Research Review: The effectiveness of multidimensional family therapy in treating adolescents with multiple behavior problems - a meta-analysis

Abstract

Background:
Multidimensional family therapy (MDFT) is a well-established treatment for adolescents showing both substance abuse and/or antisocial behavior.

Method:
The effectiveness of MDFT in reducing adolescents' substance abuse, delinquency, externalizing and internalizing psychopathology, and family malfunctioning was examined by means of a (three-level) meta-analysis, summarizing 61 effect sizes from 19 manuscripts (N = 1,488 participants).

Results:
Compared with other therapies, the overall effect size of MDFT was significant, albeit small in magnitude (d = 0.24, p < .001), and similar across intervention outcome categories. Moderator analysis revealed that adolescents with high severity problems, including severe substance abuse and disruptive behavior disorder, benefited more from MDFT than adolescents with less severe conditions.

Conclusions:
It can be concluded that MDFT is effective for adolescents with substance abuse, delinquency, and comorbid behavior problems. Subsequently, it is important to match specific characteristics of the adolescents, such as extent of impairment, with MDFT.

Full text

Research Review: The effectiveness of
multidimensional family therapy in treating
adolescents with multiple behavior problems –a
meta-analysis
Thimo M. van der Pol,
1,2
Machteld Hoeve,
3
Marc J. Noom,
3
Geert Jan J. M. Stams,
3
Theo A. H. Doreleijers,
2
Lieke van Domburgh,
2
and Robert R. J. M. Vermeiren
1,2
1
Department of Child and Adolescent Psychiatry, Curium –Leiden University Medical Centre, Leiden;
2
Department of
Child and Adolescent Psychiatry, VU University Medical Centre, Amsterdam;
3
Research Institute of Child
Development and Education, University of Amsterdam, Amsterdam, The Netherlands
Background: Multidimensional family therapy (MDFT) is a well-established treatment for adolescents showing both
substance abuse and/or antisocial behavior. Method: The effectiveness of MDFT in reducing adolescents’ substance
abuse, delinquency, externalizing and internalizing psychopathology, and family malfunctioning was examined by
means of a (three-level) meta-analysis, summarizing 61 effect sizes from 19 manuscripts (N=1,488 participants).
Results: Compared with other therapies, the overall effect size of MDFT was significant, albeit small in magnitude
(d=0.24, p<.001), and similar across intervention outcome categories. Moderator analysis revealed that
adolescents with high severity problems, including severe substance abuse and disruptive behavior disorder,
benefited more from MDFT than adolescents with less severe conditions. Conclusions: It can be concluded that
MDFT is effective for adolescents with substance abuse, delinquency, and comorbid behavior problems.
Subsequently, it is important to match specific characteristics of the adolescents, such as extent of impairment,
with MDFT. Keywords: Meta-analysis; adolescence; addiction; delinquency; disruptive behavior.
Introduction
Substance abuse disorders (SUD) in adolescents
predispose to a variety of behavior problems, such
as delinquency, externalizing and internalizing psy-
chopathology (Grella, Hser, Joshi, & Rounds-Bryant,
2001; Merikangas et al., 2010), and family malfunc-
tioning (Colins et al., 2011; Cuellar, McReynolds, &
Wasserman, 2006; Hoeve, McReynolds, & Wasser-
man, 2013; McReynolds & Wasserman, 2011). The
incidence of SUD-related comorbidity is estimated
to reach up to 75% (Grella et al., 2001), which
influences treatment outcome substantially. For
instance, the presence of externalizing psychopathol-
ogy in combination with SUD increases the likelihood
of engaging (Anderson, Ramo, Schulte, Cummins, &
Brown, 2007; Anderson, Tapert, Moadab, Crowley, &
Brown, 2007; Monahan, 2003) and persisting in
delinquent behavior (Lodewijks, De Ruiter, & Dorelei-
jers, 2010; Wasserman, McReynolds, Fisher, &
Lucas, 2003). The same pattern has been observed
in adolescents with internalizing psychopathology
(Loeber, Stouthamer-Loeber, & Raskin White, 1999).
As such, the presence of multiple behavior problems
in adolescence creates major societal and public
health concerns (Johnston & Hauser, 2008; Moffit,
1993). Hence, effective prevention and treatment
programs to address the complex problems of
adolescents with SUD are direly needed (Hall et al.,
2016; Merikangas et al., 2010).
In the last 30 years, several treatments have been
developed to effectively reduce SUD, delinquency,
and comorbid behavior problems. Various system-
atic literature reviews and meta-analyses have con-
cluded that family-based treatments and cognitive
behavioral therapy (CBT) are effective in treating
adolescents with SUD, delinquency, and comorbid
psychopathology (Carr, 2009; Von Sydow, Retzlaff,
Beher, Haun, & Schweitzer, 2013; Waldron &
Turner, 2008). A promising family-based treatment
program is multidimensional family therapy (MDFT;
Liddle, 2002). The present meta-analysis focuses on
the effectiveness of MDFT compared with other
treatments in reducing adolescents’ substance
abuse, delinquency, externalizing and internalizing
psychopathology, and family malfunctioning.
Multidimensional family therapy
Multidimensional family therapy is a manualized,
evidence-based, intensive intervention program with
assessment and treatment modules focusing on four
areas: (a) the individual adolescents’ issues regard-
ing SUD, delinquency, and comorbid psychopathol-
ogy, (b) the parents’ child-rearing skills and personal
functioning, (c) communication and relationship
between adolescent and parent(s), and (d) interac-
tions between family members and key social sys-
tems (Liddle, 2002). MDFT is based on the family
Conflict of interest statement: No conflicts declared.
©2017 Association for Child and Adolescent Mental Health.
Published by John Wiley & Sons Ltd, 9600 Garsington Road, Oxford OX4 2DQ, UK and 350 Main St, Malden, MA 02148, USA
Journal of Child Psychology and Psychiatry **:* (2017), pp **–** doi:10.1111/jcpp.12685

therapy foundation established by Minuchin (1974)
and Haley (1976) and on the ecological systems
theory of Bronfenbrenner (1979) which states that
human development is shaped by the interaction of
the individual with his or her surrounding social
contexts. Within each adolescent’ environment,
there are multiple risk and protective factors that
influence and reinforce each other (Brook, White-
man, & Finch, 1992). Therefore, MDFT was devel-
oped to intervene in multiple systems, addressing
these risks and strengthening protective factors in
the adolescents’ environments (Liddle, 1999). MDFT
is operational and expanding briskly in Europe and
in the United States and targets youth from diverse
ethnic and socioeconomic backgrounds in a variety
of settings (Liddle, 2002; Rigter et al., 2010).
The effectiveness of MDFT
Three previous meta-analyses (Baldwin, Christian,
Berkeljon, & Shadish, 2012; Filges, Andersen, &
Jørgensen, 2015; Tanner-Smith, Wilson, & Lipsey,
2013) summarized the results of studies that
examined the effectiveness of MDFT alone or
together with other family-based treatments. Tan-
ner-Smith et al. (2013) concluded that for sub-
stance abuse, family therapy is the treatment with
the strongest evidence of comparative effectiveness.
The overall effect compared with nonfamily treat-
ments was small (d=0.26). Similarly, Baldwin et al.
(2012) found family therapies to have a small effect for
substance abuse and delinquency compared with
treatment as usual (d=0.21) and alternative treat-
ments, such as group therapy (GT), psychodynamic
family therapy, individual therapy, parent groups,
and family education (d=0.26). It must be noted that
the Baldwin et al. study did not include any follow-up
data of the studies they reviewed in their meta-
analysis. Filges et al. (2015) concluded that MDFT
was successful in reducing adolescents’ substance
abuse in the short run, but not in the long run (no
Cohen’s dwas reported).
The studies included in the meta-analyses revealed
substantial variability in the effectiveness of MDFT,
which may be explained by differences in study
characteristics. For example, differences in MDFT
effectiveness could be related to the severity of
substance abuse and/or psychopathology of partici-
pants. However, the effect of substance abuse sever-
ity, psychopathology, and other potentially important
moderators were not taken into account in previous
meta-analyses. The authors of the three meta-
analyses mentioned not being able to perform exten-
sive moderator analyses due to a limited number of
studies. Therefore, further comprehensive research is
needed. Insight into moderating factors of the effec-
tiveness of MDFT is important for identifying which
adolescents may benefit most from MDFT; this knowl-
edge is crucial for improvement of assessment and
referral practices.
The present meta-analysis
The goal of this study was to provide a meta-analytic
overview of the studies examining the effects of
MDFT compared with other interventions for ado-
lescents with SUD and comorbid behavior problems.
First, we examined the overall effectiveness of MDFT
regarding substance abuse, delinquency, external-
izing and internalizing psychopathology, and family
functioning. Also, the mean effects of MDFT as
compared to CBT, GT, and combined treatments
(CT) were examined. Second, we conducted moder-
ator analyses to investigate whether study charac-
teristics contributed to the effectiveness of MDFT.
The most important question to be investigated was
if adolescents with severe substance abuse and
severe externalizing psychopathology benefited
more from MDFT than adolescents with less severe
conditions, which is from now on called ‘the severity
gradient’. To test this severity gradient (Henderson,
Dakof, Greenbaum, & Liddle, 2010), a three-level
meta-analysis was utilized. This novel three-level
analytic method makes it possible to include more
effect sizes per study and account for differences
between effect sizes both within and between stud-
ies, which prevents important data and information
loss, increases statistical power and the number
of moderators that can be tested (Assink et al.,
2015).
Methods
Sample of studies
Three criteria guided the selection of studies. First, the study
had to examine the effectiveness of MDFT. Second, the study
had to report results for one or more of the following
outcome measures: substance abuse, delinquency,
externalizing and internalizing psychopathology, and family
functioning, or provide enough details to calculate a
bivariate test statistic. Third, in view of study quality, a
study had to report the results of a randomized controlled
trial (RCT).
Candidate studies meeting the selection criteria with data
either published by the 29th of February 2016 or available
from primary authors (unpublished manuscripts) were col-
lected as follows. First, the electronic databases PubMed,
PsycINFO, Embase, and Web of Science were searched for
articles, books, chapters, paper presentations, dissertations,
and reviews. Our purpose was to find as many studies as
possible, and therefore, a variety of terms related to MDFT were
used. Search terms, such as multidimensional*, famil*, and
MDFT, were cross-referenced with therap*and treat*in
English, Dutch, French, and German: ((Multidimensional
Family Therap*) OR (Multidimensional Family Treat*)OR
(MDFT AND (Family OR Therapy OR Multidimensional)).
Subsequently, manual searches of references lists from these
publications were conducted to identify relevant studies not
found in the electronic databases.
If multiple publications were found that reported on the
same study, we only included manuscripts which reported a
different outcome measure or a subsample of the original
study. Furthermore, we contacted the authors of the publica-
tions to check for unpublished materials. Seven manuscripts
were received of which one submitted paper (Liddle, 2015), and
four reports (Grichting, Haug, Nielsen, & Schaub, 2011; Phan,
©2017 Association for Child and Adolescent Mental Health.
2Thimo M. van der Pol et al.

