Children and Youth Services Review 34 (2012) 1689–1694

Contents lists available at SciVerse ScienceDirect

Children and Youth Services Review

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

Predicting PTSD from the Child Behavior Checklist: Data from a field study with
children and adolescents in foster care
Rita Rosner a, b,⁎, Josephine Arnold a, Eva-Maria Groh a, Maria Hagl a
a
Department of Psychology, Ludwig-Maximilians-University Munich, Germany
b
Catholic University Eichstaett-Ingolstadt, Germany

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

Article history: Previous publications have suggested that treatment providers can screen children for posttraumatic stress
Received 17 January 2012 disorder (PTSD) by using specific subscales derived from the Child Behavior Checklist (CBCL). We tested
Received in revised form 12 April 2012 three CBCL-PTSD scales for their utility to screen for PTSD. 36 traumatized children and adolescents in foster
Accepted 13 April 2012
care were interviewed using the Clinician Administered PTSD Scale of Children and Adolescents (CAPS-CA).
Available online 21 April 2012
The children's foster parents completed the CBCL. CBCL-PTSD scales showed no or small to moderate, but
Keywords:
nonsignificant correlations with the number of PTSD symptoms and symptom severity. Overall, predictive
Child Behavior Checklist properties of the respective scales were not sufficient. Therefore, instead of using CBCL-PTSD subscales as
Posttraumatic stress disorder screens, we recommend the application of specific instruments for screening for PTSD.
Foster children © 2012 Elsevier Ltd. All rights reserved.
Screening

1. Introduction internalizing symptoms such as intrusions or feelings of detachment.

The American Academy of Child and Adolescent Psychiatry
(AACAP) practice parameter highlights “the importance of early
identification of posttraumatic stress disorder” (Cohen et al., 2010, p.
414) in order to prevent its chronification and associated impairment.
Yet assessing children and adolescents for posttraumatic stress
disorder (PTSD) is difficult for a number of reasons. First of all,
almost all measures begin by asking for a traumatic event, following
the DSM-IV definition of PTSD which requires the presence of trauma
exposure as a necessary condition for PTSD diagnosis. If no traumatic
event can be specified, in some instruments further assessment of
PTSD is deemed unnecessary. This procedure might result in a
substantial number of false negatives: Some children and adolescents
are reluctant to disclose a traumatic event; they may be too
embarrassed, distressed, or afraid to talk about what has happened;
perpetrators may have forbidden them to talk about it; or they do not
remember the event. Secondly, assessment is hampered by the
possible distress of the interviewees while discussing the traumatic
event, which makes some clinicians fear retraumatization or
aggravation of symptoms. Therefore, some clinicians avoid asking
explicitly about traumatic events and posttraumatic stress (PTS)
symptoms, especially when they do not have a specific reason to
suspect traumatization. Thirdly, PTSD is difficult to assess because it
occurs as a mixture of internalizing and externalizing symptoms.
Especially younger children have difficulties with reporting
⁎ Corresponding author at: Catholic University Eichstaett-Ingolstadt, Ostenstr, 26, D-
85071 Eichstätt, Germany. Tel.: + 49 8421 93‐1581; fax: + 49 8421 93‐2033.
E-mail address: rita.rosner@ku-eichstaett.de (R. Rosner).
A certain amount of insight is needed to recognize changes in one's
own mood, cognition, and behavior, especially concerning avoidance
(cf. Scheeringa, Zeanah, Drell, & Larrieu, 1995), and these symptoms
are not easily observed by caretakers either. With regard to
externalizing symptoms, however, caretakers' observations might be
a more reliable source. Yet Meiser-Stedman, Smith, Glucksman, Yule,
and Dalgleish (2007) found little evidence of, for example, hyperar-
ousal symptoms showing greater parent–child agreement than
reexperiencing symptoms. Overall, parent–child agreement for PTSD
was rather poor (see also Shemesh et al., 2005). Fourthly, PTS
symptoms seem to manifest differently in children than in adults, and
especially so in younger children (De Young, Kenardy, & Cobham,
2011; Scheeringa, Wright, Hunt, & Zeanah, 2006; Scheeringa et al.,
1995), so current DSM-IV criteria might not be appropriate for this
age group (Scheeringa, Peebles, Cook, & Zeanah, 2001; Scheeringa et
al., 1995). Finally, even if children are old enough to be regarded as
valid information source, the acknowledged clinical interviews for
PTSD, such as the Clinician Administered PTSD Scale of Children and
Adolescents (CAPS-CA; Nader et al., 1996), are time-consuming and
therefore not a part of routine assessment.
For all these reasons, clinicians and researchers search for
efficient, economic, and unobtrusive screening measures for PTS
symptoms. Because many health-care institutions routinely use the
Child Behavior Checklist (CBCL; Achenbach & Edelbrock, 1983) as
part of their standard assessment battery, developing a CBCL subscale
for PTS symptoms seems a sensible strategy. Several studies
examined the CBCL as a possible screening instrument for PTSD or
PTS symptoms. While some studies endeavored to discriminate
between children with and without PTSD by using the established

