MDFT remained in treatment longer than
youth receiving residential treatment (average length of
stay 6.5 [SD = 2.0] vs. 3.7 [SD = 3.0] months; t (111) =
5.81, p < .001).
 In addition, youth in MDFT were more likely to be retained in
treatment for three months than those receiving residential
treatment [2 (1, N = 113) = 22.50, p < .001, OR = 11.5).
Intake to 2 Months Following Intake
Substance Use Problem Severity
Aggression
Delinquency
Internalizing Symptoms
Substance Use: Baseline to 2 months
 Both treatments show significant declines in substance use
 From intake to 2 month follow up, all youth showed a
significant decline in substance use problem severity as
measured by the PEI (Mean Slope = -12.39, standard
error [SE] = 1.13, pseudo z = -10.69, p < .001).
 Contrary to our hypothesis – no difference between
outpatient MDFT and residential treatment.
 There was not a significant treatment difference during
this initial treatment phase despite our hypothesis that
the residential treatment would improve more (treatment
coefficient for slope = -3.88, SE = 2.56, pseudo z = 1.52), as both treatments showed large decreases in
substance use.
Delinquency & Aggression Symptoms:
Baseline to 2 months
 Youth in both treatments show significant declines
 As with substance use problem severity, youth in both treatments showed
declines in delinquency and aggression symptoms during the first two
months of treatment according to both parent and youth reports (Parent
Report Delinquency: Mean Slope = -9.26, SE = 1.37, pseudo z = -6.76, p <
.001; Youth Report Delinquency: Mean Slope = -5.94, SE = 0.81, pseudo z
= -7.37, p < .001; Parent Report Aggression: Mean Slope = -3.15, SE =
0.86, pseudo z= -3.67, p < .001; Youth Report Aggression: Mean Slope = 0.99, SE = 0.14, pseudo z= -6.89, p < .001).
Delinquency & Aggression Symptoms:
Baseline to 2 months
 Parent report - Parents of residential treatment youths report a more
rapid decrease than MDFT parents
 With respect to treatment differences, in this early phase of treatment parents of youth
receiving residential treatment reported a more rapid decrease in both delinquency and
aggressive symptoms in their teen than did parents of teen who received MDFT (Parent
Report Delinquency: treatment coefficient for slope = -11.78, SE= 2.43, pseudo z = 4.93, p < .001, 95% CI = -16.64 to 6.92; Parent Report Aggression: treatment coefficient
for slope = -6.04, SE = 1.55, pseudo z = -3.89, p < .001, 95% CI = -9.14 to -2.94).
 Youth report – No treatment differences according to youth self report
(Youth Report Delinquency: treatment coefficient for slope = -1.38, SE
= 1.52, pseudo z = -0.90, ns; Youth Report Aggression: treatment
coefficient for slope = 1.92, SE = 1.02, pseudo z = 1.88, ns ) with both
groups reporting a similar decrease in delinquency symptoms.
Internalizing Symptoms: Baseline to 2
months
 Both treatment groups decrease internalizing symptoms
 Youth in both treatments reported significant decreases in internalizing symptoms
during early treatment (Mean Slope = -1.36, SE = 0.50, pseudo z= -2.71, p < .01).
 But parent rated internalizing symptoms did not concur
 However, parent-rated symptoms did not decrease (Mean Slope = -0.46, SE =
0.65, pseudo z = -0.71, ns).
 Comparing the treatments – MDFT youth show significantly
greater decreases in internalizing symptoms
 Youth receiving MDFT reported greater decreases than youth receiving ATP
(slope coefficient on treatment = 2.60, SE = 0.92, pseudo z = 2.81, p < .01, 95%
CI = 0.76 to 4.44).
 Parents reports on decreases in internalizing symptoms
 There were no treatment differences according to parents’ reports (slope
coefficient on treatment = -2.15, SE = 1.21, pseudo z = -1.78, ns).
Anxiety/Depressive Symptoms and
Withdrawl: Baseline to 2 months
 Youth in both treatments reported significant decreases in
both anxiety/depressive symptoms and withdrawal during
early treatment (Anxiety/Depression: Mean Slope = -1.40,
SE = 0.50, pseudo z= -2.81, p < .01; Withdrawal: Mean
Slope = -1.67, SE = 0.50, pseudo z= -3.37, p < .01).