2011; Tossmann & Jonas, 2010; Verbanck et al., 2010) were
eligible to be included in the meta-analysis.
In total 210 manuscripts were found, of which we selected
71 on the basis of information in the abstract. After assessing
the 71 articles, 19 manuscripts on effects of MDFT met our
criteria and were included in the present meta-analysis. For
standardization of the effect sizes and the possibility to
examine the influence of severe behavior problems through
moderator analyses, we asked the authors of the manuscripts
for supplementary information on substance abuse and psy-
chopathology. The 19 manuscripts together with the retrieved
supplementary information yielded 61 effect sizes, resulting
from eight independent studies with a total of 1,488 subjects.
Figure 1 presents a flowchart of the selection procedure.
File drawer problem
The tendency of journals to exclude manuscripts reporting
nonsignificant findings, referred to as publication bias, may
have implications for the final conclusions of the meta-analysis
(Rosenthal, 1991; Van IJzendoorn, 1998). For this, Rosenthal
coined the term ‘file drawer problem’ (1979). Several methods
exist to address potential effects of publication bias, but each
has its own shortcomings (Rothstein, 2008). The best solution
in preventing effects of publication bias is to make extensive
efforts to obtain all unpublished materials (Mullen, 2013;
Rosenthal, 1991). Following the advice of Rothstein (2008),
three methods addressing publication bias were applied. First,
we calculated a fail-safe number, which estimates the number
Citations from four databases:
PsycINFO: 125
Embase: 63
PubMed: 47
Web of Science: 125
Number of citations
(records) identified through
database searching
363
Number of citations
(records) identified through
other sources
7
Number of duplicate citations
removed
160
Number of manuscripts assessed for
eligibility
71
Number of citations excluded based
on abstract
139
Number of manuscripts included in
meta-analytic review
19
Number of manuscripts
excluded
52
Number of citations
screened
210
Figure 1 Flowchart of literature search and screening
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 3

of unretrieved studies reporting null results needed to bring
the overall combined effect size to a level at which it would no
longer be statistically significant (Rosenthal, 1991). The fail-
safe number, 2,554, exceeded Rosenthal’s (1995) critical value
(61*5+10 =315). This indicates that the number of unpub-
lished studies with nonsignificant results that would be
required to reduce significant results to nonsignificant results
was sufficient, suggesting no evidence for publication bias.
A second method of examining publication bias is inspecting
the distribution of each individual study’s effect size on the
horizontal axis against its sample size and standard error or
precision (the reciprocal of the standard error) on the vertical
axis. The distribution of effect sizes should form a funnel shape
if no publication bias is present, as studies with small sample
sizes are expected to show a larger variation in effect size
magnitude, whereas studies with large effect sizes are expected
to result in effect sizes closer to the overall mean. A violation of
funnel plot symmetry reflects publication bias, that is, a
selective inclusion of studies showing positive or negative
outcomes (Sutton, Duval, Tweedie, Abrams, & Jones, 2000).
Figure 2 depicts the funnel plot of effect sizes. In this study,
funnel plot asymmetry was tested by regressing the standard
normal deviate, defined as the effect size, divided by its
standard error, against the estimate’s precision (the inverse
of the standard error), which largely depends on sample size
(Egger, Smith, Schneider, & Minder, 1997). If there is asym-
metry, the regression line does not run through the origin and
the intercept significantly deviates from 0. The intercept did
not significantly deviate from zero (z=1.490, p=.136), indi-
cating no publication bias.
Third, we utilized the P-curve method, which was recently
introduced by Simonsohn, Nelson, and Simmons (2014). The
rationale of the method is that if a set of statistically significant
studies contains real evidential value in favor of rejecting a
joint null hypothesis, p-values extracted from these studies
should display a larger share of p-values closer to zero as
compared with p-values in the upper ranges just below the
critical value (p<.05) of statistical significance. Likewise, if
there are signs of p-hacking, that is, if a nonsignificant p-value
is pushed past the critical value for statistical significance, a
larger share of the p-values should be observed just below the
threshold of statistical significance rather than closer to zero.
The P-curve analyzes whether MDFT is being more or less
effective than the compared therapies. The P-curve test was
performed on all of the statistically significant two-tailed
p-values in our sample. When testing the two-tailed p-values,
the right-skew p-value was <.0001 (Figure 3). The P-curve
showed statistically significant signs of evidential value, and
the statistical power estimated was 85%. It can be concluded
that the results indicate no evidence of p-hacking.
Coding of the study outcomes and characteristics
We retrieved the study results (test statistic and value) or data
to calculate the effect size from the manuscripts. Next, infor-
mation on sample descriptors, treatment descriptors, research
design, and manuscript characteristics were collected.
For the sample descriptors, we categorized the effect sizes
into five primary outcome measures: substance abuse, delin-
quency, externalizing and internalizing psychopathology, and
family functioning. We coded the geographical location where
the study had been conducted (Europe and United States). As
for demographic characteristics, we collected data on age,
gender, socioeconomic status (SES), and ethnicity. We coded
age of the subjects at the start of treatment. Gender was
defined as percentage of males in the sample. The SES was
characterized by calculating the mean family income in euros.
Furthermore, we defined the percentage of Caucasian, Afro-
American, Hispanic, Asian, and other ethnicities (e.g. Carib-
bean and North-African). The percentage of adolescents in the
sample with additional psychiatric disorders was also coded
for conduct disorder (CD), oppositional defiant disorder (ODD),
and disruptive behavior disorder (DBD; i.e. the presence of
either CD and/or ODD), attention deficit hyperactivity disorder
(ADHD), generalized anxiety disorder (GAD), and depression.
Moreover, we collected data on the type of substance abuse
and calculated the percentage of cannabis, alcohol, and other
drug use in the sample. Finally, we retrieved information on
the severity of cannabis use. Using the benchmark established
by Hendriks, Van Der Schee, and Blanken (2011) and also
used in Rigter et al. (2013), we retrieved the percentage of
adolescents who reported using substances more than 64 of
the 90-day intake assessment period.
For the treatment descriptors, we distinguished three treat-
ment comparison groups: CBT, GT, and CT. We assigned the
comparison group in the Rigter et al. (2013) overarching
multisite trial to the CBT category, because in all sites the
comparison group consisted of either CBT alone or CBT
complemented with other treatment approaches. CT was coded
if more than one treatment module was combined. The
following combinations were found: CBT and motivational
enhancement therapy sessions (Dennis et al., 2004), CBT with
GT and family interventions (The Adolescent Community
Reinforcement Approach; Dennis et al., 2004), CBT, motiva-
tion enhancement therapy sessions, and family interventions
(Family Support Network; Dennis et al., 2004), and CBT and
GT (Residential Substance Abuse Treatment; Liddle, 2015).
Finally, treatment duration was collected.
For the research design characteristics, we coded whether
studies were conducted by the developers of the treatment or
by others (developers and nondevelopers), to test the assump-
tion that studies carried out by the developers yield higher
effect sizes. In this category, overall sample size, treatment
group size, comparison group size, and study follow-up dura-
tion were analyzed as well.
For the manuscript characteristics, we coded the year of
publication. If the manuscript had not been published, we
used the year that the manuscript was written. Finally, the
impact factor of the journal in which the manuscript was
published was inventoried.
Inter-rater reliability
The first and third author coded the effect sizes and study
characteristics. Reliability of the coding scheme was examined
by having a subset of the study characteristics coded by two
research assistants. Ten manuscripts were randomly selected.
Observed outcome
2.18
4.28
6.38
8.48
10.58
–0.50 0.00 0.50 1.00
Figure 2 Funnel plot of effect sizes
©2017 Association for Child and Adolescent Mental Health.
4Thimo M. van der Pol et al.

