Children and Youth Services Review 34 (2012) 748–757

Contents lists available at SciVerse ScienceDirect

Children and Youth Services Review

journal homepage: www.elsevier.com/locate/childyouth

The data behind the dissemination: A systematic review of trauma-focused cognitive

behavioral therapy for use with children and youth
Colleen E. Cary ⁎, J. Curtis McMillen
University of Chicago, School of Social Service Administration, Chicago, IL, United States

a r t i c l e i n f o a b s t r a c t

Article history: Trauma-Focused Cognitive Behavioral Therapy (TF-CBT) is one of the most widely disseminated mental
Received 21 July 2011 health interventions for children and youth. The purpose of this study is to systematically review the evi-
Received in revised form 30 December 2011 dence of TF-CBT's ability to reduce symptoms of post-traumatic stress, depression and behavior problems
Accepted 1 January 2012
in children and youth who have survived trauma. A search was conducted to locate studies that evaluated
Available online 14 January 2012
TF-CBT or interventions highly similar to TF-CBT. Ten studies (twelve articles) were selected for inclusion
Keywords:
in three sets of meta-analyses. Findings were consistent amongst meta-analyses; pooled estimates were sim-
TF-CBT ilar whether we were analyzing the effects of interventions that were highly similar to TF-CBT, or if we were
Post-traumatic stress exclusively analyzing the effects of the branded intervention. Results show that there is a significant differ-
Depression ence between the TFCBT condition and comparison conditions in its ability to reduce symptoms of PTSD
Behavior problems (g = .671), depression (g = .378) and behavior problems (g = .247) immediately after treatment completion.
This difference held for PTSD at twelve months after treatment completion (.389) but did not hold for depres-
sion or behavior problems. There was not a significant difference between the TF-CBT condition and alterna-
tive active control conditions immediately after treatment completion. Therefore, TF-CBT is an effective
intervention for the treatment of PTSD in youth.
© 2012 Elsevier Ltd. All rights reserved.

1. Introduction who work with traumatized children tend to be most familiar with
the trauma-focused cognitive behavioral intervention developed by
Trauma-Focused Cognitive Behavioral Therapy (TF-CBT) is one of Cohen et al. (2006). We refer to this as the branded version of TF-
the most widely disseminated mental health interventions for chil- CBT because it has been manualized and widely disseminated in
dren and youth (Cohen and Mannarino, 2008; Cohen, Mannarino, this form. It is a highly structured, conjoint parent/child intervention,
and Deblinger, 2006; Saunders, 2011; Saunders, Smith, and Best, consisting of sequential 90-minute weekly sessions. A trained clini-
2010). Despite its popularity, a systematic review of its effects has cian moves the client through a series of 8 components, pacing the
not yet been published. Systematic reviews are unique in their ability progression of the treatment with the client's clinical readiness. The
to reveal the overall effects of interventions, pooling and analyzing components include: psychoeducation and parenting skills (P), relax-
the results of every trial in which an intervention has been evaluated ation (R), affective expression and regulation (A), cognitive coping
while considering the quality of each of those trials. The purpose of (C), trauma narrative development and processing (T), in vivo gradu-
this study is to systematically review the evidence of TF-CBT's al exposure (I), conjoint parent/child sessions (C) and enhancing
ability to reduce symptoms of post-traumatic stress, depression and safety/future development (E). Together these components comprise
behavior problems in children and youth who have survived at least the P.R.A.C.T.I.C.E. acronym.
one traumatic event. This branded version has been actively disseminated. In addition
to the hardback treatment manual published by Guilford (2006),
there is a web-based training program maintained by the Medical
2. Background University of South Carolina (TF-CBT.musc.edu) that, as of May
2011, had 90,970 registered users (Saunders, 2011), including clini-
2.1. What is TF-CBT? cians from more than 111 countries (Saunders et al., 2010). The treat-
ment developers have maintained an active training schedule for a
While there are a number of trauma-focused interventions for number of years, supplemented by a cadre of sanctioned train-the-
children that employ cognitive treatment components, clinicians trainer clinicians and learning collaboratives (Cohen and Mannarino,
2008). Much of this work has been promoted and funded through
⁎ Corresponding author. the National Child Traumatic Stress Network (NCTSN), a program
E-mail address: cecary@gmail.com (C.E. Cary). established by Congress in 2000 in the interest of linking traumatized

0190-7409/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.childyouth.2012.01.003
 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757 749

children and their families with effective resources including and individual trauma-focused cognitive behavioral interventions
evidence-based interventions. outperformed waitlist/community treatment conditions in reducing
This branded version of TF-CBT has a substantial history, as it is a posttrauma symptoms, performed as well as eye-movement desensi-
combination of and expansion upon earlier trauma-focused interven- tization and reprocessing therapy (EMDR) and outperformed other
tions developed by the Cohen/Mannarino and Deblinger teams to non-EMDR treatments (Bisson & Andrew, 2009). Second, there have
treat child sexual abuse survivors (Cohen and Mannarino, 1993, been a number of randomized trials of both the branded TF-CBT and
1996a, 1996b, 1998, 2000; Cohen, Mannarino, and Staron, 2005; other trauma focused cognitive interventions that have shown posi-
Deblinger and Heflin, 1996; Deblinger, McLeer, and Henry, 1990; tive outcomes. The TF-CBT website maintained at the Medical Univer-
Deblinger, Stauffer, and Steer, 2001; Stauffer and Deblinger, 1996). sity of South Carolina lists nine randomized trials with positive
The prior interventions used by the Cohen/Mannarino team went by outcomes (http://tfcbt.musc.edu/resources.php?p=5) and 3 trials of
a variety of names (Structured Parent Counseling–Child Psychothera- other interventions similar to branded TF-CBT. The TF-CBT book
py (SPC–CP), Cognitive-Behavioral Therapy adapted for Sexually cites five randomized controlled trials with positive outcomes
Abused Pre-school children (CBT-SAP) and Sexual Abuse-Specific Cog- (Cohen et al., 2006), all of which were included on the website.
nitive Behavioral Therapy (SAS-CBT)), but all of the earlier forms of A number of organizations have given the branded version of TF-
this intervention were cognitive-behavioral in nature. While differing CBT their highest endorsements. In their report sponsored by the
slightly, they shared a focus on (1) exploring the impact of sexual U.S. Department of Justice, Saunders, Berliner, and Hanson (2002)
abuse on the family, (2) developing a sense of self-efficacy in the sur- reviewed the research on 24 interventions for child maltreatment.
vivor of sexual abuse and (3) encouraging an understanding of how Only one, TF-CBT, received their highest classification rating, “well-
the experience of abuse was carried out in subsequent behaviors and supported, efficacious”. The Kauffman Best Practices Project, con-
relationships (Cohen et al., 2006). Their 1993 study employing SCP– ducted by the Kauffman Best Practices (2004), similarly gave TF-CBT
CP provided the earliest foundation for the intervention as it exists the most rigorous classification of all the interventions they evaluat-
today (Cohen and Mannarino, 1993). Their next round of trials ed, considering it the “best practice” in the field of child abuse treat-
(Cohen and Mannarino, 1996a, 1996b, 1997) used the same interven- ment. The California Evidence Based Clearinghouse for Child
tion, but with a new name (SBT-SAP). These interventions were large- Welfare gave TF-CBT it's most rigorous ranking, a 1, asserting that it
ly based on cognitive reframing and included both psychoeducation is “strongly supported by research evidence” (California Evidence-
and caregiver participation. Cognitive reframing remained the central Based Clearinghouse for Child Welfare, 2011). The National Registry
element of SAS-CBT, the intervention tested in their 1998 trial, but this of Evidence-Based Programs and Practices (NREPP), a sector of the
version also included a stress management component. The interven- US Dept of Health and Human Services (SAMHSA), gave TF-CBT be-
tion did not, at that stage, include structured formalized exposure, al- tween a 3.6 and a 3.8 out of 4.0 possible points on its ability to effec-
though the abuse was discussed; the later inclusion of this important tively treat PTSD, depression and behavior problems and a 3.6 out of
component was largely a result of Deblinger's earlier work. 4.0 on its quality of research rating (SAMHSA, 2008).
Deblinger's cognitive behavioral treatment manual (Deblinger and Systematic reviews provide a number of advantages over examin-
Heflin, 1996) was centered around gradual exposure techniques, in- ing studies one at a time and over the results of organizational en-
cluding in vivo exposure to reminders of the trauma and writing ex- dorsements such as those named above, even when these
ercises in which traumatized children were encouraged to describe endorsements are based on the results of randomized trials. First, sys-
the details of the trauma as well as associated feelings and thoughts. tematic reviews include all of the eligible trials conducted to date in
Deblinger also focused on the therapeutic role of the parent (Cohen their analyses; some rating systems award high marks to an interven-
et al., 2006). Cohen and Mannarino and Deblinger merged their ap- tion if two randomized controlled trials have shown significantly pos-
proaches to treating traumatized youth in 1997; the manual for the itive effects, even if other studies or better studies showed null
branded TF-CBT was available on the internet for a number of years effects. Secondly, implicit in a systematic review is an assessment of
and was published in 2006 in book form. the quality of the included research studies, weighing when neces-
While the branded version of TF-CBT is the most well known and sary the results of some studies over others. Rating organizations
widely disseminated, other child-focused trauma treatments employ- may include a study in their research base that is of questionable
ing many of the same intervention components are available to clini- quality or has produced misleading results. Finally, numerous ana-
cians, have been actively disseminated (to varying degrees), and have lyses can be conducted within a systematic review, which provides
been evaluated in clinical trials. The most similar of these interven- a more complete picture than a rating or endorsement. For example:
tions is the Cognitive Behavioral Intervention for Trauma in Schools systematic reviewers can assess the effectiveness of an intervention
(CBITS). This intervention shares nearly all treatment components on a variety of measured outcomes, at a variety of times post comple-
with the branded version of TF-CBT, but rotates between a group tion, and against a variety of neutral or active conditions.
and individual format in the school setting and does not consistently Systematic reviews can also reveal inconsistencies across studies
include a caregiver component (Stein et al., 2003; Jaycox et al., 2010). and outcomes with practical implications for organizations and prac-
A number of other Cognitive Behavioral Treatments that have been titioners who are considering adopting the candidate intervention.
adapted to meet the needs of traumatized children similarly share These advantages were in play in Littell et al.'s review of Multi-
the majority of treatment components (Berger, Pat-Horenczyk, and Systemic Treatment (MST, Littell, 2005; Littell, Popa, and Forsythe,
Gelkopf, 2007; Celano, Hazzard, Webb, and McCall, 1996; Deblinger, 2005), another child intervention that has been highly endorsed.
Lippmann, and Steer, 1996; King et al., 2000; Smith et al., 2007). They concluded that effects across studies were not consistent and
that the most rigorous analyses found no significant differences
2.2. Evaluation and dissemination between MST and usual services in reducing restrictive out-of-home
placements, arrests or convictions. This finding resulted in a mild fire-
Despite the fact that there has been no published systematic re- storm, with complaints from the treatment developers (Henggeler,
view of the branded or other versions of TF-CBT for children and Schoenwald, Swenson, and Borduin, 2006) and a response from the
youth, there are a number of reasons to expect that these interven- lead systematic reviewer (Littell, 2006). Our choice to conduct a sys-
tions are effective at reducing symptoms of post-traumatic stress, de- tematic review of TF-CBT was based on our interest in looking across
pression and behavior problems following trauma. Firstly, a 2007 multiple studies and outcomes, pooling results, to evaluate the extent
Cochrane Collaboration systematic review of trauma-focused cogni- to which TF-CBT was having a positive impact on the lives of trauma-
tive behavioral interventions for adults concluded that both group tized children.
750 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757