 Parents reported decreases in withdrawal (Mean Slope = 1.54, SE = 0.76, pseudo z= -2.02, p < .05) but not
anxiety/depressive symptoms (Mean Slope = -0.23, SE =
0.53, pseudo z= -0.44, ns).
Results- Anxiety/Depressive
Symptoms and Withdrawl: Baseline to
2 months
 MDFT youth report significantly greater decreases in
anxiety – depressive symptoms and withdrawl
 Comparing the treatments, youth receiving MDFT reported greater
decreases than youth receiving residential treatment in anxiety/depressive
symptoms and withdrawal (Anxiety/Depression: slope coefficient on
treatment = 2.00, SE = 0.95, pseudo z= 2.09, p < .05, 95% CI = 0.10 to
3.90; Withdrawal: slope coefficient on treatment = 1.09, SE = 0.17, pseudo
z= 6.32, p < .001, 95% CI = 0.75 to 1.43).
 No treatment differences according to parents self report
 There were no treatment differences in either outcome according to
parents’ reports (Anxiety/Depression: slope coefficient on treatment = 1.03, SE = 1.05, pseudo z = -0.98, ns; Withdrawal: slope coefficient on
treatment = -2.21, SE = 1.50, pseudo z= -1.48, ns).
Outcomes at 18 Months
Following Intake
Longer-term outcomes to determine the
sustainability of changes following early
treatment (approximately 2 months after
intake) through 18 months after intake.
Substance Use at 18 Months
 From 2 to 18 months substance use problem severity
remained relatively low in comparison to intake status and
stable (Mean Slope = 0.12, SE = 0.11, pseudo z = 1.17, ns).
 MDFT youths maintain previous decreases. Residential
youths increase substance use problem severity
 When comparing the treatments, youth receiving MDFT
maintained their early treatment gains; while youth
receiving residential treatment reported increased
substance use problem severity over time (slope
coefficient for treatment = 0.72, SE = 0.22, pseudo z =
3.28, p < .01, 95% CI = 0.28 to 1.16; see Figure 1).
 Note: Although youth who received residential treatment
showed increased substance use problems in comparison
to youth who received MDFT, this increase did not reach
baseline levels.
Drug Use Problem Severity
70
65
Drug Abuse Problem Severity
60
55
MDFT
50
RT
45
40
35
30
Intake
2 Months
4 Months
12 Months
18 Months
Delinquency-Related Symptoms and
Aggression at 18 months
 Parents of MDFT youths report continued decreases at 18 months
 Comparing the treatments, parents of youth receiving MDFT, in comparison
to parent reports from youth who received residential treatment, indicate a
continuing decrease in symptoms of delinquency and aggression over the
follow-up period (Parent Report Delinquency: treatment coefficient for slope
= 1.22, SE = 0.39, pseudo z = 3.11, p < .01, 95% CI = 0.44 to 2.00; Parent
Report Aggression: slope coefficient for treatment = 0.89, SE = 0.22,
pseudo z = 4.02, p < .001, 95% CI = 0.44 to 1.32) (see Figure 2).
 Youths in MDFT vs. residential report more decreases in aggressive behaviors
at 18 months
 Youth in MDFT report more pronounced decreases over time in aggression
than youth from residential treatment (Youth Report Aggression: slope
coefficient for treatment = 0.07, SE = 0.03, pseudo z = 2.10, p < .05, 95%
CI = 0.01 to 0.13). There were no differences, however, in youth reports of
delinquency (Youth Report Delinquency: treatment coefficient for slope =
0.10, SE = 0.17, pseudo z = 0.56, ns) with youth in both treatments
reporting a general maintenance of decreased delinquency symptoms.
Delinquent Behavior
100
95
90
Delinquency Symptoms
85
80
MDFT
75
RT
70
65
60
55
50
Intake
2 Months
4 Months
12 Months
18 Months
Internalizing Symptoms at 18 Months
 Between 2 and 18 months, parents reported a decrease in
their teen’s internalizing symptoms (Mean Slope = -0.32,
SE = 0.08, pseudo z = -3.97, p < .001).
 Youth reports indicated these symptoms remain reduced –
data show a trend toward significant decreases (Mean
Slope = -0.12, SE = 0.06, pseudo z= -1.85, p < .10).
 There were no treatment differences according to both
parents (slope coefficient for treatment = -0.18, SE = 0.14,
pseudo z = -1.25, ns) and youth (slope coefficient for
treatment = 0.05, SE= 0.16, pseudo z = 0.31, ns).