0190-7409/$ – see front matter © 2012 Elsevier Ltd. All rights reserved.
doi:10.1016/j.childyouth.2012.04.019
1690 R. Rosner et al. / Children and Youth Services Review 34 (2012) 1689–1694

CBCL syndrome scales (e.g., Saigh, Yasik, Oberfield, Halamandaris, & maintained 15 CBCL items (see Table 1). Their modified CBCL-PTSD
McHugh, 2002), other authors developed specific CBCL-PTSD sub- scale correlated significantly (r = .66) with the number of PTS
scales to screen for symptomatic children (Dehon & Scheeringa, 2006; symptoms reported by a caretaker in a semi-structured interview
Ruggiero & McLeer, 2000; Sim et al., 2005; Wolfe, Gentile, & Wolfe, preceding the CBCL assessment. The Internalizing (r = .57) and
1989). Wolfe et al. (1989) proposed a subset of 20 CBCL-PTSD items Externalizing (r = .42) scales also correlated significantly with
(see Table 1) by selecting items that conformed to the DSM-III criteria reported PTS symptoms. Furthermore, the authors showed in a
of PTSD. The authors studied 71 sexually abused children and regression analysis that the CBCL-PTSD scale had significant incre-
compared their scores with the CBCL normative sample. Sexually mental predictive power. Sim et al. (2005) applied a new strategy to
abused children scored about five times higher on the CBCL-PTSD determine whether the CBCL adequately screens for PTSD. They
scale. Ruggiero and McLeer (2000) further evaluated Wolfe et al.'s developed an item list based on expert ratings of all CBCL items and
item set. The authors compared 63 sexually abused children tested it with a confirmatory factor analysis in a sample of over 1700
(between 6 and 16 years of age) with two control groups (a children. The authors derived a PTSD subscale, a dissociation subscale,
psychiatric outpatient sample and a non-clinical school sample) and and a combined PTSD/dissociation scale from the CBCL. With the
found adequate internal consistency for the CBCL-PTSD scale PTSD scale, six items were identical to Wolfe et al.'s (1989) scale and
(Cronbach's α = .85), but questionable concurrent and poor discrim- one new item was added (see Table 1). However, the new CBCL-PTSD
inant validity. The scale did not discriminate between sexually abused scale did not significantly correlate (r = .26) with a self-report
children and not sexually abused psychiatric outpatients. However, measure for PTS symptoms that Sim et al. used for validation in a
within the sexually abused group, the CBCL-PTSD scale correlated subsample of 56 children (aged 8 to 12). Furthermore, the authors
significantly with the total number of PTS symptoms endorsed on a found no differences in CBCL-PTSD scores between clinical groups
structured interview for PTSD (r = .57), and the score was signifi- when comparing a not sexually abused psychiatric sample with a
cantly higher in sexually abused children with PTSD vs. sexually sample of children who were sexually abused. Sim and colleagues
abused children without a PTSD diagnosis. Nevertheless, the same conclude that the CBCL is not a good measure for PTSD in children and
was true for several other CBCL subscales, and also for the adolescents, and that parents may be poor raters for their children's
Internalizing scale and the Total Problem score. The authors posttraumatic symptomatology. Moreover, the CBCL-PTSD scale
calculated sensitivity and specificity and reported good sensitivity seems to reflect “generic distress, as opposed to trauma-related,
and weak to moderate specificity for a cut-off score of 8. Levendosky, distress” (p. 697). Recently, Loeb, Stettler, Gavila, Stein, and Chinitz
Huth-Bocks, Semel, and Shapiro (2002) assessed 62 preschool (2011) evaluated Dehon and Scheeringa's (2006) CBCL-PTSD scale in
children who had been exposed to domestic violence. For children a sample of 51 preschool-aged children with high trauma exposure
at the age of three, the respective CBCL version was used resulting in a and receiving outpatient child–parent psychotherapy for PTSD. In
14-item adaptation of the CBCL-PTSD scale of Wolfe et al. (1989). For particular, they tested the scale's validity in comparison to clinicians'
older preschool children, the authors used a 22-item adaptation, but DSM-IV diagnoses and to diagnoses based on the Diagnostic