Inter-rater agreement was analyzed for each of the study
outcomes and study characteristics by calculating the per-
centage of agreement for all study characteristics, Kappa for
categorical variables and intraclass correlation for interval and
ratio variables. The inter-rater reliability was good, with Kappa
ranging from .93 (93% agreement) for comparison group to
1.00 for outcome, geographic location, and independence of
researchers (100% agreement); intraclass correlations ranged
from .96 for follow-up period (91% agreement) to 1.00 for effect
size (91% agreement), SES (91% agreement), average age
(100% agreement), and percentage of males (100% agreement).
Analyses
For each study outcome, a Cohen’s deffect size was coded or
calculated. When not provided, formulae provided by Lipsey
and Wilson (2001) to transform test statistics into Cohen’s dor
to calculate don the basis of means and standard deviations
were used. Effect sizes of d=0.20, d=0.50, and d=0.80 were
considered as small, medium, and large group differences,
respectively, whereas d=0.00 would indicate no difference
between the experimental and comparison groups (Cohen,
1988). Using standardized z-values larger than 3.29 or smaller
than 3.29 (Tabachnick & Fidell, 1989), no outliers were
identified. Each continuous moderator variable was centered
around its mean. For the categorical variables, we made
dichotomous dummy variables. The extent of the variation in
effect sizes was examined by conducting a test for homogeneity
of effect sizes.
Independence of study results is desirable when conducting
a meta-analysis in order to prevent a particular study being
weighted more strongly than others (Lipsey & Wilson, 2001;
Mullen, 2013; Rosenthal, 1991). To deal with dependency of
study results, we applied a three-level random effects model
(Cheung, 2014; Van den Noortgate, L
opez-L
opez, Mar
ın-
Mart
ınez, & S
anchez-Meca, 2013). This model accounts for
three sources of variance: sampling variance (level 1 variance),
variance between effect sizes from the same study (level 2
variance), and variance between studies (level 3 variance; Hox,
2002; Van den Noortgate et al., 2013). A three-level random
effects model, therefore, accounts for the hierarchical struc-
ture of the data in which the effect sizes or study results (the
lowest level) are nested within studies (the highest level).
A likelihood ratio test was used to examine between-study and
within-study heterogeneity (Raudenbush & Bryk, 2002).
Moderator analyses were conducted by extending the model
with study and effect size characteristics. For these models
including moderators, an omnibus test of the fixed-model
parameters was conducted, which tests the null hypothesis
that the group mean effect sizes are equal. The Knapp and
Hartung (2003) adjustment was applied to control for Type I
error rates. We used the metafor package (Viechtbauer, 2010)
for the R environment (Version 3.2.3; R Core Team, 2015) for
modeling a three-level random effects model as described by
Van den Noortgate et al. (2013). Parameters were estimated
using the restricted maximum-likelihood procedure.
Results
The 19 manuscripts included in the meta-analysis
reported on eight studies and presented 61 effect
sizes. These studies examined 1,488 adolescents in
total, of whom 699 received MDFT and 789 CBT, GT,
or CT. The effect sizes from the individual studies
ranged from d=0.62 to 1.16. An overview of the
characteristics of the 19 manuscripts and the 61
effect sizes is presented in Table 1.
Results indicated that the overall mean effect size
for MDFT was beneficial compared to adolescents
receiving another form of therapy, d=0.24, p<.01.
For effect sizes, variance between effect sizes within
studies (level 2 variance), r
2
=.012, v
2
(1) =23.00,
p=.14, was nonsignificant, whereas variation
between studies (level 3 variance), r
2
=.048,
v
2
(1) =32.77, p<.001, was significant, resulting in
the examination of the extent to which potential
moderators explained effect size variability.
Moderator analyses
Table 2 summarizes the results of the moderator
analyses. Two moderators yielded a positive contri-
bution to effect size. Percentage of severe substance
abusers in the study sample was associated with
larger effects favoring MDFT, F(1, 45) =6.150,
77%
4%
8% 8%
4%
0%
10%
20%
30%
40%
50%
60%
70%
80%
90%
100%
0 0.01 0.02 0.03 0.04 0.05
Percentage of test results
p-value
Figure 3 P-curve, testing possible p-hacking. The Observed P-curve includes 26 statistically significant (p<.05) results, of which 22 are
p<.025. There were no nonsignificant results entered [Colour figure can be viewed at wileyonlinelibrary.com]
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 5

Table 1 Description of major characteristics of studies used in the meta-analysis
No. Study Year NNtarget Ncontrol
Comparison
condition
Study
ID
Age
mean Follow-up % Males % Minorities
% Conduct
disorder
% Severe
cannabis
users Outcome measure
Effect
size
1 Dakof et al. 2015 112 55 57 Group 1 16.00 18 88 100 52 48 Substance abuse 0.05
2 2015 54 24 30 Group 1 16.26 18 –100 –100 Substance abuse 0.09
3 2015 58 31 27 Group 1 15.90 18 –100 –0 Substance abuse 0.01
4 2015 112 55 57 Group 1 16.00 18 88 100 52 48 Delinquency 0.14
5 2015 112 55 57 Group 1 16.00 18 88 100 52 48 Externalizing 0.21
6 2015 112 55 57 Group 1 16.00 18 88 100 52 48 Internalizing 0.24
7 Liddle et al. Draft 113 57 56 CT 2 15.36 12 75 88 77 70 Substance abuse 0.05
8 Draft 79 43 36 CT 2 15.33 12 –––100 Substance abuse 0.29
9 Draft 34 14 20 CT 2 15.44 12 ––– 0 Substance abuse 0.62
10 Draft 113 57 56 CT 2 15.36 12 75 88 77 70 Delinquency 0.13
11 Draft 113 57 56 CT 2 15.36 12 75 88 77 70 Externalizing 0.20
12 Draft 113 57 56 CT 2 15.36 12 75 88 77 70 Internalizing 0.11
13 Schaub et al. 2014 450 212 238 CBT 3 16.30 6 85 40 –48 Externalizing 0.05
14 2014 450 212 238 CBT 3 16.30 6 85 40 –48 Internalizing 0.10
15 2014 450 212 238 CBT 3 16.30 6 85 40 –48 Family 0.18
16 Rigter et al. 2013 450 212 238 CBT 3 16.30 6 85 40 –48 Substance abuse 0.25
17 2013 237 108 129 CBT 3 16.39 6 –––100 Substance abuse 0.35
18 2013 213 104 109 CBT 3 16.09 6 ––– 0 Substance abuse 0.11
19 Hendriks et al. 2012 109 55 54 CBT 3 16.80 6 80 28 29 51 Externalizing 0.25
20 2012 109 55 54 CBT 3 16.80 6 80 28 29 51 Internalizing 0.42
21 Hendriks et al. 2011 109 55 54 CBT 3 16.80 6 80 28 29 51 Substance abuse 0.14
22 2011 50 28 22 CBT 3 17.03 6 ––29 100 Substance abuse 0.41
23 2011 53 24 29 CBT 3 17.03 6 ––29 0 Substance abuse 0.04
24 2011 37 21 16 CBT 3 16.80 6 ––84 57 Substance abuse 1.16
25 2011 109 55 54 CBT 3 16.80 6 80 28 29 51 Delinquency 0.00
26 Liddle et al. 2011 154 76 78 CT 4 15.40 3 83 84 43 33 Delinquency 0.30
27 Phan 2011 101 38 63 CT 3 16.29 6 89 33 –57 Substance abuse 0.14
28 2011 58 20 38 CT 3 16.48 6 89 33 –100 Substance abuse 0.19
29 2011 43 18 25 CT 3 16.02 6 89 33 –0 Substance abuse 0.38
30 Grichting 2011 60 30 30 CT 3 16.07 6 92 67 –33 Substance abuse 0.00
31 2011 20 8 12 CT 3 16.15 6 92 67 –100 Substance abuse 0.28
32 2011 40 22 18 CT 3 16.03 6 92 67 –0 Substance abuse 0.21
33 Tossmann 2010 120 59 61 CT 3 16.21 6 83 30 –55 Substance abuse 0.51
34 2010 66 33 33 CT 3 16.42 6 83 30 –100 Substance abuse 0.70
35 2010 54 26 28 CT 3 15.96 6 83 30 –0 Substance abuse 0.26
36 Verbanck 2010 60 30 30 CBT 3 16.60 6 93 37 –65 Substance abuse 0.65
37 2010 39 19 20 CBT 3 16.67 6 93 37 –100 Substance abuse 0.53
38 2010 21 11 10 CBT 3 16.48 6 93 37 –0 Substance abuse 0.83
(continued)
©2017 Association for Child and Adolescent Mental Health.
6Thimo M. van der Pol et al.