South Carolina revealed that many of the studies often cited as evi-
dence of TF-CBT's effects did not use the branded version of TF-CBT.
Several used interventions similar to TF-CBT, but those interventions
did not use all of the eight P.R.A.C.T.I.C.E. components. In order to con-
sider what other TF-CBT interventions to include in this review, we
went to the writings of the treatment developers of the branded ver-
sion to see which treatment components they viewed as key. In 2000
(Cohen, Mannarino, Berliner, and Deblinger, 2000), they asserted that
(1) exposure, (2) cognitive processing and reframing, (3) stress man-
agement, and (4) parental treatment were the “four major compo-
nents” of TF-CBT (p. 1202). In their 2006 book, they also call
psychoeducation “one of the major components of TF-CBT” (2006, p.
59). Many of the studies of the Cohen/Mannarino and Deblinger
teams that used pre-branded versions of TF-CBT did not include all
five components and neither did many of the other TF-CBT interven-
tions of different lineage.
Ultimately, we decided it was necessary to conduct multiple ver-
sions of the analyses that pooled results across studies. In the first
version, we would include only studies that evaluated the branded
version of TF-CBT. In the second version, we would include studies
that evaluated TF-CBT interventions that used all five of the compo-
nents listed above. In the third version, we would include studies
that evaluated TF-CBT interventions that used any four of the five
key components listed above.
In addition, we included a study in our review if it:

• used a randomized trial design with a non-TF-CBT comparison

condition;
• included study participants who were under the age of 18;
• included study participants that had survived at least one traumatic
event;
Fig. 1. Search and review strategy. • assessed symptoms of posttraumatic stress disorder (PTSD); and
• was published between 1990 and 2011 (We determined that it was
unlikely we would find TF-CBT trials with all of these components
There have been a number of narrative reviews that set out to de- prior to 1990.)
termine which of the evidence-based interventions is most effective
in its treatment of PTSD symptoms in traumatized youth (Dorsey, 3.2. Search strategy
Briggs, and Woods, 2011; Dowd and McGuire, 2011; Feeny, Foa,
Treadwell, and March, 2004; Rolfsnes and Idsoe, 2011; Silverman et Fig. 1 depicts the strategies we used for retrieval. We used six da-
al., 2008; Wethington et al., 2008). All of these studies review cogni- tabases to locate potential research studies for inclusion: MEDLINE,
tive behavioral interventions and some specifically include TF-CBT in PsycINFO, CINAHL, Child Development and Adolescent Studies, SCO-
their database. Those that include TF-CBT tend to state that its effects PUS, Social Work Abstracts, ISI Web of Knowledge and ERIC. Search
are superior to other forms of treatment (Dowd and McGuire, 2011; terms are shown in Table 1. They included synonyms for the interven-
Silverman et al., 2008). However, none of these studies (1) exclusive- tion, the population and the study design. Literature was included in
ly review TF-CBT or (2) include TF-CBT's effectiveness when treating all languages, from all countries and from all types of publications.
PTSD, depression and behavior problems. Those search terms yielded 1621 articles for review.
Also reviewed were the referenced sources listed on the TF-CBT
3. Methods website that were considered to be either “highly related” or “similar”
to TF-CBT. Thirteen of the sixteen listed studies were returned in our
3.1. Study inclusion database search. We included the three that were not in our list of po-
tential studies. We also reached out to the Cohen, Mannarino and
Our first and most vexing task was to determine which versions of Deblinger team. They verified that all of their study results had been
TF-CBT and which TF-CBT studies should be included in the systemat- published and that they did know of any other study results from
ic review. A cursory examination of the studies cited by the treatment other teams on the effects of TF-CBT that had not yet been published
developers and on the TF-CBT website at the Medical University of (Mannarino, 2011).

Table 1
Search terms.

Search category Search terms

1: CBT “Trauma focused cognitive behavioral therapy” or “trauma focused cognitive behavioral treatment” or “trauma focused
cognitive behavioral tx” or “trauma focused cognitive behavioral” or “trauma-focused cognitive behavioral” or “TF-CBT”
or “TF-CBT” or “cognitive behavioral” or “cognitive behavioral” or “cognitive behavioral therapy” or “cognitive behavioral
treatment” or “cognitive therapy” or “cognitive treatment” or “CBT”
2: For use with youth Child* or youth
3: In a study Treatment or interven* or eval* or outcome or trial or effect or experiment*
4: For use with youth who have been traumatized Trauma or abuse or maltreat* or violen* or disaster or accident or crash or death or grief
 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757 751

The abstracts of the 1624 studies were read by both authors to see

2) Lacked Stress Management component for youth (present only for mothers)
whether the abstracts warranted reading the full research article to
determine eligibility criteria. If either reviewer thought it was war-
ranted, the article was read in full. The abstract review yielded 58 ar-

1) Lacked Exposure component for youth (present only for mothers)

ticles for further consideration. The 58 articles were read to
determine whether they should be discussed as potentially meeting

1) Too little information provided on treatment components.