Anxiety/Depressive Symptoms and
Withdrawal at 18 Months
 Between 2 and 18 months, anxiety/depressive symptoms and withdrawal
remained stable according to youth reports (Anxiety/Depression: Mean
Slope = -0.09, SE = 0.07, pseudo z= -1.28, ns; Withdrawal: Mean Slope = 0.06, SE = 0.08, pseudo z= -0.76, ns).
 In contrast with the early treatment results, which showed no change,
parents reported decreases in both anxiety/depressive symptoms and
withdrawal (Anxiety/Depression: Mean Slope = -0.31, SE = 0.07, pseudo z=
-4.17, p < .001; Withdrawal: Mean Slope = -0.27, SE = 0.10, pseudo z= 2.72, p < .01).
 There were no treatment differences according to youth or parent reports in
either outcome (Parent Report Anxiety/Depression: slope coefficient on
treatment = -0.02, SE = 0.15, pseudo z= -0.16, ns; Parent Report
Withdrawal: slope coefficient on treatment = 0.13, SE = 0.20, pseudo z=
0.63, ns; Youth Report Anxiety/Depression: slope coefficient on treatment =
-0.22, SE = 0.16, pseudo z = -1.41, ns; Withdrawal: slope coefficient on
treatment = -0.03, SE = 0.16, pseudo z= -0.19, ns).]
Latent Class Pattern Mixture Modeling:
Drug Use and Delinquent Activity
 Results of the LCPMM indicated that three latent classes provided





the best representation of the heterogeneity in placement patterns.
The first class (Early Placement) represented 18% of participants
who showed high probabilities of being placed in a controlled
environment toward the beginning of the follow up period (defined
between 3 – 9 months from intake).
By the end of the follow up period, defined as between 10 – 18
months from intake, these individuals tended to be discharged
from their placements and living at home.
Members of the second class (Late Placement; 11%) were also
institutionalized at a high rate, but the placement tended to occur
later in follow-up.
Earlier in the follow-up period—during months 3 to 9—these
individuals showed moderate probabilities of being placed in a
controlled environment.
The third class consisted of 72% of the sample (Minimal
Placement) who showed fairly low probabilities of being
institutionalized throughout the follow-up period.
Minimal Placement Class
 Drug use frequency outcome at 2 months
 Thirty-day drug use frequency among youth in the Minimal Placement class decreased
similarly across treatments during the first two months of treatment (slope coefficient for
treatment = 1.01, SE = 2.34, pseudo z= -4.18, p < .001)
 Drug use frequency outcome at 18 months
 This pattern changed in the follow up period where youth who received residential
treatment showed a greater increase in drug use in comparison to MDFT youth (b = 0.41,
SE = 0.20, pseudo z= 1.96, p < .05, 95% CI = 0.00 to 0.81)
 Treatment effects for the Minimal Placement class were significant for frequency of
delinquent behaviors.
 Delinquent behavior outcomes at 2 months
 During early treatment, residential treatment youths showed a trend toward decreasing
their delinquent behavior more rapidly than those who received MDFT (slope coefficient
for treatment = -0.23, SE = 0.14, pseudo z= -1.67, p = .10).
 Delinquent behavior outcomes at 18 months
 However, during the follow-up period, residential treatment youths increased their
delinquent activity, while MDFT youths remained stable and maintained their treatment
gains (slope coefficient for treatment = 0.04, SE = 0.02, pseudo z = 2.43, p < .05).
Three study hypotheses were tested
 The first hypothesis addressed the feasibility of a family based
outpatient alternative to residential treatment.
 A majority of youth in both residential and MDFT remained in
treatment for 90-days or longer.
 Youth in MDFT, however, remained in treatment almost 3 months
longer than did youth in residential treatments.
Comment
 A fundamental question in this study concerned the feasibility of an outpatient




alternative treatment for youth who had been deemed in need of residential
treatment.
Strong documentation exists supporting the challenges of treating
adolescents with the characteristics of the current sample across treatment
modalities and levels of care (Wong et al 2002).
Although there is no national standard for the prescribed length of stay in
residential treatment programs, and in fact, there is significant variability on
what should constitute an adequate or preferred amount of treatment dose,
some standards have been used with adolescent samples.