Table 1 (continued)
No. Study Year NNtarget Ncontrol
Comparison
condition
Study
ID
Age
mean Follow-up % Males % Minorities
% Conduct
disorder
% Severe
cannabis
users Outcome measure
Effect
size
39 Henderson et al. I 2010 45 22 23 CBT 5 15.53 12 –––100 Substance abuse 0.51
40 2010 178 90 88 CBT 5 15.36 12 81 –– 0 Substance abuse 0.33
41 Henderson et al. II 2010 50 25 25 CT 4 15.52 3 –––100 Substance abuse 0.23
42 2010 104 51 53 CT 4 15.36 3 ––– 0 Substance abuse 0.34
43 Henderson et al. 2009 83 40 43 Group 6 13.73 6 74 84 39 2 Family 0.44
44 Liddle et al. 2009 83 40 43 Group 6 13.73 6 74 97 39 2 Substance abuse 1.07
45 2009 83 40 43 Group 6 13.73 6 74 97 39 2 Delinquency 0.31
46 2009 83 40 43 Group 6 13.73 6 74 97 39 2 Internalizing 0.54
47 Liddle et al. 2008 224 112 112 CBT 5 15.40 12 81 97 50 21 Substance abuse 0.59
48 Hogue et al. 2008 136 74 62 CBT 5 15.50 6 81 80 50 –Substance abuse 0.47
49 2008 136 74 62 CBT 5 15.50 6 81 80 50 –Externalizing 0.56
50 2008 136 74 62 CBT 5 15.50 6 81 80 50 –Internalizing 0.76
51 Hogue et al. 2004 51 25 26 CBT 5 15.20 12 67 75 69 –Substance abuse 0.47
52 2004 51 25 26 CBT 5 15.20 12 67 75 69 –Externalizing 0.62
53 2004 51 25 26 CBT 5 15.20 12 67 75 69 –Internalizing 0.74
54 Dennis et al. I 2004 200 100 100 CT 7 16.00 9 83 39 53 –Substance abuse 0.06
55 Dennis et al. II 2004 200 100 100 CT 7 16.00 9 83 39 53 –Substance abuse 0.26
56 Liddle et al. I 2001 152 47 105 Group 8 15.90 12 80 49 –1 Substance abuse 0.25
57 2001 152 47 105 Group 8 15.90 12 80 49 –1 Externalizing 0.10
58 2001 152 47 105 Group 8 15.90 12 80 49 –1 Family 0.61
59 Liddle et al. II 2001 152 47 105 CT 8 15.90 12 80 49 –1 Substance abuse 0.85
60 2001 152 47 105 CT 8 15.90 12 80 49 –1 Externalizing 0.35
61 2001 152 47 105 CT 8 15.90 12 80 49 –1 Family 0.31
CBT, cognitive behavioral therapy; CT, combined treatment; Group: group therapy.
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 7

Table 2 Results for moderators
Moderator
variables
No. of
studies
No. of
ES b
0
, mean d
a
(95% CI) b
1
(95% CI) Omnibus test
Variance
level 2
b
Variance
level 3
c
Sample descriptors
Outcome characteristics
Substance
abuse (rc)
8 37 0.252 (0.069; 0.436)** F(4, 56) =0.383 .015 .049***
Delinquency 5 5 0.212 (0.049; 0.473) 0.041 (0.273; 0.192)
Externalizing 5 8 0.168 (0.058; 0.393) 0.085 (0.261; 0.092)
Internalizing 5 7 0.297 (0.063; 0.531)* 0.044 (0.143; 0.232)
Family
functioning
3 4 0.252 (0.015; 0.520) 0.000 (0.230; 0.230)
Geographic location
United
States (rc)
7 36 0.226 (0.277; 0.728) F(1, 59) =0.003 .011 .061***
Europe 1 25 0.240 (0.037; 0.443)* 0.015 (0.527; 0.557)
Independence researchers
Yes (rc) 5 38 0.226 (0.277; 0.728) F(1, 59) =0.003 .011 .061***
No 6 23 0.240 (0.037; 0.443)* 0.015 (0.527; 0.557)
Age of the
subjects
8 61 0.224 (0.052; 0.397)* 0.058 (0.240; 0.123) F(1, 59) =0.521 .012 .045***
Socioeconomic
status
5 23 0.410 (0.212; 0.609)*** 0.000 (0.000; 0.000) F(1, 21) =0.626 .018 .029**
% Males 8 52 0.262 (0.102; 0.423)** 0.856 (2.501; 0.789) F(1, 50) =1.093 .011 .039***
% Caucasians 8 54 0.281 (0.080; 0.481)** 0.145 (0.469; 0.759) F(1, 52) =0.225 .008 .062***
% African-
Americans
8 42 0.243 (0.076; 0.411)** 0.429 (0.299; 1.158) F(1, 40) =1.419 .005 .042***
% Hispanics 8 42 0.262 (0.071; 0.454)** 0.040 (0.709; 0.790) F(1, 40) =0.012 .005 .058***
% Asians 8 49 0.232 (0.048; 0.416)* 2.676 (6.430; 11.782) F(1, 47) =0.349 .008 .058***
% Others 8 43 0.226 (0.045; 0.408)* 0.585 (1.812; 0.641) F(1, 41) =0.929 .009 .053***
% Cultural
minority
8 40 0.256 (0.075; 0.437)** 0.139 (0.555; 0.834) F(1, 38) =0.165 .006 .051***
% CD 7 29 0.272 (0.018; 0.525)* 0.736 (0.234; 1.705) F(1, 27) =2.421 .000 .096***
% ODD 3 18 0.349 (0.082; 0.616)* 0.271 (0.879; 1.421) F(1, 16) =0.250 .000 .040
% DBD 2 7 0.494 (0.317; 0.670)*** 1.371 (0.432; 2.311)* F(1, 5) =14.072** .000 .000
% ADHD 5 15 0.199 (0.088; 0.487) 1.118 (5.008; 2.773) F(1, 13) =0.385 .000 .075**
% GAD 4 14 0.223 (0.202; 0.648) 0.051 (3.533; 3.430) F(1, 12) =0.001 .000 .125***
% Depression 6 22 0.273 (0.008; 0.553) 0.320 (3.111; 3.751) F(1, 20) =0.038 .000 .090***
% Cannabis 7 54 0.221 (0.040; 0.402)* 0.236 (0.993; 0.521) F(1, 52) =0.393 .005 .047***
% Alcohol 5 18 0.210 (0.075; 0.472) 0.490 (1.899; 0.919) F(1, 16) =0.543 .000 .079***
% Other drugs 5 18 0.200 (0.108; 0.508) 0.413 (3.727; 2.902) F(1, 16) =0.070 .000 .092***
% Severe
substance
abuse
7 47 0.282 (0.067; 0.496)* 0.264 (0.050; 0.479)* F(1, 45) =6.150** .001 .059**
Treatment descriptors
Comparison condition
Cognitive
behavioral
therapy
2 32 0.281 (0.034; 0.529)* F(2, 58) =0.218 .012 .060***
Combined
treatment (rc)
4 16 0.257 (0.044; 0.469) 0.025 (0.222; 0.172)
Group
therapies
3 13 0.178 (0.087; 0.444) 0.103 (0.420; 0.214)
Duration of
treatment in
months
8 61 0.263 (0.102; 0.423)** 0.059 (0.035; 0.154) F(1, 59) =1.579 .011 .039***
Research design
Total sample
size
8 61 0.246 (0.077; 0.416)** 0.000 (0.001; 0.000) F(1, 59) =2.448 .009 .048***
Sample size
treatment
group
8 61 0.247 (0.079; 0.414)** 0.001 (0.002; 0.000) F(1, 59) =2.374 .009 .047***
Sample size
comparison
group
8 61 0.246 (0.076; 0.417)** 0.001 (0.002; 0.000) F(1, 59) =2.491 .009 .049***
Follow-up
(in months)
8 46 0.237 (0.053; 0.422)* 0.012 (0.043; 0.019) F(1, 44) =0.573 .010 .057***
(continued)
©2017 Association for Child and Adolescent Mental Health.
8Thimo M. van der Pol et al.