1) Not all study participants had survived a traumatic event
criteria. Of the 58, we established that 23 had the potential to meet
full criteria. Twelve articles, representing ten unique studies met cri-
teria for inclusion in the systematic review. Nine studies and eleven
articles were excluded due to (1) their lack of a measure of PTSD,
(2) their lack of two or more core components, (3) their lack of a

3) Lacking stress management component

non-CBT comparison group or (4) their lack of all participants
experiencing at least one traumatic event. Because inclusion/exclu-

1) No non-CBT comparison group

1) No non-CBT comparison group

1) No non-CBT comparison group

sion of these studies is crucial to understanding the results of the sys-

2) Lacking exposure component

tematic review, we provide in Table 2 the reasons why studies that
reached the final round of consideration were excluded from the sys-

1) No measure of PTSD

1) No measure of PTSD
1) No measure of PTSD
tematic review.

Reason for exclusion

It is important to note that some of Cohen and Mannarino's early
trials (Cohen and Mannarino, 1996a, 1996b, 1997, 1998, 2000) were
excluded from our analyses. This was not due to the fact that the in-
terventions were not highly similar to the branded version of TF-
CBT; some of them were (Cohen and Mannarino, 1998, 2000). The ex-
clusion of these studies was based on the fact that they did not mea-

A school-based, teacher-mediated prevention program (ERASE-Stress) for reducing terror-related traumatic reactions in
sure symptoms of PTSD.

A Pilot Randomized Controlled Trial of Combined Trauma-Focused CBT and Sertraline for Childhood PTSD Symptoms

Comparative Efficacies of Supportive and Cognitive Behavioral Group Therapies for Young Children Who Have Been
Trauma‐focused cognitive behavioral therapy for children: impact of the trauma narrative and treatment length
3.3. Outcomes of interest

Children's mental health care following Hurricane Katrina: A field trial of trauma-focused psychotherapies
While PTSD symptoms are our primary outcome of interest, we

A Treatment Study for Sexually Abused Preschool Children: Outcome During a One-Year Follow-up
also decided to review TF-CBT studies in relation to two other out-
comes: depression symptoms and behavior problems. This decision
was made in light of the fact that Cohen, Mannarino and Deblinger
clearly state in their treatment manual that TF-CBT was intended to Cohen and Mannarino, 1996a, A Treatment Outcome Study for Sexually Abused Preschool Children: Initial Findings
target “affective trauma symptoms” and “behavioral trauma symp- Factors that Mediate Treatment Outcome of Sexually Abused Preschool Children

Interventions for Sexually Abused Children: Initial Treatment Outcome Findings

toms” (p. 6–14, Cohen et al., 2006). They include depression as a com-
mon affective trauma symptom, and explain that traumatized
children may develop behaviors that enable them to avoid painful
feelings. Further, they make special mention of the effectiveness of

A Comparison of CBT and EMDR for Sexually-abused Iranian Girls

the parenting skills section of TF-CBT for children who experience be-
Predictors of Treatment Outcome in Sexually Abused Children

havioral problems in response to trauma (p. 67, Cohen et al., 2006).

Lastly, Cohen, Mannarino and Deblinger's trials evaluating the brand-
ed version of TF-CBT include measures of PTSD, depression and be-
havior problems, and show effects for all (Cohen, Deblinger,
Sexually Abused and Their Nonoffending Mothers

Israeli youth: a quasi-randomized controlled trial

Mannarino, and Steer, 2004; Cohen, Mannarino, and Iyengar, 2011;

Cohen, Mannarino, and Knudsen, 2005; Deblinger, Mannarino,
Cohen, and Steer, 2006).

3.4. Evaluation of studies

Systematic reviews assess the quality of each included study in

order to determine whether certain studies should be given more or
less weight due to their strengths and weaknesses. This review can
Article name

be facilitated through the use of scaled instruments that measure dif-

ferent forms of validity and sum to total quality scores, or by engaging
in qualitative assessment. A number of authors have now recom-
Studies excluded in the final round.

mended against the use of quality rating instruments due to the con-
flation of measures that are not differentiated in summation (Higgins
Cohen and Mannarino, 1998
Cohen and Mannarino, 2000
Authors, year of publication

Cohen, Mannarino, Perel, &

Gelkopf and Berger, 2008

and Green, 2006; Littell, Corcoran, and Pillai, 2008). High scores on
Jaberghaderi et al., 2004

some sections of the scale may contribute to a high overall score de-
Deblinger et al., 2011
Deblinger et al., 2001

Jaycox et al., 2010

spite troubling scores on other sections, resulting in misleading re-

Staron, 2007

sults. Given these concerns we decided to examine the quality of

1996b, 1997

the research studies in our review qualitatively, assessing (1) study

design; (2) selection bias; (3) unaccounted for confounders; (4)
Table 2

data collection; (5) handling of missing data; (6) intervention integ-

rity and (7) analyses.
752 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757