without specifying which items they added. Results showed that the Classification of Mental Health and Developmental Disorders of Infancy
number of PTS symptoms directly reported by the mothers on a PTSD and Early Childhood (DC:0–3; Zero to Three, 2005) which provides a
rating scale adhering to DSM-IV criteria did not significantly correlate more developmentally sensitive and appropriate PTSD definition for
with the CBCL-PTSD scale. Dehon and Scheeringa (2006) assessed 62 very young children. Additionally, caregiver reports on trauma
traumatized children between the age of 2 and 6 years. Similar to exposure and symptoms were assessed using an established PTSD-
Levendosky et al. (2002), they removed all items from Wolfe et al.'s screening instrument. While children with clinician-based DSM-IV
scale that did not appear in the preschool form of the CBCL and diagnosis showed significantly higher CBCL-PTSD scores, there were
no significant differences in scores regarding PTSD diagnosis based on
DC:0–3 criteria. Moreover, receiver operator characteristic analyses
Table 1 yielded no adequate results for the CBCL-PTSD scale as a screening
Comparison of the CBCL-PTSD scales.
tool. CBCL-PTSD scores correlated significantly (r = .40) with the
Item Wolfe Dehon and Sim Item wording number of symptoms endorsed by parents on the PTSD screening, but
number et al. Scheeringa et al. not with the number of criteria endorsed by treating clinicians. Like
(1989) (2006) (2005)
Sim et al. (2005), Loeb et al. (2011) conclude that the CBCL‐PTSD
Item 3 X X Argues a lot subscale is not specific enough for screening for PTSD.
Item 8 X X Can't concentrate or can't pay Table 1 displays the CBCL‐PTSD subscales discussed above and
attention for long
Item 9 X X Can't get his/her mind off
their respective items.
certain thoughts; obsessions While all studies assessed PTSD or PTS symptoms, each used a
Item 11 X X Clings to adults or too dependent different measure: a self-report scale (Sim et al., 2005), or caretaker
Item 29 X X X Fears certain animals, situations, or rating (Levendosky et al., 2002), or a semi-structured caretaker
places other than school
interview (Dehon & Scheeringa, 2006), or caretaker rating plus
Item 34 X Feels others are out to get him/her
Item 45 X X X Nervous, high-strung, or tense clinician's diagnosis (Loeb et al., 2011). Ruggiero and McLeer (2000)
Item 47 X X X Nightmares used clinical diagnosis based on an interview that combined caretaker
Item 50 X X X Too fearful or anxious and self-report. Only Sim et al. (2005) and Ruggiero and McLeer
Item 52 X Feels too guilty (2000) assessed samples with children old enough to take into
Item 56b X Headaches
account self-reported symptoms too. Therefore, one reason for the
Item 56c X X Nausea, and feels sick
Item 56f X X Stomachaches mixed results so far might be the application of different assessment
Item 56g X X Vomiting, throwing up methods, the use of different sources of information, as well as the
Item 69 X Secretive, keeps things to self study of different age groups. Furthermore, the two studies with
Item 76 X Sleeps less than most kids
samples of school-aged children (i.e., Ruggiero & McLeer, 2000; Sim
Item 86 X X Stubborn, sullen, or irritable
Item 87 X X Sudden changes in mood or feelings et al., 2005) focused on PTSD according to DSM criteria, which have
Item 100 X X X Trouble sleeping been developed on the basis of research on adults, and have been
Item 103 X X Unhappy, sad, or depressed criticized as not appropriate for children and adolescents (Scheeringa
Item 111 X X Withdrawn, doesn't get involved et al., 2001). Therefore, alternative criteria are of interest. One set of
with others
criteria has been proposed by Scheeringa et al. (2001; Scheeringa,
 R. Rosner et al. / Children and Youth Services Review 34 (2012) 1689–1694
Zeanah, Myers, & Putnam, 2003) for children and adolescents.
Another set of criteria can be found in the WHO classification of
mental disorders, ICD-10 (WHO, 1992), which is the mandatory
coding system in most countries. In adults, the ICD-10 diagnosis of
PTSD is known to be less strict than the DSM-IV diagnosis.