p=.017. This suggests that adolescents with more
severe substance abuse benefit more from MDFT
than from the comparison treatments. In addition,
percentage of DBD was positively related to the effect
size, F(1, 5) =14.072, p=.013, indicating that sam-
ples with higher percentages of DBD responded
better to MDFT. Year of publication yielded a trend,
F(1, 59) =3.638, p=.061, showing relatively smal-
ler effects in newer studies.
The effect sizes for the outcome measures sub-
stance abuse, delinquency, externalizing and inter-
nalizing psychopathology, and family functioning
were found to be in the same range, all indicating a
small incremental effect over other established treat-
ments with no significant differences between the
effect sizes for the five outcome categories. Further-
more, for treatment groups, no significant differences
in effect size were found between studies that com-
pared MDFT with CBT and studies that compared
MDFT with CT, respectively GT. The geographic
location where studies were conducted (i.e. Europe
vs. United States) had no impact on study results.
Studies led by the developers of MDFT had similar
outcomes as those led by independent researchers.
No moderating effects were found for adolescents’
age, gender, SES, ethnic background, duration of
therapy, and duration of the follow-up period. More-
over, the rates of depression, GAD, ADHD, CD, and
ODD in the sample, and percentage of cannabis,
alcohol, and other drugs were not associated with
effect size. Finally, study sample sizes and impact
factor had no moderating effect.
Model with multiple moderators
To examine the unique contribution of each moder-
ator to the variance in effect size, a model with
multiple moderators was tested. Variables associ-
ated with effect sizes with a p<.20 in the bivariate
moderator analyses reported above were entered in
the model. To retain sufficient power in the model
with multiple moderators, only the variables for
which the number of effect sizes was at least k=30
were included. The following variables were
included: percentage of adolescents with severe
substance abuse, sample size, and year of publica-
tion. The model was found to be significant,
F(3, 43) =5.779, p=.002, k=47. Two moderators
were significant predictors of effect size: severe
substance abuse, b=.26, p=.016, and year of
publication, b=.09, p=.002. Thus, studies with
a larger proportion of subjects with severe substance
abuse and older studies yielded larger effect sizes,
favoring MDFT. To illustrate the effect of severe
substance abuse in samples, Table 3 (Neyeloff,
Fuchs, & Moreira, 2012) includes a forest plot that
depicts studies with low (0%), moderate (1–99%),
and high (100%) severe substance abusers. The
forest plot illustrates that in general, MDFT gener-
ated larger effect sizes for samples with a higher
percentage of severe cannabis users. The computed
mean effect sizes for relatively low, moderate, and
high severe substance abusers showed that effects of
MDFT were nonsignificant for nonsevere substance
abusers (d=0.09), small for moderate abusers
(d=0.28), and small to moderate for severe sub-
stance abusers (d=0.38).
Discussion
The purpose of this meta-analysis was first to
examine the effectiveness of MDFT, compared to
other (active) treatments, and second to inventory
the effects of severe behavior problems and other
potential moderators. Overall, compared to other
treatments and across outcome categories, MDFT
showed a significant effect size, d=0.24, which
corresponds to a success rate difference (Kraemer
& Kupfer, 2006) of approximately 13%. These find-
ings supporting the effectiveness of MDFT are in line
with the meta-analyses of other multiple systems-
based treatments, such as multisystemic therapy
(Van der Stouwe, Asscher, Stams, Dekovi
c, & Van
Der Laan, 2014). In addition, MDFT was found to be
most effective in adolescents with severe substance
abuse and/or DBD.
Table 2 (continued)
Moderator
variables
No. of
studies
No. of
ES b
0
, mean d
a
(95% CI) b
1
(95% CI) Omnibus test
Variance
level 2
b
Variance
level 3
c
Manuscript characteristics
Year of
publication
8 61 0.213 (0.015; 0.410)* 0.031 (0.063; 0.001) F(1, 59) =3.638
+
.006 .068***
Impact factor 8 61 0.240 (0.062; 0.417)** 0.006 (0.048; 0.036) F(1, 59) =0.084 .012 .053***
No. of studies: number of independent studies; No. of ES: number of effect sizes; mean d, mean effect size; CI, confidence interval; rc,
reference category; ADHD, attention deficit hyperactivity disorder; DBD, disruptive behavior disorder; GAD, generalized anxiety
disorder.
a
For continuous predictors, the mean effect size indicates the mean effect size of a participant with an average value on the
corresponding predictor.
b
Variance between the effect sizes from the same study.
c
Variance between studies.
+
p<.1; *p<.05; **p<.01; ***p<.001.
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 9

Table 3 Foster plot of individual effect sizes 95% confidence intervals
–1 –0.5 0 0.5 1 1.5 2
No. Study Year Effect Size 95% CI Forest Plot
1 Dakof et al. 2015 0.05 –0.32 0.42
2 - 2015 0.09 –0.28 0.46
3 - 2015 0.01 –0.36 0.38
4 - 2015 0.14 –0.31 0.59
5 - 2015 0.21 –0.50 0.92
6 - 2015 0.24 –0.13 0.61
7 Liddle et al. draft 0.05 –0.03 0.13
8 - draft 0.29 –0.08 0.66
9 - draft –0.62 –0.99 –0.25
10 - draft 0.13 0.05 0.21
11 - draft 0.20 0.12 0.28
12 - draft 0.11 –0.26 0.48
13 Schaub et al. 2014 0.05 –0.85 0.95
14 - 2014 0.10 –0.55 0.75
15 - 2014 0.18 –0.33 0.69
16 Rigter et al. 2013 0.25 0.07 0.43
17 - 2013 0.35 0.17 0.53
18 - 2013 0.11 –0.07 0.29
19 Hendriks et al. 2012 0.25 0.07 0.43
20 - 2012 0.42 0.17 0.67
21 Hendriks et al. 2011 0.14 –0.13 0.41
22 - 2011 0.41 0.04 0.78
23 - 2011 –0.04 –0.41 0.33
24 - 2011 1.16 0.79 1.53
25 - 2011 0.00 –0.37 0.37
26 Liddle et al. 2011 0.30 –0.27 0.87
27 Phan 2011 0.14 –0.23 0.51
28 - 2011 –0.19 –0.68 0.30
29 - 2011 0.38 –0.15 0.91
30 Grichting 2011 0.00 –0.53 0.53
31 - 2011 0.28 –0.23
–0.58
0.79
32 - 2011 –0.21 0.16
33 Tossmann 2010 0.51 0.00 1.02
34 - 2010 0.70 –0.20 1.60
35 - 2010 0.26 –0.37 0.89
36 Verbanck 2010 0.65 0.24 1.06
37 - 2010 0.53 –0.02 1.08
38 - 2010 0.83 0.22 1.44
39 Henderson et al. I 2010 0.51 –0.06 1.08
40 - 2010 0.33 –0.22 0.88
41 Henderson et al. II 2010 0.23 –0.36 0.82
42 - 2010 –0.34 –0.63 –0.05
43 Henderson et al. 2009 0.44 –0.01 0.89
44 Liddle et al. 2009 1.07 0.72 1.42
45 - 2009 0.31 0.04 0.58
46 - 2009 0.54 0.27 0.81
47 Liddle et al. 2008 0.59 0.16 1.02
48 Hogue et al. 2008 0.47 0.04 0.90
49 - 2008 0.56 0.13 0.99
50 - 2008 0.76 0.49 1.03
51 Hogue et al. 2004 0.47 0.14 0.80
52 - 2004 0.62 0.27 0.97
53 - 2004 0.74 0.39 1.09
54 Dennis et al. I 2004 –0.06 –0.41 0.29
55 Dennis et al. II 2004 –0.26 –0.61 0.09
56 Liddle et al. I 2001 0.25 –0.10 0.60
57 - 2001 –0.10 –0.45 0.25
58 - 2001 0.61 0.26 0.96
59 Liddle et al. II 2001 0.85 0.28 1.42
60 - 2001 0.35 –0.36 1.06
61 - 2001 0.31 –0.24 0.86
Samples with severe substance abuse
Samples with non-severe substance abuse
©2017 Association for Child and Adolescent Mental Health.
10 Thimo M. van der Pol et al.