3.5. Synthesis of data approximation. We also used the pretreatment SD scores as a proxy
for the Stein study, which similarly did not include posttreatment
Data was entered manually into meta-analytic software (Compre- SD scores. Lastly, the 2011 Cohen, Mannarino and Iyengar study was
hensive Data Analysis) to conduct three separate meta-analyses. (1) the only one to use a total CBCL score. In the third condition of the
The first set of analyses had the strictest of inclusion criteria: it in- second and third sets of meta analyses, their total CBCL score is com-
cluded only those trials testing the branded version of TF-CBT. (2) pared with the Externalization scores in the comparison studies.
The second set of analyses had slightly looser criteria: it included tri-
als testing the branded TF-CBT and interventions that had 5 out of 5 4. Results
key treatment components in common with branded TF-CBT. (3)
The third set of analyses had the loosest of inclusion criteria; it includ- 4.1. Studies selected for inclusion
ed trials testing branded TF-CBT and interventions that had 4 out of 5
key treatment components in common with branded TF-CBT. All of Table 3 provides descriptions of the ten studies included in this re-
these analyses (a) compared TF-CBT interventions and an attention view, and the analyses in which each study was included. Of the ten
control condition, waitlist control condition or standard community studies, three evaluated the branded version of TF-CBT. The other
care condition immediately after the termination of treatment and seven interventions, (1) Overshadowing the Threat of Terrorism
(b) compared TF-CBT interventions and an attention control condi- (OTT) (Berger et al., 2007); (2) Recovering From Abuse Program
tion or standard community care condition at 12 months after termi- (RAPP) (Celano et al., 1996), (3) Mother and Child Cognitive Behav-
nation of treatment. The second and third analyses compared TF-CBT ioral Therapy (Deblinger et al., 1996), (4) Family Cognitive Behavioral
interventions with another active treatment condition immediately Therapy (King et al., 2000), (5) Trauma-Focused Cognitive Behavioral
after the termination of treatment (none of the studies included in Therapy (TF-CBT, based on CBT-SAP) (Scheeringa et al., 2011), (6)
the first analysis included alternative active treatment conditions). Cognitive Behavioral Therapy (CBT) (Smith et al., 2007) and (7) Cog-
We considered it important to separately examine studies with an ac- nitive Behavioral Intervention for Trauma in Schools (CBITS) (Stein et
tive treatment comparison condition, because one would expect al., 2003), were highly similar to the branded version of TF-CBT
smaller effect sizes for these studies. Grouping them with the non- (shared at least four of five key treatment components). Two of
active comparison studies would likely dampen the effect size in the these studies had multiple comparison conditions (Deblinger et al.,
pooled analyses, leading to an underestimation of treatment effect. 1996; King et al., 2000). These two studies were included in both
Hedges' g scores and confidence intervals were calculated to ex- the comparison of TF-CBT with an attention control, standard com-
amine the standardized mean differences in the post outcomes of munity care or wait list control and of TF-CBT with alternative active
PTSD symptoms, depression symptoms and problem behaviors for treatment conditions. The other eight studies compared the active TF-
all three groups in the three analyses. Hedges' g is a measure of effect CBT condition with one attention control condition (Berger et al.,
size that includes an adjustment for small sample size (Littell et al., 2007; Celano et al., 1996; Cohen et al., 2011; Cohen, Mannarino, and
2008) and was chosen over Cohen's d in light of the small sample Knudsen, 2005; Cohen et al., 2004; Scheeringa et al., 2011; Smith et
sizes used in the studies we reviewed. al., 2007 and Stein et al., 2003). Three of the studies captured data
The TF-CBT interventions in the review differed in their imple- at twelve months after completion of the intervention (Cohen,
mentation, such as number of sessions, length of sessions, type of cli- Mannarino, and Knudsen, 2005; Deblinger et al., 2006; Deblinger et
nician and population, including age and differences in traumatic al., 1999); the other studies captured data at the immediate end of
event experienced. This heterogeneity led us to base our estimated treatment.
pooled effect sizes on a random effects model. Random effect models Seven of these studies appear to be efficacy trials (Celano et al.,
are more appropriate than fixed models when there are reasons to 1996; Cohen, Mannarino, and Knudsen, 2005; Cohen et al., 2004;
believe that the differences in effect sizes between studies are due Deblinger et al., 1996; King et al., 2000; Scheeringa et al., 2011;
to subject level sampling error and between-study variance (Lipsey Smith et al., 2007), as therapy was conducted by university-trained
and Wilson, 2000; Littell et al., 2008). By way of this model, the effect clinicians and exclusion criteria were strictly defined. Three of these
size is calculated by using an inverse variance weight that takes both trials could be considered effectiveness trials (Berger et al., 2007;
subject level sampling error and other random variance into Cohen et al., 2011; Stein et al., 2003), as therapy was conducted by
consideration. community clinicians, school counselors or teachers under “real-
Not all of the studies were consistent in the manner in which they world” circumstances. Four of the studies implemented a 12-week
reported their findings. When the authors were unable to provide TF-CBT treatment model with an 80–90 minute session per week
exact scores for the purposes of this review, we first contacted (Cohen, Mannarino, and Knudsen, 2005; Cohen et al., 2004;
study authors to ask for additional information and if it was not forth- Deblinger et al., 1996; Scheeringa et al., 2011). Three of the studies
coming, we adopted methods to compensate for these inconsistencies implemented an 8-week treatment model, one with 90-minute ses-
or left out what we could not closely approximate. First, Cohen et al. sions (Berger et al., 2007) and two with 60-minute sessions (Celano
(2004, 2006) did not provide a total PTSD mean or standard deviation et al., 1996; Cohen et al., 2011). Smith et al. (2007) and Stein et al.
(SD) at post, instead providing mean and SDs of three PTSD symptom (2003) implemented a 10-week treatment model. King et al. (2000)
clusters. To come up with comparable indicators at immediate and implemented a 20-week treatment model with 50-minute sessions.
12 months post, we established a grand mean by multiplying each Sample sizes for the study conditions ranged from 12 to 92.
cluster mean by its n, adding those totals, and dividing them by the
total n of all three groups (Lipsey and Wilson, 2000). To get the 4.2. Study quality
pooled standard deviation, we multiplied the variance of each cluster
by the n minus 1, added those totals, divided by the total n of the 3 Study quality was consistently high amongst all ten studies. All
groups after subtracting 3 and took the square root of that number used controlled, randomized designs, widely-used measurement
(Lipsey and Wilson, 2000). scales and appropriate analytic methods. Further, each employed
Second, Cohen et al. (2011) reported their findings by using some methods to ensure intervention integrity and most employed
change scores as opposed to reporting means. To come up with com- extensive methods. Studies 1, 2 and 3 trained and/or used some of
parable scores, we deducted the negative change score from the pre- their therapists to complete both interventions to avoid bias that
treatment score mean and used the pretreatment SD scores as a proxy may result from a particularly effective therapist. Studies 3, 4, 5, 7,
for the posttreatment SD scores to obtain a reasonable 8, 9 and 10 monitored the content of sessions by recording a portion
Table 3
Studies included in systematic review (n = 10).

Study #, authors, year of Sample sizes for Sample Other eligibility criteria TF-CBT and dosage Comparison Condition/s Included in
publication each condition age and Dosage meta analysis #:

(1) Berger et al., 2007 OTT = 70 2nd to 6th Children who (1) were part of the specific school Overshadowing the Threat of Terrorism (OTT) Wait List Control (WLC) 3
WLC = 72 grade population and (2) had a consenting parent. 80–90-minute sessions in social studies class;
2 psychoeducation sessions for parents.
(2) Celano et al., 1996 RAP, N = 15 8–13 years Girls who (1) experienced confirmed sexual abuse Recovering from Abuse Program (RAP) Treatment As Usual (TAU) 3
TAU, in the past 3 years; (2) could disclose the abuse 8 weekly sessions, 60 min; 30 min spent with 8 weekly sessions, 60 min;
N = 17 during evaluation and (3) had a consenting, child and 30 min spent with parent average of 50% spent with child,
non-perpetrating parent willing to participate. (2–3 sessions conjoint) 43% spent with parent and 6% spent
in conjunction.
(3a) Cohen et al., 2004 TF-CBT, N = 92. 8–14 years Children who (1) experienced confirmed TF-CBT (branded) Child Centered Therapy (CCT) 1, 2, 3
(3b) Deblinger et al., 2006. CCT, N = 91. contact sexual abuse; (2) had 5 PTSD 12 weekly sessions 12 weekly sessions

C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757
symptoms; (3) were English-speaking 9 sessions: 9 sessions:
90 min, 45 for each 90 min, 45 for each
3 sessions:
30 min combined
30 for each
(4) Cohen et al., 2011 TF-CBT, N = 64. 7–14 years Children who (1) had at least 5 IPV-related PTSD TF-CBT (branded) Child Centered Therapy (CCT) 1, 2, 3
CCT, N = 60. symptoms; (2) were English-speaking; (3) had an 8 weekly sessions: 8 weekly sessions
English-speaking mother who was a direct IPV victim. 90 min, 45 for each 90 min, 45 for each
2 sessions:
Part was spent with parent/child together
Added following:
(1)safety component moved to the beginning;
(2) trauma narrative focused on sharing
child's IPV experiences;
(3) goal to discriminate between real danger
and generalized fears
(5) Cohen, Mannarino, TF-CBT, N = 41. 8–15 years Children who (1) experienced confirmed sexual abuse in the past TF-CBT (branded) Non-directive supportive therapy (NST) 1, 2, 3
and Knudsen, 2005 NST, N = 41. 6 months; (2) had “significant symptomatology” related to sexual 12 weekly sessions 12 weekly sessions
abuse; (3) had a non-offending, English-speaking parent who was 90 min, 45 for each 90 min, 45 for each
available to participate in treatment.
(6a) Deblinger et al., 1996 Mother and Child 8–14 years Children who (1) had experienced confirmed sexual abuse; (2) Mother and Child Cognitive Behavioral Therapy Child Cognitive Behavioral Therapy 2, 3
(6b) Deblinger, Steer, CBT, N = 22. had at least 3 symptoms of PTSD (including 1 symptom of 12 weekly Sessions 12 weekly sessions
and Lippmann, 1999 Child CBT, N = 24. avoidance or re-experiencing). 80–90 min 45 min with child
SCC, N = 22. 1st phase: each seen for 45 min separately (parents given periodic updates)
Later phase: seen for 30 min together Standard Community Care (SCC)
(7) King et al., 2000 Family CBT, N = 12. 5–17 years Children who (1) experienced confirmed sexual abuse; (2) had at Family Cognitive Behavioral Therapy Child Cognitive Behavioral Therapy 2, 3
Child CBT, N = 12. least 3 symptoms of PTSD; (3) were English-speaking. 20 weekly sessions for youth, 50 min 20 weekly sessions for youth
WLC, N = 12. 20 weekly sessions for parent, 50 min 50 min
(parents received no form of treatment)
Wait List Control (WLC)
(8) Scheeringa, Weems, TF-CBT, N = 40 3–6 years Children who (1) experienced a life-threatening traumatic event; Trauma-focused Cognitive Behavioral Therapy, Wait list Control (WLC) 3
Cohen, Amaya-Jackson, WLC = 24 (2) were 36 to 83 months at the time of the most recent trauma Immediate Treatment (adapted from CBT-SAP)
and Guthrie, 2011 and at the time of enrollment; (3) had 4 or more PTSD symptoms 12 weekly sessions, Includes joint caregiver/child
with at least one being a reexperiencing symptom or an avoid- sessions and TV monitoring of sessions by caregivers.
ance symptom.
(9) Smith et al. (2007) CBT = 12 8–18 years Children who (1) had PTSD (relating to a single traumatic event) Cognitive Behavioral Therapy (CBT) Wait List Control (WLC) 3
WLC = 12 as the main presenting problem and (2) were fluent in English. 10 weekly individual sessions with child,
parents usually seen by therapist after
session; conjoint sessions as necessary
(10) Stein et al., 2003 CBITS = 61 6th grade Children who (1) had substantial exposure to violence (2) had Cognitive Behavioral Intervention for Wait List Control (WLC) 3
WLC = 65 symptoms of PTSD in the clinical range; (3) had symptoms of PTSD Trauma in Schools (CBITS)
related to exposure of violence (4) did not appear too disruptive to 10 sessions (most often meeting once a week)
participate in group therapy (5) received parental consent.