In national studies DATOS-A study (Grella et al 2001), and the SAMHSA
CSAT Adolescent Treatment Study (Dennis, 2007), the preferred treatment
length was 90 days (also see Gottfredson et al).
In the DATOS-A study 58% of the adolescents referred for residential
treatment remained in the residential programs for 3 months.
Comment
 In the present study, the residential treatment program met this benchmark, retaining





54% of the adolescents at 3 months.
One of the most consistent findings in the adolescent substance abuse treatment
outcome literature pertains to program completion and time in treatment.
MDFT participants – average length of 195 days vs. 111 days for residential treatment
MDFT in this study and in other trials engages and retains adolescents and their
families at comparatively higher rate
Another point related to the meaning of the retention outcomes can be noted. In the
DATOS-A study, the residential sample youths who had criminal involvement, of the
kind evidenced in the current study sample, demonstrated significantly worse retention
and higher rates of substance abuse post discharge (Galaif et al 2002).
Current study - multiple diagnoses youths who were referred for residential treatment
and largely juvenile justice involved (81%) were able to engage in the family-based
outpatient alternative and improve on several important dimensions, including
substance abuse, unlike the adolescents in the Galaif et al (2002) study.
Comment
 A second hypothesis predicted that in the early phase of treatment, residential




treatment youths, because of the greater intensity of the treatment and the
restrictive environment, would show greater reduction in substance use,
externalizing symptoms, delinquent behaviors, and internalized distress compared
to the outpatient alternative youths receiving MDFT.
The results did not support this hypothesis.
First, it should be recognized that for both treatments the greatest decline on all
variables was from intake to 2 months.
Youth in residential treatment did not report better outcomes than youth in MDFT
on any of the domains examined: substance use, aggression, delinquency, and
internalizing symptoms.
In fact, youth in MDFT reported significantly greater reduction in internalizing
symptoms than residential youth during the first 2 months of treatment.
Comment
 The third hypothesis was based on existing evidence for long-term effects of
MDFT in previous trials and its family-focus, targeting known characteristics and
processes related to substance use and antisocial behavior.
 We hypothesized that the pattern of improvements would reverse later in
treatment and over the follow-up period, per other residential treatment
evaluations, the gains made in early treatment (2 months) would not be
maintained by residential treatment participant but would be maintained by MDFT
youth at the 18-month follow up assessment.
 The results appear to support this hypothesis with one exception, that is,
internalizing disorders, where even though MDFT reported greater symptom
reduction than residential youth during the first two months of treatment, ultimately
there were no treatment differences during the 2 – 18 month period.
 However, with respect to drug use, and symptoms of conduct disorder, namely
delinquency and aggression, both youth and parent reports on all variables
measured are consistent: youth randomized to the outpatient family-based
treatment showed greater symptom reduction than youth randomized to
residential treatment from 2 to 18 months following intake into treatment.
 Limitations: only one site administered the residential treatment and another
the outpatient alternative.
 Because multisite trials indicate that outcomes can vary significantly according the treatment site
(Helgerson et al 2005), and despite the RT site’s representative in terms of program features,
and the intensive program fidelity analysis, the use of a single residential treatment setting is
factor that limits the study’s generalizability.
 Comparatively few numbers of non-hispanic whites and girls included in the
sample
 Strengths include strong methods, including full randomization, intent to treat
design, multi-source outcomes, multiple measures of key outcomes with
standardized and objective indicators, assessors blind to client’s treatment,
and solid data capture rates. The study tested two well defined, theory driven,
and well-defined treatments (Epstein (2004; Curry, 1991) and adherence
checking indicated that treatments were delivered as planned.
Is It Possible to Create an Effective, Outpatient
Alternative to Residential Treatment?
Summary and Conclusion
 Parents and youth referred for residential treatment could be
retained in an outpatient, family based treatment.
 Youth in both treatments achieved considerable gains over the first,
early phase of each treatment.
 However, only MDFT youths maintained or enhanced these gains
through the 18 month follow up, while the youths receiving
residential treatment did not maintain their in-treatment gains.
 48 month results are forthcoming
 In this study outpatient MDFT demonstrated clinically significant
and stable effectiveness as an alternative for multiply diagnosed
youths referred for residential treatment
Summing up
A trainer’s experience.
“...that’s just the best”.
http://www.youtube.com/watch?v=dSp_XW2K6gI&sns=em