This ‘severity gradient’ supported by our finding
that MDFT is more effective for those with high
severity problems, such as severe substance abuse,
is in line with previous research, showing that
adolescents with severe cannabis abuse (Rigter
et al., 2013) and severe cannabis or substance
abusers with comorbid externalizing psychopathol-
ogy benefit most from MDFT (Henderson et al., 2010;
Hendriks, Van Der Schee, & Blanken, 2012). This is
not surprising, as the treatment goals of MDFT have
been designed to serve a broad, heterogeneous group
of adolescents with substance use disorders and
diverse and complex behavior problems (Henderson
et al., 2010; Weisz & Kazdin, 2010). Over the years,
different versions of MDFT have been designed and
tested in different countries, in samples with differ-
ent ages, gender, psychopathology, and in different
settings, including clinical and juvenile justice set-
tings. From our findings, it seems that MDFT is
effective in a variety of settings, and for different
adolescents, however, the largest effects are found
for those with high severity problems. Our finding is
consistent with the risk principle of the Risk-Need-
Responsivity model (Andrews, Bonta, & Hoge, 1990;
Andrews, Bonta, & Wormith, 2006, 2011) which
states that the intensity of interventions should
match recidivism risk: those with increased recidi-
vism risk (i.e. with more severe conditions) should
receive more intensive treatment. Our findings sup-
port the notion that treatment effectiveness of inten-
sive, comprehensive treatment programs is better for
severely affected youths. Specifically, for MDFT this
means that although MDFT is applicable for a broad
spectrum of problems, the treatment appears to have
surplus value for the most severely impaired youth.
In the model with multiple moderators, an effect of
year of publication was found. In early publications,
effect sizes for MDFT were larger than in later
publications. One possible explanation for this find-
ing would be the ‘decline effect’, a term coined by
Ioannidis (2005). He stated that early research is
usually small and may be more likely to produce
positive results supporting the hypotheses examined
than later, larger studies, in which regression to the
mean might occur. However, given that we did not
find a moderating effect of sample size, this expla-
nation is not likely. It is more likely that confounding
moderators, not examined in this meta-analysis,
may explain the effect of publication year. Although
we have coded many study characteristics, data on
features of the intervention, such as different ver-
sions of MDFT, or levels of treatment integrity were
not available, and therefore, we did not examine
these potential moderating characteristics.
Further, effects of MDFT on different treatment
outcomes, including substance abuse, delinquency,
externalizing and internalizing psychopathology,
and family functioning, were about equal in effect
size. This suggests that MDFT affects a broad
range of domains which may be explained by the
multifocused approach of MDFT (Liddle, 2002; Lid-
dle & Rigter, 2013). An important finding, enhancing
the applicability of MDFT is that this therapy
appeared to be similarly effective for boys and girls
and for adolescents with different ages, SES, and
ethnic background, as these were no significant
moderators of the effectiveness of MDFT. With regard
to age, this is not consistent with an earlier study,
which found MDFT to be more effective when the
intervention was aimed at younger adolescents
(Hendriks et al., 2011); however, this study has a
relative small sample size, not representative com-
pared to the current meta-analysis. Some studies
postulate the development of specific interventions
aimed at girls (e.g. Hipwell & Loeber, 2006); the
present meta-analysis found that MDFT is beneficial
for a varied group of male and female adolescents
from different ethnic backgrounds.
To our knowledge, this is the first meta-analysis on
MDFT, using three-level analytic techniques. This
novel three-level analytic method makes it possible
to study the influence of moderators more exten-
sively and increases statistical power, which allowed
us to test the described severity gradient. Another
strength of the present meta-analysis is that we only
included randomized control trials (RCTs) comparing
MDFT with other evidence-based, effective therapies,
which is considered to be the most robust research
design and best equipped to handle threats to a
study’s internal validity (Weisburd, Lum, & Pet-
rosino, 2001; Welsh, Peel, Farrington, Elffers, &
Braga, 2011). Notwithstanding the strength of the
present meta-analysis, our findings should be inter-
preted in the context of some limitations. First, there
is a lack of studies that examined family functioning
as an outcome measure. Family functioning is con-
sidered to be a major focus in the treatment model
for MDFT (Dakof, Cohen, & Duarte, 2009). There-
fore, more studies regarding family functioning are
necessary. Second, although a RCT is considered to
be the best research design, there are scholars postu-
lating that due to the selection procedure of RCTs, we
should be cautious to generalize the findings in exper-
imental settings to routine youth care (Waldron &
Turner, 2008). Within clinical samples, there is gen-
erally much heterogeneity in adolescent characteris-
tics (e.g. age, substance abuse, delinquency, and
psychiatric comorbidity). Therefore, adolescent sub-
groups, within these clinical samples, may differ
considerably in treatment outcome (Chan, Dennis, &
Funk, 2008; Daudin et al., 2010). Finally, in the
current meta-analysis, we were unable to examine
various types of criminal behavior, which could gen-
erate additional insight. In the five studies that
reported delinquency, only one study analyzed the
influence of MDFT on various types of criminal behav-
ior (e.g. person crimes, theft; Dakof et al., 2015).
For future research, we strongly suggest other
established treatments addressing substance abus-
ing adolescents with comorbid behavior problems to
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 11

test the severity gradient for substance abuse,
externalizing disorders and possible other important
variables, to be able to better match treatment with
the characteristics of an adolescent (Bell, Marcus, &
Goodlad, 2013; Leijten et al., 2015).
Specific for MDFT, one of the directions of future
research should be to intensively investigate family
functioning as a moderator of the effectiveness of
MDFT. Some studies addressed this quintessential
topic for MDFT (Henderson, Rowe, Dakof, Hawes, &
Liddle, 2009; Schmidt, Liddle, & Dakof, 1996;
Shelef, Diamond, Diamond, & Liddle, 2005). Never-
theless, more research on family functioning is
necessary. A further research topic of interest is to
study the impact of MDFT on different indices of
criminal behaviors (Dakof et al., 2015). This type of
research could provide more precise information for
which type of adolescents MDFT is the most effective.
Moreover, MDFT is an intensive treatment, which is
considered to be more expensive than most alterna-
tive therapies, and therefore, conducting cost-effec-
tiveness studies carries substantial relevance.
Practical implications of the present meta-analysis
are that treatment delivery systems should aim to
provide different treatment modules matching the
severity of problem behaviors of the youth. MDFT
has addressed this issue extensively, by developing
diverse modules and researching varied subgroups
of adolescents (Brown & Zucker, 2015; Weisz &
Kazdin, 2010); most other treatments targeting this
heterogeneous group of adolescents are advised to
follow suit. The feasibility of this suggestion can be
debated; however, for society, the improvement of
the quality of care for this group of adolescents is of
major importance.
Finally, MDFT, although suitable for a broad
spectrum of adolescents with behavior problems,
may be most suitable for adolescents with severe
problems, severe substance abuse, and DBD in
particular. Furthermore, this finding could indicate
that other less intensive and expensive treatments,
for example, individual CBT, may be as appropriate
for addressing SUD and comorbid psychopathology
in adolescents with less severe problem behavior.
In summary, we conclude that MDFT has an incre-
mental, 13% advantage over other established treat-
ments. As a unique asset, MDFT can be successfully
deployed in male and female adolescents from diverse
ethnic backgrounds in a variety of settings, with SUD,
delinquency, and diverse comorbid conditions,
notwithstanding their age. Furthermore, MDFT was
found to be more effective for adolescents with severe
problem behavior. As such, MDFT can be regarded as
a valuable therapy, especially when treating the most
challenging group of youth.
Acknowledgements
No external funding was received as part of this work.
The authors have declared that they have no competing
or potential conflicts of interest. They thank Maya
Boustani and Henk Rigter for their support and helping
in our search for relevant articles. The authors are also
grateful to Tale Evenhuis for conducting the literature
search and Craig Henderson for his support and help in
providing the additional data. Finally, they thank
Annemieke ter Harmsel and Lise Swinkels for helping
to code the studies.
Correspondence
Thimo M. van der Pol, Department of Child and
Adolescent Psychiatry, VU University Medical Centre,
Meibergdreef 5, 1105 AZ, Amsterdam, The Netherlands;
Email: thimovanderpol@gmail.com
Key points
•Multidimensional family therapy (MDFT) is a promising treatment for adolescents with substance abuse and
comorbid behavior problems.
•Multidimensional family therapy is effective for adolescents with substance use disorders, delinquency, and
comorbid behavior problems.
•Compared with other treatments, MDFT is more effective for adolescents with disruptive behavior disorder.
•Compared with other treatments, MDFT is most effective for adolescents with severe substance abuse.
•It is important to match the treatment with specific characteristics of the adolescents, such as extent of
impairment.
References
Anderson, K.G., Ramo, D.E., Schulte, M.T., Cummins, K., &
Brown, S.A. (2007). Substance use treatment outcomes for
youth: Integrating personal and environmental predictors.
Drug Alcohol Dependence,88,42–48.
Anderson, K.G., Tapert, S.F., Moadab, I., Crowley, T.J., &
Brown, S.A. (2007). Personality risk profile for conduct
disorder and substance use disorders in youth. Addictive
Behaviors,32, 2377–2382.
Andrews, D.A., Bonta, J., & Hoge, R.D. (1990). Classification
for effective rehabilitation: Rediscovering psychology. Crim-
inal Justice and Behavior,17,49–52.
Andrews, D.A., Bonta, J., & Wormith, J.S. (2006). The recent
past and near future of risk and/or need assessment. Crime
and Delinquency,52,7–27.
©2017 Association for Child and Adolescent Mental Health.
12 Thimo M. van der Pol et al.