753
754 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757

of sessions usually sessions chosen randomly. Study 6 required clini- those studies that evaluated interventions sharing only 4 out of 5
cians to complete a procedure checklist for every session. Studies 3, 4, key components (1) measured effects at 12 months or (2) compared
7 and 8 measured their intervention integrity rates to be >90% (Study the intervention with an active alternative intervention. Therefore,
10 provided a mean intervention integrity rate of 96%). pooled effect sizes are consistent in the 12 months and active com-
Despite the high level of study quality, there remained two concerns parison categories for meta analyses 2 and 3.
for specific studies. The first concerns rates of attrition. Study 8 had seri- We will focus the remainder of this result section on the results
ous retention problems, attributed to the fact that Hurricane Katrina produced in the third set of meta analyses which included studies
struck six months after randomization. Forty children were randomized that shared at least four out of five key components with branded
to the TF-CBT group and 13 (33%) completed the intervention. Twenty- TF-CBT. As shown in Table 5, Studies 1, 3, 4, 6, 7, 9 and 10 revealed
four were randomized to the comparison group, but only 3 (13%) provid- a significant difference between the TF-CBT conditions and compari-
ed information at post. Studies 2 and 4 had moderate retention issues. In son conditions in their ability to reduce symptoms of PTSD at imme-
study 2, 25 subjects were randomized to the TF-CBT group and 15 (60%) diate post. Studies 2, 5 and 8 were the only studies that did not show
completed the intervention. In study 4, 64 subjects were randomized to a significant difference between the two conditions. Subsequently,
the TF-CBT group and 43 (67%) completed the intervention. Researchers the overall pooled estimates suggested a medium effect size (Cohen,
in this study attributed this drop out rate to the fact they did not preclude 1988) for PTSD at immediate post. As seen in Table 6, the pooled es-
actively drug-abusing mothers from participation. Otherwise, treatment timate suggested a more modest effect size for PTSD at 12 month
completion rates at immediate post were excellent: a 100% rate in Study post-treatment.
1; a 78% rate in Study 3, a 78% rate in Study 5, an 88% rate in Study 6, a Overall pooled estimates shown in Table 5 revealed small effect
75% rate in Study 7, a 100% rate in Study 9 and a 93% rate in Study 10. sizes for depression and behavior problems at immediate post. Stud-
Where reported, 12-month assessments were conducted on 72% (3) ies 5, 6, 8 and 9 showed medium to large effect sizes for the reduction
and 76% (6) of the original sample. In light of these rates, researchers of depression symptoms, but the pooled estimate showed less of a
in Studies 3, 4, 5, 6, 7 and 9 conducted Intent to Treat (ITT) analyses to difference between TF-CBT and the comparison condition at immedi-
account for those who dropped out along the course of the intervention. ate post. Studies 2, 6 and 8 also showed medium effect sizes for the
The second quality concern relates to blind assessment. The extent to reduction of behavior problems, but the remaining studies showed
which those who conducted the post-treatment assessments were blind small effect sizes; these contribute to the small pooled effect size for
to treatment conditions is clear in Studies 1, 2, 3, 5 and 9. There is a pos- behavior problems at immediate post. Both of these small pooled ef-
sibility that Studies 4, 6, 7 and 8 used parties who were aware of partic- fect sizes, for depression and for behavior problems, decreased sub-
ipant assignment to conduct some or all of their post-treatment stantially and become insignificant at 12 months after the
assessments. Study 10 used self-administered questionnaires. completion of the intervention, as shown in Table 6.
In light of these findings, we examined each study with a twelve-
4.3. Effects of TF-CBT month follow-up to see whether the effects of TF-CBT faded or if
symptoms of depression and behavior problems in the comparison
As previously mentioned, three separate sets of meta analyses group improved over time. We found that the insignificant difference
were conducted in this review: (1) the first included studies that between the groups at 12 months was not the result of declining ef-
evaluated branded TF-CBT; (2) the second included studies that fects of TF-CBT. All 3 studies that showed effects at 12 months post
shared five out of five key treatment components with branded TF- (Studies 3, 5 and 6) revealed that depression improved modestly for
CBT and (3) the third included studies that shared four out of five the groups who did not receive TF-CBT over the course of the year
key treatment components with branded TF-CBT. As shown in after termination of treatment. The same mean trend appeared
Table 4, we discovered that findings were consistent amongst meta- when we evaluated for behavior problems; Studies 3 and 6 revealed
analyses. Pooled estimates were similar whether we were analyzing that behavior problems also improved modestly for the groups who
the effects of interventions that were highly similar to TF-CBT, or if did not receive TF-CBT over the course of the year after termination
we were exclusively analyzing the effects of the branded interven- of treatment, but Study 5 revealed that behavior problems in the at-
tion. However, it is important to note that none of the trials that eval- tention control group decline until 6 months post, and then spiked
uated the branded version of TF-CBT compared it against an between 6 and 12 months post. In short, children who received TF-
alternative active condition. It is also important to note that none of CBT were more likely to have posttrauma depression and behavior

Table 4
Pooled results of meta analyses 1 (branded), 2 (5 shared components) and 3 (4 shared components).

Analysis PTSD Depression Behavior problems

Effect size/ 95% CI: PTSD Z p Effect size/ 95% CI: Z p Effect size/ 95% CI: Z p
SE: PTSD SE: depress depress SE: behave behave

Inclusion criteria: branded version of TF-CBT

1 — post .407 (.102) .206, .607 3.975 .000 .320 (.102) .121, .520 3.152 .002 .202 (.102) .002, .401 1.983 .047
1 — 12 months .350 (.131) .094, .606 2.675 .007 .174 (.129) −.079, .428 1.347 .178 .086 (.132) −.173, .344 .650 .516
1 — active control – – – – – – – – – – – –

Inclusion criteria: 5 out of 5 components

2 — post .487 (.095) .302, .673 5.142 .000 .357 (.094) .173, .540 3.804 .000 .237 (.094) .054, .421 2.532 .011
2 — 12 months .389 (.123) .149, .630 3.174 .002 .172 (.121) −.065, .409 1.426 .154 .083 (.124) −.160, .326 .670 .503
2 — active control .100 (.234) −.359, .558 .427 .669 .035 (.234) −.423, .493 .152 .879 .333 (.236) −.129, .794 1.413 .158

Inclusion criteria: 4 out of 5 components

3 — post .671 (.074) .527, .815 9.129 .000 .378 (.081) .218, .537 4.648 .000 .247 (.081) .089, .405 3.064 .002
3 — 12 months .389 (.123) .149, .630 3.174 .002 .172 (.121) −.065, .409 1.426 .154 .083 (.124) −.160, .326 .670 .503
3 — active control .100 (.234) −.359, .558 .427 .669 .035 (.234) −.423, .493 .152 .879 .333 (.236) −.129, .794 1.413 .158

Note: PTSD = symptoms of post-traumatic stress disorder; depress = symptoms of depression; behave = behavior problems.
Analyses 2 and 3 include the same studies in the 12-month and active control results.
 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757 755

Table 5
Meta analysis 3: 4 out of 5 components of TF-CBT v. Comparison treatment condition at immediate post.