Concordance between ICD and DSM diagnoses ranges between 35%
and 53% (Andrews, Slade, & Peters, 1999; Rosner & Powell, 2009).
Overall, the different outcomes of these different methods present
a puzzling picture. Studies that assessed PTS symptoms or PTSD with
a thorough clinical interview (i.e. Dehon & Scheeringa, 2006;
Ruggiero & McLeer, 2000) found significant correlation of CBCL-
PTSD scales and number of PTS symptoms. Studies that assessed PTS
symptoms with a caregiver symptom rating showed mixed results,
with Levendosky et al. (2002) reporting no correlation, but Loeb et al.
(2011) reporting significant correlation of the CBCL-PTSD scale with
the number of PTS symptoms. Using a self-report instrument, Sim et
al. (2005) found a moderate, but nonsignificant correlation. Finally,
while the use of an age-appropriate set of criteria seemed to have
worked well in the Dehon and Scheeringa (2006) study, using the
DC:0–3 criteria seemed to weaken the correlation in the Loeb et al.
(2011) study.
In summary, the empirical basis for the use of CBCL subscales to
screen for PTSD is still weak, and needs further investigation. In the
current study, we compared the various published versions of CBCL-
PTSD scales in their ability to screen for PTSD as measured in a
standardized clinical interview in a sample of foster children and
adolescents. Furthermore, we examined sensitivity and specificity not
only for DSM-IV criteria, but for ICD-10 criteria and the alternative
diagnostic criteria proposed by Scheeringa et al. (2001) as well.
2. Method
2.1. Participants
Participants were 36 foster children and their non-kin foster
parents. This sample was part of a larger study that examined
traumatic events in foster children, their current living conditions,
and the connection to psychological distress and PTS symptoms
(Rosner, Groh, Arnold, & Hagl, in preparation). Foster families were
recruited over a period of 18 months. Inclusion criteria were that the
foster child should be (a) between 10 and 18 years old, (b) living at
least for more than a month with the foster family, (c) living in
Bavaria, Germany; and (d) that the child should not have a mental
retardation (to the knowledge of the foster parents), i.e., is able to
answer interview questions. The study was approved by the
University's Ethics Review board.
For the purposes of this article, only data from children who
answered all questions of the clinical interview to assess PTSD were
included. This was done to ensure that no false negative cases were
included in the sample. Due to the aforementioned problem of
incomplete assessment when no index trauma is identified, a
diagnosis of PTSD could not be reliably rejected for the children
who did not report such a trauma. Participating children and
adolescents were on average 13.94 years old (SD = 2.44). 44% were
female and they lived between 1 month and 16 years with their foster
family, with a mean duration of almost 7 years (SD = 2.33).
2.2. Instruments
PTSD symptoms were assessed with the German version of the
Clinician Administered PTSD Scale for Children and Adolescents
(CAPS-CA; Nader et al., 1996; German version: Steil & Füchsel, 2006).
This structured clinical interview can be regarded as a gold standard
and reflects DSM-IV criteria for PTSD. Before asking for PTS
symptoms, the CAPS-CA offers a list of 17 possible traumatic events.
PTS symptoms according to DSM-IV are rated for frequency and
1691
intensity, and a total severity score can be calculated. In addition, a
categorical diagnosis according to DSM-IV or ICD-10 criteria is
available.
The Child Behavior Checklist (CBCL) by Achenbach and Edelbrock
(1983; German version: Döpfner et al., 1998) is a widely used and
well-validated measure for a range of behavior problems in children.
It consists of a list of 120 items answered on a three-point scale (i.e.,
not true, somewhat/sometimes true, very/often true; 0–2) by a
parent or caregiver. The CBCL yields eight syndrome scales (Aggres-
sive Behavior, Anxious/Depressed, Attention Problems, Rule-Breaking
Behavior, Social Problems, Somatic Complaints, Thought Problems,
and Withdrawn/Depressed) and three broad-band scales: Internaliz-