Andrews, D.A., Bonta, J., & Wormith, J.S. (2011). The risk-
need-responsivity (RNR) model does adding the good lives
model contribute to effective crime prevention? Criminal
Justice and Behavior,38, 735–755.
Assink, M., Van Der Put, C.E., Hoeve, M., De Vries, S.L.,
Stams, G.J.J.M., & Oort, F.J. (2015). Risk factors for
persistent delinquent behavior among juveniles: A meta-
analytic review. Clinical Psychology Review,42,47–61.
Baldwin, S.A., Christian, S., Berkeljon, A., & Shadish, W.R.
(2012). The effects of family therapies for adolescent delin-
quency and substance abuse: A meta-analysis. Journal of
Marital and Family Therapy,38, 281–304.
Bell, E.C., Marcus, D.K., & Goodlad, J.K. (2013). Are the parts
as good as the whole? A meta-analysis of component
treatment studies. Journal of Consulting and Clinical Psy-
chology,81, 722.
Bronfenbrenner, U. (1979). Contexts of child rearing: Problems
and prospects. American Psychologist,34, 844.
Brook, J.S., Whiteman, M.M., & Finch, S. (1992). Childhood
aggression, adolescent delinquency, and drug use: A longitu-
dinal study. The Journal of Genetic Psychology,153, 369–383.
Brown, S.A., & Zucker, R.A. (2015). The Oxford handbook of
adolescent substance abuse. New York: Oxford University
Press.
Carr, A. (2009). The effectiveness of family therapy and
systemic interventions for child-focused problems. Journal
of Family Therapy,31,3–45.
Chan, Y., Dennis, M.L., & Funk, R.R. (2008). Prevalence and
comorbidity of major internalizing and externalizing prob-
lems among adolescents and adults presenting to substance
abuse treatment. Journal of Substance Abuse Treatment,34,
14–24.
Cheung, M.W.-L. (2014). Modeling dependent effect sizes with
three-level meta-analyses: A structural equation modeling
approach. Psychological Methods,19, 211.
Cohen, J. (1988). Statistical power analysis for the behavioral
sciences. New York: Academic.
Colins, O., Vermeiren, R., Vahl, P., Markus, M., Broekaert, E.,
& Doreleijers, T. (2011). Psychiatric disorder in detained
male adolescents as risk factor for serious recidivism. The
Canadian Journal of Psychiatry,56,44–50.
Cuellar, A.E., McReynolds, L.S., & Wasserman, G.A. (2006). A
cure for crime: Can mental health treatment diversion
reduce crime among youth? Journal of Policy Analysis and
Management,25, 197–214.
Dakof, G.A., Cohen, J.J.B., & Duarte, J. (2009). Increasing
family reunification for substance-abusing mothers and
their children: Comparing two drug court interventions in
Miami. Juvenile and Family Court Journal,60,11–23.
Dakof, G.A., Henderson, C.E., Rowe, C.L., Boustani, M.,
Greenbaum, P.E., Wang, W., ... & Liddle, H.A. (2015). A
randomized clinical trial of family therapy in juvenile drug
court. Journal of Family Psychology,29, 232.
Daudin, M., Cohen, D., Edel, Y., Bonnet, N., Bodeau, N.,
Consoli, A., ... & Guil
e, J. (2010). Psychosocial and clinical
correlates of substance use disorder in an adolescent
inpatient psychiatric population. Journal of the Canadian
Academy of Child and Adolescent Psychiatry,19, 264.
Dennis, M., Godley, S.H., Diamond, G., Tims, F.M., Babor, T.,
Donaldson, J., ... & Funk, R. (2004). The Cannabis Youth
Treatment (CYT) Study: Main findings from two randomized
trials. Journal of Substance Abuse Treatment,27, 197–213.
Egger, M., Smith, G.D., Schneider, M., & Minder, C. (1997).
Bias in meta-analysis detected by a simple, graphical test.
BMJ,315, 629–634.
Filges, T., Andersen, D., & Jørgensen, A.K. (2015). Effects of
Multidimensional Family Therapy (MDFT) on nonopioid
drug abuse: A systematic review and meta-analysis.
Research on Social Work Practice,1,1–16.
Grella, C.E., Hser, Y., Joshi, V., & Rounds-Bryant, J. (2001).
Drug treatment outcomes for adolescents with comorbid
mental and substance use disorders. The Journal of Nervous
and Mental Disease,189, 384–392.
Grichting, E., Haug, S., Nielsen, P., & Schaub, M. (2011).
INCANT Hauptstudie. Nationaler Schlussbericht Schweiz.
Zurich, Switzerland: Institut f€
ur Sucht- und Gesundheits-
forschung.
Haley, J. (1976). Problem-solving therapy: New strategies for
effective family therapy. San Francisco: Jossey-Bass.
Hall, W.D., Patton, G., Stockings, E., Weier, M., Lynskey, M.,
Morley, K.I., & Degenhardt, L. (2016). Why young people’s
substance use matters for global health. The Lancet Psychi-
atry,3, 265–279.
Henderson, C.E., Dakof, G.A., Greenbaum, P.E., & Liddle, H.A.
(2010). Effectiveness of multidimensional family therapy
with higher severity substance-abusing adolescents: Report
from two randomized controlled trials. Journal of Consulting
and Clinical Psychology,78, 885–897.
Henderson, C.E., Rowe, C.L., Dakof, G.A., Hawes, S.W., &
Liddle, H.A. (2009). Parenting practices as mediators of
treatment effects in an early-intervention trial of multidi-
mensional family therapy. The American Journal of Drug and
Alcohol Abuse,35, 220–226.
Hendriks, V., Van Der Schee, E., & Blanken, P. (2011).
Treatment of adolescents with a cannabis use disorder:
Main findings of a randomized controlled trial comparing
multidimensional family therapy and cognitive behavioral
therapy in The Netherlands. Drug Alcohol Dependence,119,
64–71.
Hendriks, V., Van Der Schee, E., & Blanken, P. (2012).
Matching adolescents with a cannabis use disorder to
multidimensional family therapy or cognitive behavioral
therapy: Treatment effect moderators in a randomized
controlled trial. Drug Alcohol Dependence,125, 119–126.
Hipwell, A.E., & Loeber, R. (2006). Do we know which
interventions are effective for disruptive and delinquent
girls? Clinical Child and Family Psychology Review,9, 221–
255.
Hoeve, M., McReynolds, L.S., & Wasserman, G.A. (2013).
Service referral for juvenile justice youths: Associations with
psychiatric disorder and recidivism. Administration and
Policy in Mental Health,41, 379–389.
Hox, J.J. (2002). Multilevel analysis: Techniques and applica-
tions. Mahwah, NJ: Lawrence Erlbaum.
Ioannidis, J.P. (2005). Why most published research findings
are false. Chance,18,40–47.
Johnston, S.C., & Hauser, S.L. (2008). Peer review at National
Institutes of Health: Small steps forward. Annals of Neurol-
ogy,64, A15–A17.
Knapp, G., & Hartung, J. (2003). Improved tests for a random
effects meta-regression with a single covariate. Statistics in
Medicine,22, 2693–2710.
Kraemer, H.C., & Kupfer, D.J. (2006). Size of treatment effects
and their importance to clinical research and practice.
Biological Psychiatry,59, 990–996.
Leijten, P., Dishion, T.J., Thomaes, S., Raaijmakers, M.A.,
Orobio de Castro, B., & Matthys, W. (2015). Bringing
parenting interventions back to the future: How randomized
microtrials may benefit parenting intervention efficacy.
Clinical Psychology: Science and Practice,22,47–57.
Liddle, H.A. (1999). Theory development in a family-based
therapy for adolescent drug abuse. Journal of Clinical Child
Psychology,28, 521–532.
Liddle, H.A. (2002). Multidimensional Family Therapy for
adolescent cannabis users. Cannabis youth treatment (CYT)
series (Vol. 5). Rockville, MD: Center for Substance Abuse
Treatment, Substance Abuse and Mental Health Services
Administration.
Liddle, H.A. (2015). Multidimensional family therapy as a
community-based alternative to residential drug treatment
for multiply-diagnosed adolescents: a randomized controlled
trial. Manuscript submitted for publication.
©2017 Association for Child and Adolescent Mental Health.
The effectiveness of MDFT 13