Study Effect size/ 95% CI: Z p Effect size/ 95% CI: Z p Effect size/ 95% CI: Z p Instruments used:
# SE: PTSD PTSD SE: depress depress SE: behave behave

1 1.053 (.178) .704, 1.403 5.909 .000 – – – – – – – – UCLA PTSD,

child version
2 .206 (.346) −.472, .885 .596 .551 – – – – .689 (.356) −.008, 1.387) 1.936 .053 CITES-R
CBCL
3 .526 (.151) .230, .822 3.482 .000 .402 (.149) .110, .693 2.701 .007 .287 (.150) −.006, .581 1.921 .055 K-SADS-PL
CDI
CBCL
4 .362 (.180) .009, .715 2.010 .044 .083 (.179) −.267, .434 .467 .640 .206 (.179) −.145, .557 1.153 .249 K-SADS-PL
CDI
CBCL
5 .222 (.219) −.208, .652 1.012 .311 .506 (.222) .070, .942 2.275 .023 .011 (.219) −.418, .440 .050 .960 TSC-C
CDI
CBCL
6 .86 (.310) .253, 1.468 2.775 .006 .729 (.306) .129,1.329 2.381 .017 .531 (.305) −.067,1.128 1.741 .082 K-SADS-E
CDI
CBCL
7 1.194 (.430) .351, 2.038 2.776 .006 .281 (.396) −.495, 1.058 .710 .478 .283 (.396) −.493, 1.060 .715 .474 ADIS, child version
CDI
CBCL
8 1.118 (.637) −.131, 2.366 1.755 .079 .726 (.619) −.487, 1.940 1.173 .241 .684 (.618) −.526, 1.895 1.108 .268 PAPA
9 1.53 (.452) .645, 2.416 3.387 .001 .707 (.407) −.091, 1.505 1.736 .083 – – – – CAPS
DSRS
10 .934 (.194) .038, .554 4.813 .000 .361 (.186) −.003, .725 1.943 .052 .127 (.184) −.234, .489 .691 .490 CPSS
CDI
TCRS
Pooled .671 (.074) .527, .815 9.129 .000 .378 (.081) .218, .537 4.648 .000 .247 (.081) .089, .405 3.064 .002

Note: PTSD = symptoms of post-traumatic stress disorder; depress = symptoms of depression; behave = behavior problems; UCLA PTSD, child version = UCLA index for DSM IV,
child version; CITES-R = children's impact of traumatic events scales — revised; K-SADS-PL = kiddie schedule for affective disorders and schizophrenia for school aged children —
present and lifetime version; K-SADS-E = kiddie schedule for affective disorders and schizophrenia for school aged children — epidemiological version; CDI = children's depression
inventory; CBCL = child behavior checklist; TSC-C = trauma symptom checklist for children, ADIS = anxiety disorders interview schedule; PAPA = preschool age psychiatric as-
sessment; CAPS = clinician administered PTSD scale; DSRS = depression self-rating scale; CPSS = the child PTSD symptom scale; TCRS = teacher-child rating scale.

problems resolve more quickly than children in comparison condi- Table 7 also revealed that there was no significant difference be-
tions, although in many cases, children in the comparison condition tween the TF-CBT condition and the alternative active conditions in
caught up to children in the TF-CBT by 12 months after intervention. their ability to reduce symptoms of PTSD, depression or behavior

Table 6
Meta analysis 3: 4 out of 5 components of TF-CBT v. Comparison condition at 12 months.

Study # Effect size/ 95% CI: Z p Effect size/ 95% CI: Z p Effect size/ 95% CI: Z p Instruments
SE: PTSD PTSD SE: depress depress SE: behave behave used:

3 .285 (.162) −.033, .602 1.759 .079 .174 (.160) −.141, .488 1.082 .279 .116 (.165) −.208, .440 .703 .482 K-SADS-PL
CDI
CBCL
5 .472 (.222) .037, .907 2.128 .033 .176 (.219) −.254, .605 .802 .423 .032 (.219) −.396, .461 .148 .882 K-SADS-PL
CDI
CBCL
6 .678 (.354) −.015, 1.371 1.918 .055 .159 (.338) −.503, .821 .471 .638 .062 (.363) −.648, .773 .172 .863 K-SADS-E
CDI
CBCL
Pooled .389 (.123) .149, .630 3.174 .002 .172 (.121) −.065, .409 1.426 .154 .083 (.124) −.160, .326 .670 .503

Note: PTSD = symptoms of post-traumatic stress disorder; depress = symptoms of depression; behave = behavior problems; K-SADS-PL = schedule for affective disorders and
schizophrenia for school aged children — present and lifetime version; K-SADS-E = kiddie schedule for affective disorders and schizophrenia for school aged children — epidemi-
ological version; CDI = children's depression inventory; CBCL = child behavior checklist.

Table 7
Meta analysis 3: 4 out of 5 components of TF-CBT v. Alternative active treatment condition at immediate post.

Study # Effect size/ 95% CI: PTSD Z p Effect size/ 95% CI: depress Z p Effect size/ 95% CI: behave Z p Instruments used:
SE: PTSD SE: depress SE: behave

6 .037 (.290) −.532, .605 .126 .900 .046 (.290) −.522, .615 .160 .873 .425 (.293) −.150, 1.00 1.448 .147 K-SADS-E
CDI
CBCL
7 .218 (.395) −.557, .993 .550 .582 .015 (.394) −.757, .788 .039 .969 .166 (.395) −.608, .940 .420 .675 ADIS, child version
CDI
CBCL
Pooled .100 (.234) −.359, .558 .427 .669 .035 (.234) −.423, .493 .152 .879 .333 (.236) −.129, .794 1.413 .158

Note: PTSD = symptoms of post-traumatic stress disorder; depress = symptoms of depression; behave = behavior problems; K-SADS-PL = kiddie schedule for affective disorders and
schizophrenia for school aged children — present and lifetime version; K-SADS-E = kiddie schedule for affective disorders and schizophrenia for school aged children — epidemiological
version; CDI = children's depression inventory; CBCL = child behavior checklist; SDQ = strengths and difficulties questionnaire; ADIS = anxiety disorders interview schedule.
756 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757