Liddle, H.A., & Rigter, H. (2013). How developmental research
and contextual theory drive clinical work with adolescents
with addiction. Harvard Review of Psychiatry,21, 200–204.
Lipsey, M.W., & Wilson, D.B. (2001). Practical meta-analysis.
Thousand Oaks, CA: Sage.
Lodewijks, H.P., De Ruiter, C., & Doreleijers, T.A. (2010). The
impact of protective factors in desistance from violent
reoffending: A study in three samples of adolescent offend-
ers. Journal of Interpersonal Violence,25, 568–587.
Loeber, R., Stouthamer-Loeber, M., & Raskin White, H. (1999).
Developmental aspects of delinquency and internalizing
problems and their association with persistent juvenile
substance use between ages 7 and 18. Journal of Clinical
Child Psychology,28, 322–332.
McReynolds, L.S., & Wasserman, G.A. (2011). Self-injury in
incarcerated juvenile females: Contributions of mental
health and traumatic experiences. Journal of Traumatic
Stress,24, 752–755.
Merikangas, K.R., He, J., Burstein, M., Swanson, S.A.,
Avenevoli, S., Cui, L., ... & Swendsen, J. (2010). Lifetime
prevalence of mental disorders in US adolescents: Results
from the National Comorbidity Survey Replication-Adoles-
cent Supplement (NCS-A). Journal of the American Academy
of Child and Adolescent Psychiatry,49, 980–989.
Minuchin, S. (1974). Families and family therapy. Oxford, UK:
Harvard University Press.
Moffit, R. (1993). Solving the health care crisis: A prescription
for reform. Journal of the Medical Association of Georgia,82,
293–296.
Monahan, G. (2003). An international perspective on the rights
of juvenile offenders or “children living in exceptionally
difficult conditions”. Substance Use and Misuse,38, 1531–
1538.
Mullen, B. (2013). Advanced basic meta-analysis: Version
1.10. Hilsdale, NJ: Erlbaum.
Neyeloff, J.L., Fuchs, S.C., & Moreira, L.B. (2012). Meta-
analyses and Forest plots using a Microsoft excel spread-
sheet: Step-by-step guide focusing on descriptive data
analysis. BMC Research Notes,5, 52.
Phan, O. (2011). 
Etude de comparaison de l’efficacit
edela
MDFT (multidimensional family therapy) versus la TAU
(treatment as usual) dans le traitement de l’abus et de la
d
ependance au cannabis. MD/PhD thesis, Universit
e Pierre
et Marie Curie, Paris.
R Core Team (2015). R: A languageand environment for statistical
computing. Vienna, Austria: R Foundation for Statistical
Computing (Retrieved from http://www.Rproject.org/).
Raudenbush, S.W., & Bryk, A.S. (2002). Hierarchical linear
models: Applications and data analysis methods. Thousand
Oaks, CA: Sage.
Rigter, H., Henderson, C.E., Pelc, I., Tossmann, P., Phan, O.,
Hendriks, V., ...& Rowe, C.L. (2013). Multidimensional family
therapy lowers the rate of cannabis dependence in adoles-
cents: A randomised controlled trial in Western European
outpatient settings. Drug Alcohol Dependence,130,85–93.
Rigter, H., Pelc, I., Tossmann, P., Phan, O., Grichting, E.,
Hendriks, V., & Rowe, C. (2010). INCANT: A transnational
randomized trial of multidimensional family therapy versus
treatment as usual for adolescents with cannabis use
disorder. BMC Psychiatry,10, 28.
Rosenthal, R. (1979). The file drawer problem and tolerance for
null results. Psychological Bulletin,86, 638–641.
Rosenthal, R. (1991). Meta-analytic procedures for social
research. Newbury Park, CA: Sage.
Rosenthal, R. (1995). Writing meta-analytic reviews. Psycho-
logical Bulletin,118, 183–192.
Rothstein, H.R. (2008). Publication bias as a threat to the
validity of meta-analytic results. Journal of Experimental
Criminology,4,61–81.
Schmidt, S.E., Liddle, H.A., & Dakof, G.A. (1996). Changes in
parenting practices and adolescent drug abuse during
multidimensional family therapy. Journal of Family Psychol-
ogy,10,12–27.
Shelef, K., Diamond, G.M., Diamond, G.S., & Liddle, H.A.
(2005). Adolescent and parent alliance and treatment out-
come in multidimensional family therapy. Journal of Con-
sulting and Clinical Psychology,73, 689–698.
Simonsohn, U., Nelson, L.D., & Simmons, J.P. (2014). P-curve:
A key to the file-drawer. Journal of Experimental Psychology:
General,143, 534.
Sutton, A.J., Duval, S.J., Tweedie, R.L., Abrams, K.R., &
Jones, D.R. (2000). Empirical assessment of effect of pub-
lication bias on meta-analyses. BMJ,320, 1574–1577.
Tabachnick, B.G., & Fidell, L.S. (1989). Using multivariate
statistics. New York: Harper Collins.
Tanner-Smith, E.E., Wilson, S.J., & Lipsey, M.W. (2013). The
comparative effectiveness of outpatient treatment for ado-
lescent substance abuse: A meta-analysis. Journal of Sub-
stance Abuse Treatment,44, 145–158.
Tossmann, P., & Jonas, B. (2010). Ergebnisbericht der kontrol-
lierten Behandlungsstudie INCANT [Final report on the
controlled treatment study INCANT]. Berlin, Germany:
delphi-Gesellschaft f€
ur Forschung, Beratung und Projekte-
nentwicklung mbH.
Van den Noortgate, W., L
opez-L
opez, J.A., Mar
ın-Mart
ınez, F.,
&S

anchez-Meca, J. (2013). Three-level meta-analysis of
dependent effect sizes. Behavior Research Methods,45, 576–
594.
Van der Stouwe, T., Asscher, J.J., Stams, G.J.J.M., Dekovi
c,
M., & Van Der Laan, P.H. (2014). The effectiveness of
multisystemic therapy (MST): A meta-analysis. Clinical Psy-
chology Review,34, 468–481.
Van IJzendoorn, M.H. (1998). Meta-analysis in early childhood
education: Progress and problems. In B. Spodek A.D.
Pellegrini O.N. Saracho (Eds.), Issues in early childhood
educational research. Yearbook in early childbook education
(pp. 156–176). New York: Teachers College Press.
Verbanck, P., Pelc, I., Glaesener, M., Gladsteen, Y., Spapen, P.,
van Gerwen, K., ... & Angelidis, T. (2010). INCANT: Eindac-
tiviteitenverslag. Periode: November 2006 –April 2010
[INCANT final report. Period: November 2006 –April 2010].
Brussels, Belgium: CHU Brugmann, Instelling voor Psychi-
atrie en Medische Psychologie (also available in French).
Viechtbauer, W. (2010). Conducting meta-analyses in R with
the metafor package. Journal of Statistical Software,36,1–
48.
Von Sydow, K., Retzlaff, S., Beher, S., Haun, M.W., &
Schweitzer, J. (2013). The efficacy of systemic therapy for
childhood and adolescent externalizing disorders: A system-
atic review of 47 RCT. Family Process,52, 576–618.
Waldron, H.B., & Turner, C.W. (2008). Evidence-based psy-
chosocial treatments for adolescent substance abuse. Journal
of Clinical Child and Adolescent Psychology,37, 238–261.
Wasserman, G.A., McReynolds, L.S., Fisher, P., & Lucas, C.
(2003). Psychiatric disorders in incarcerated youths. Journal
of the American Academy of Child and Adolescent Psychia-
try,42, 1011.
Weisburd, D., Lum, C.M., & Petrosino, A. (2001). Does
research design affect study outcomes in criminal justice?
The Annals of the American Academy of Political and Social
Science,578,50–70.
Weisz, J.R., & Kazdin, A.E. (2010). Evidence-based psy-
chotherapies for children and adolescents. New York: Guil-
ford Press.
Welsh, B.C., Peel, M.E., Farrington, D.P., Elffers, H., & Braga,
A.A. (2011). Research design influence on study outcomes in
crime and justice: A partial replication with public area
surveillance. Journal of Experimental Criminology,7, 183–
198.
Accepted for publication: 21 November 2016
©2017 Association for Child and Adolescent Mental Health.
14 Thimo M. van der Pol et al.