problems at post. However, these findings must be interpreted with on all of the studies that others might consider similar in nature. We
caution, as there were only two studies that included an active alter- did not want to base this review only on the three studies that
native treatment. employed the branded version of TF-CBT because there are other in-
terventions that are highly similar. However, we also thought that
5. Discussion basing a systematic review exclusively on studies that share less
than four critical components with the branded version of TF-CBT
This is the first systematic review of TF-CBT for children and youth would not be helpful in determining the effects of this intervention.
to provide pooled estimates of the intervention's ability to reduce It's for these reasons that we decided to conduct three separate sets
symptoms of PTSD, depression and behavior problems. The pooled of meta-analyses. We have discovered that while the branded version
estimates strongly suggest that TF-CBT is more effective than atten- performs extremely well, and has the advantage of being clearly out-
tion control, standard community care and waitlist control conditions lined in a manual and supported by trainings and learning collabora-
at reducing symptoms of PTSD in youth, both immediately and tives, clinicians may consider using highly similar alternatives with
12 months after the termination of treatment. Therefore, enthusiasm the presumption that these alternatives are likely to deliver similar
for this intervention appears to be justified: TF-CBT effectively helps results.
traumatized youth who experience symptoms of PTSD. This is an im- These results should be understood with respect to the limitations
portant finding, as it confirms that clinicians across the world are in a of this study. First, while ten trials of an intervention are considered a
position to help traumatized children by being trained in and utilizing large number in the children's services world, by meta-analytic stan-
TF-CBT interventions. dards it is still a relatively small number. Further, the sample sizes in
Further, the use of TF-CBT interventions may also help traumatized some of those studies were also small. Secondly, none of the studies
youth with their symptoms of depression and their problem behaviors. included in this review was designed to examine mediating and mod-
The pooled estimates strongly suggest that TF-CBT is more effective erating effects of the intervention components, as the sizes of the
than attention control, standard community care and waitlist control samples in these studies would make the chances of detecting such
conditions at reducing the symptoms of depression and problem behav- effects extremely small. A future study that includes an analysis of
iors at immediate post, although children who received an attention con- the isolated effects of each component would be highly beneficial to
trol, standard community care or waitlist condition often caught up to the advancement of this field of inquiry. Deblinger et al. (2011)
gains made in the TF-CBT condition by 12 months post. have begun this important work by conducting a study that examines
These findings may be the result of a variety of factors, and may the unique effects of the trauma narrative, comparing those in a TF-
reflect a fundamental qualitative difference in the nature of the expe- CBT group who received that component of the intervention with
rience of PTSD and the experience of depression and/or behavior those who did not.
problems. It's possible that a child's experience of depression and Deblinger's recent article (2011) is also relevant in its attempt to
their exhibition of behavior problems following trauma may naturally understand the most effective length of treatment. The studies
lessen over time, even when they are not receiving any therapeutic employing TF-CBT in this review vary widely in regards to treatment
support. It's also possible that the onset time of trauma-related symp- length (with the shortest intervention lasting eight weeks and the
toms may differ; natural symptom reductions in the areas of depres- longest lasting twenty weeks) and session length (with sessions vary-
sion and behavior problems may occur later than symptom reduction ing between 45/50 min and 90 min). Clearly, length of treatment and
in the area of PTSD after a child's experience of trauma. session can affect the outcomes of the intervention for a variety of
While the pooled estimates do not show TF-CBT to be significantly reasons. The fact that we were not able to account for treatment
more effective than alternative active interventions, it should be length or session length is a limitation of this study. Deblinger et al.
noted that each of the active treatments to which TF-CBT was com- (2011) discovered that the inclusion of the trauma narrative in an
pared contain some of the core elements of TF-CBT, including cogni- 8-week intervention model enabled parents to more effectively
tive reframing, psychoeducation, and exposure. The federal cope with their abuse-specific distress and enabled the children to
government has called for more comparison effectiveness trials more effectively cope with their fear and anxiety. However, they
(Committee on Comparative Effectiveness Research Prioritization, also discovered that those children who received a 16-week interven-
Institute of Medicine, 2009; U.S. Department of Health and Human tion model that excluded the trauma narrative were less likely to en-
Services, 2009). We believe that the field of child trauma would ben- gage in problem behaviors. These findings bring us closer to
efit from more studies that compare the effectiveness of TF-CBT with understanding how the intervention components and length contrib-
other interventions that may be chosen by clinicians over TF-CBT in ute to the overall effects of TF-CBT.
the real world. One candidate for a comparison effectiveness trial Systematic reviews can advance our understanding of how well in-
with TF-CBT is Eye Movement Desensitization and Reprocessing terventions work, who they work best for, and which symptoms they
(EMDR, Shapiro, 1996). EMDR has been shown to have positive ef- are most effectively able to treat. This systematic review was able to
fects on children experiencing symptoms of PTSD (Adler-Tapia and identify that TF-CBT is effective for the treatment of PTSD in children,
Settle, 2009a, 2009b; Chemtob, Nakashima, Hamada, and Carlson, and may hasten recovery for posttrauma depression and behavioral
2002) and performed equally as well as cognitive approaches to trau- problems. Future systematic reviews on other frequently used interven-
ma treatment in the review of adult trauma treatments (Bisson & tions would be beneficial to clinicians, administrators and parents who
Andrew, 2009). For younger children, there would be benefit in com- are looking to select the most appropriate and effective intervention
paring TF-CBT with manualized forms of play therapy for trauma sur- for the treatment of symptoms in their clients and children.
vivors (Bonner, Walker, and Berliner, 1999; Gil, 1998), as these are
interventions that community clinicians might otherwise use for chil- References
dren with symptoms of post-traumatic stress. To be clear, we are not
suggesting that TF-CBT needs to continue to show its worth in more Adler-Tapia, R., & Settle, C. (2009). Evidence of the efficacy of EMDR with children and
adolescents in individual psychotherapy: A review of the research published in
trials. We are instead suggesting that clinicians tempted to use non-
peer-reviewed journals. Journal of EMDR Practice and Research, 3(4), 232–247.
TF-CBT interventions to help traumatized children deserve to know Adler-Tapia, R. L., & Settle, C. S. (2009). EMDR psychotherapy with children. In A. Rubin,
whether these other interventions stack up well with what has be- & Springer (Eds.), Treatment of traumatized adults and children: Part of the clini-
come the intervention of choice. cian's guide to evidence based practice series. New York: Wiley.
Berger, R., Pat-Horenczyk, R., & Gelkopf, M. (2007). School-based intervention for pre-
It is important to note that the results of this review are based on vention and treatment of elementary-students' terror-related distress in Israel: A
analyses of the ten studies we chose to include in our review, and not quasi-randomized controlled trial. Journal of Traumatic Stress, 20(4), 541–551.
 C.E. Cary, J.C. McMillen / Children and Youth Services Review 34 (2012) 748–757 757

Bisson J., Andrew M. (2009) Psychological treatment of post-traumatic stress disorder Feeny, N. C., Foa, E. B., Treadwell, K. R. H., & March, J. (2004). Posttraumatic stress dis-
(PTSD). Cochrane Database of Systematic Reviews 2007, Issue 3. Art. No.: order in youth: A critical review of the cognitive and behavioral treatment out-
CD003388. doi:10.1002/14651858.CD003388.pub3. come literature. Professional Psychology: Research and Practice, 35(5), 466–476.
Bonner, B., Walker, C. E., & Berliner, L. (1999). Treatment manual for dynamic group play Gelkopf, M., & Berger, R. (2008). A school-based, teacher-mediated prevention pro-
therapy for children with sexual behavior problems and their parents/caregivers. gram (ERASE-Stress) for reducing terror-related traumatic reactions in Israeli
Washington, DC: National Clearinghouse on Child Abuse and Neglect. youth: A quasi-randomized controlled trial. Journal of Child Psychology and Psychi-
California Evidence-Based Clearinghouse for Child Welfare (2011). Rating of trauma- atry, 50(8), 962–971.
focused cognitive behavioral therapy. Retrieved June 2011 from. http://www. Gil, E. (1998). Essentials of play therapy with abused children. New York: Guilford Press.
cebc4cw.org/program/trauma-focused-cognitive-behavioral-therapy/detailed Henggeler, S. W., Schoenwald, S. K., Swenson, C. C., & Borduin, C. M. (2006). Methodo-
Celano, M., Hazzard, A., Webb, C., & McCall, C. (1996). Treatment of traumagenic beliefs logical critique and meta-analysis as Trojan horse. Children and Youth Services Re-
among sexually abused girls and their mothers: An evaluation study. Journal of Ab- view, 20, 447–457.
normal Child Psychology, 24(1), 1–17. Higgins, J. P. T., & Green, S. (Eds.). (2006). Cochrane handbook for systematic reviews of
Chemtob, C. M., Nakashima, J., Hamada, R. S., & Carlson, J. G. (2002). Brief-treatment for interventions. Chichester, UK: John Wiley & Sons.
elementary school children with disaster-related posttraumatic stress disorder: A Kauffman Best Practices Project (2004). Closing the quality chasm in child abuse treat-
field study. Journal of Clinical Psychology, 58, 99–112. ment: Identifying and disseminating best practices: Findings of the Kauffman best
Cohen, J. A., Mannarino, A. P., & Iyengar, S. (2011). Community treatment for posttrau- practices project to help children heal from child abuse. Charleston (SC)7 National
matic stress disorder for children exposed to intimate partner violence. Archives of Crime Victims Research and Treatment Center.
Pediatrics & Adolescent Medicine, 165(1), 16–21. King, N., Tonge, B. J., Mullen, P., Myerson, N., Heyne, D., Rollings, S., et al. (2000). Treat-
Cohen, J. A., & Mannarino, A. P. (2008). Disseminating and implementing trauma fo- ing sexually abused children with post-traumatic stress symptoms: A randomized
cused CBT in community settings. Trauma, Violence & Abuse, 9(4), 214–226. clinical trial. Journal of the American Academy of Child and Adolescent Psychiatry,
Cohen, J. A., Mannarino, A. P., & Deblinger, E. (2006). Treating trauma and traumatic grief 59(11), 1347–1355.
in children and adolescents. New York: Guilford. Jaberghaderi, N., Greenwald, R., Rubin, A., Zand, S. O., & Dolatabadi, S. (2004). A com-
Cohen, J. A., Mannarino, A. P., & Knudsen, K. (2005). Treating sexually abused children: One parison of CBT and EMDR for sexually abused Iranian girls. Clinical Psychology
year follow-up of a randomized controlled trial. Child Abuse & Neglect, 29, 135–145. and Psychotherapy, 11, 358–368.
Cohen, J. A., Mannarino, A. P., Perel, J. M., & Staron, V. (2007). A pilot randomized con- Jaycox, L. H., Cohen, J. A., Mannarino, A. P., Walker, D. W., Langley, A. K., Gegenheimer,
trolled trial of combined traumafocused CBT and setraline for childhood PSTD K. L., et al. (2010). Children's mental health care following Hurricane Katrina: A
symptoms. Journal of the American Academy of Child & Adolescent Psychiatry, 46, field trial of trauma-focused psychotherapies. Journal of Traumatic Stress, 23(2),
811–819. 223–231.
Cohen, J. A., Mannarino, A. P., & Staron, V. (2005). A pilot study of modified cognitive- Lipsey, M., & Wilson, D. (2000). Practical meta-analysis. Thousand Oaks, California: Sage
behavioral therapy for childhood traumatic grief (CBT-CTG). Journal of the Ameri- Publications.
can Academy of Child and Adolescent Psychiatry, 45(12), 1465–1473. Littell, J. H., Corcoran, J., & Pillai, V. (2008). Systematic reviews and meta-analysis. Ox-
Cohen, J. A., Deblinger, E., Mannarino, A. P., & Steer, R. A. (2004). A multi-site, random- ford, England: Oxford University Press.
ized controlled trial for sexually abused children with PTSD symptoms. Journal of Littell, J. H. (2006). The case for multisystemic therapy: Evidence or orthodoxy? Letter
the American Academy of Child and Adolescent Psychiatry, 43, 393–402. to the editor. Children and Youth Services Review, 28, 458–472.
Cohen, J., Mannarino, A., Berliner, L., & Deblinger, E. (2000). TF-CBT: An empirical up- Littell, J. H., Popa, M., & Forsythe, B. (2005). Multisystemic therapy for social, emotional,
date. Journal of Interpersonal Violence, 15(11), 1202–1223. and behavioral problems in youth aged 10–17. : Campbell Collaboration Library
Cohen, J. A., & Mannarino, A. P. (2000). Predictors of treatment outcome in sexually http://www.campbellcollaboration.org/frontend.asp
abused children. Child Abuse & Neglect, 24(7), 983–994. Littell, J. (2005). Lessons from a systematic review of effects of multisystemic therapy.
Cohen, J. A., & Mannarino, A. P. (1998). Interventions for sexually abused children: Ini- Children and Youth Services Review, 27, 445–463.
tial treatment findings. Child Maltreatment, 3(1), 17–26. Mannarino, A. (personal, communication, May 3, 2011).
Cohen, J. A., & Mannarino, A. (1997). A treatment study for sexually abused preschool Rolfsnes, E. S., & Idsoe, T. (2011). School-based intervention programs for PTSD symp-
children: Outcome during a one-year follow-up. Journal of the American Academy of toms: A review and meta-analysis. Journal of Traumatic Stress, 24(2), 155–165.
Child and Adolescent Psychiatry, 36, 1228–1235. SAMHSA (2008). Review of trauma-focused cognitive behavioral therapy. Retrieved
Cohen, J. A., & Mannarino, A. (1996a). A treatment outcome study for sexually abused May 2011 from. http://nrepp.samhsa.gov/ViewIntervention.aspx?id=135
preschooler children: Initial findings. Journal of the American Academy of Child and Saunders, B. (personal, communication May 16, 2011).
Adolescent Psychiatry, 35(1), 42–50. Saunders, B., Smith, D., & Best, C. (2010). Effectiveness of web-based training in dis-
Cohen, J. A., & Mannarino, A. (1996b). Factors that mediate treatment outcome of sex- seminating evidence-based trauma interventions. Presentation at the 26th annual
ually abused preschool children. Journal of the American Academy of Child and meeting of the international society of traumatic stress studies, Montreal, Canada.
Adolescent Psychiatry, 35(10), 1402–1410. Saunders, B. E., Berliner, L., & Hanson, R. F. (2002). Child physical and sexual abuse:
Cohen, J. A., & Mannarino, A. P. (1993). A treatment model for sexually abused pre- Guidelines for treatment. Charleston, SC: Office for Victims of Crime.
schoolers. Journal of Interpersonal Violence, 8(1), 115–131. Shapiro, F. (1996). Eye movement desensitization and reprocessing (EMDR): Evalua-
Cohen, J. (1988). Statistical power analysis for the behavioral sciences (second ed.). : tion of controlled PTSD research. Journal of Behavior Therapy and Experimental
Lawrence Erlbaum Associates. Psychiatry, 27, 209–218.
Committee on Comparative Effectiveness Research Prioritization, Institute of Medicine Scheeringa, M. S., Weems, C. F., Cohen, J. A., Amaya-Jackson, L., & Guthrie, D. (2011).
(2009). Initial national priorities for comparative effectiveness research. Washington, Trauma-focused cognitive behavioral therapy for posttraumatic stress disorder in
D.C: The National Academies Press. three-through six year-old children: A randomized clinical trial. Journal of Child
Deblinger, E., Mannarino, A. P., Cohen, J. A., Runyon, M. K., & Steer, R. A. (2011). Trau- Psychology and Psychiatry, 52(8), 853–860.
ma-focused cognitive behavioral therapy for children: Impact of trauma narrative Silverman, W. K., Ortiz, C. D., Viswesvaran, C., Burns, B. J., Kolko, D. J., Putnam, F. W.,
and treatment length. Depression and Anxiety, 28, 67–75. et al. (2008). Evidence-based psychosocial treatments for children and adolescents
Deblinger, E., Mannarino, A. P., Cohen, J. A., & Steer, R. A. (2006). A follow-up study of a exposed to traumatic events. Journal of Clinical Child and Adolescent Psychology,
multi-site, randomized, controlled trial for children with sexual abuse-related 37(1), 156–183.
PTSD symptoms. Journal of the American Academy of Child and Adolescent Psychiatry, Smith, P., Yule, W., Perrin, S., Tranah, T., Dalgleish, T., & Clark, D. (2007). Cognitive-be-
45(12), 1474–1484. havioral therapy for PTSD in children and adolescents: A preliminary randomized
Deblinger, E., Stauffer, L. B., & Steer, R. (2001). Comparative efficacies of supportive and controlled trial. Journal of the American Academy of Child and Adolescent Psychiatry,
cognitive behavioral group therapies for young children who have been sexually 46(8), 1051–1061.
abused and their non-offending mothers. Child Maltreatment, 6, 332–343. Stauffer, L., & Deblinger, E. (1996). Cognitive behavioral groups for non-offending
Deblinger, E., Steer, R. A., & Lippmann, J. (1999). Two-year follow-up study of cognitive mothers and their young sexually abused children: A preliminary treatment out-
behavioral therapy for sexually abused children suffering post-traumatic stress come. Child Maltreatment, 1, 65–76.
symptoms. Child Abuse & Neglect, 23, 1371–1378. Stein, B. D., Jaycox, L. H., Kataoka, S. H., Wong, M., Tu, W., Elliott, M. N., et al. (2003). A
Deblinger, E., Lippmann, J., & Steer, R. (1996). Sexually abused children suffering post- mental health intervention for schoolchildren exposed to violence: A randomized
traumatic stress symptoms: Initial treatment outcome findings. Child Maltreatment, controlled trial. JAMA, 290(5), 603–611.
1, 310–321. U.S. Department of Health and Human Services (2009). Recovery act allocates
Deblinger, E., & Heflin, A. H. (1996). Treating sexually abused children and their nonof- $1.1 Billion for comparative effectiveness research. Downloaded from. www.hhs.
fending parents: A cognitive behavioral approach. Thousand Oaks, CA: Sage. gov/recovery/programs/os/cerbios.html on December 12, 2011.
Deblinger, E., McLeer, S. V., & Henry, D. (1990). Cognitive-behavioral treatment for sex- Wethington, H. R., Hahn, R. A., Fuqua-Whitley, D. S., Sipe, T. A., Crosby, A. E., Johnson, R.
ually abused children suffering post-traumatic stress: Preliminary findings. Journal L., et al. (2008). The effectiveness of interventions to reduce psychological harm
of the American Academy of Child and Adolescent Psychiatry, 29(747–752). from traumatic events among children and adolescents: A systematic review.
Dorsey, S., Briggs, E. C., & Woods, B. A. (2011). Cognitive-behavioral treatment for post- American Journal of Preventive Medicine, 35(3), 287–313.
traumatic stress disorder in children and adolescents. Child and Adolescent
Psychiatric Clinics of North America, 20(2), 255–269.
Dowd, H., & McGuire, B. E. (2011). Psychological treatment of PTSD in children: An
evidence-based review. The Irish Journal of Psychology, 32(1–2), 25–39.