ing, Externalizing, and Total Problems. One foster parent completed
the CBCL, in most cases the foster mother (97%). The various CBCL-
PTSD scales were calculated by summing up the respective items.
2.3. Procedure
Most of the foster families were reached via youth welfare offices
and by a number of foster family organizations in Bavaria.
Additional foster children and their caregivers were recruited by
flyers and articles in daily newspapers and in a magazine for foster
parents, and by word of mouth. As we had no direct access to official
records of foster families, we depended on the support of the
aforementioned organizations and do not know how many foster
families were reached altogether or how many declined to
participate. Foster parents who were interested in participating
contacted our project office. A brief telephone screening about age,
time spent in the foster family, and mental retardation was
conducted to determine eligibility. Interviews were conducted in
the family's home by the second or third author, and lasted between
two and 4 h. Each child or adolescent received a voucher, worth €
15, for participating.
Altogether, 74 foster children were interviewed. Thirty-eight
children (51%) were excluded because they did not answer all
CAPS-CA questions. Most of them (30 children, i.e., 40% overall) were
not able to give detailed information about a possible traumatic event
due to their young age when the event occurred. These children knew
about an accident or violent incident only from hearsay. Eight
children (11%) reported a traumatic event, but were too distressed
to talk about it: They either asked to terminate the interview or the
interviewer discontinued because the child showed signs of ongoing
dissociation. Although these children most likely fulfilled PTSD
criteria, they were excluded from current analysis because the
diagnosis could neither be confirmed nor disconfirmed. Again, this
procedure was chosen to avoid false negatives. All in all, 36 foster
children and their foster parents were included in the current
analysis.
2.4. Data analysis
No child had to be excluded because of missing CBCL data.
Individual missing items were replaced by the mean of the respective
subscale if 80% of the subscale had been answered, which was always
the case. To compare sensitivity (for identifying true positives) and
specificity (for identifying true negatives) for the different CBCL-PTSD
scales, we calculated receiver operating characteristic (ROC) curves.
ROC curves are graphical plots of the sensitivity vs. specificity at
different cut-off points. The area under the curve (AUC) quantifies the
overall ability of a scale to discriminate between those individuals
with PTSD and those without. A truly useless scale has an area of 0.5
(which means true positives could be identified equally accurate by
flipping a coin). A perfect scale has an area of 1, indicating a
sensitivity and specificity of 100%.
1692 R. Rosner et al. / Children and Youth Services Review 34 (2012) 1689–1694

3. Results

3.1. Overall distress and PTS symptoms

On average, children and adolescents showed elevated scores on

all three CBCL broad-band scales: Internalizing (M = 57.86;
SD = 9.08), Externalizing (M = 59.94; SD = 11.29), and Total Prob-
lems (M = 61.69; SD = 9.46) were all about one standard deviation
above German norms (Döpfner et al., 1998). Means and SDs of the
respective CBCL-PTSD scales can be found in Table 2.
Thirty-three children (92%) reported one or more traumatic events
in their past. The average number of PTS symptoms endorsed on the
CAPS-CA was 6 (M = 5.58; SD = 4.07), with a minimum of 0 and a
maximum of 15 symptoms per child. The most frequently reported
symptoms were recurrent and intrusive distressing recollections of
the event (78%), avoidance of thoughts, feelings, or conversations
associated with the trauma (58%), and intense psychological distress
when exposed to internal or external cues that symbolize or resemble
an aspect of the traumatic event (44%). In this sample, 4 of the 36
children (11%) met diagnostic criteria for PTSD according to DSM-IV,
but 16 children (44%) according to the less stringent ICD-10 criteria. If
the alternative diagnostic criteria proposed by Scheeringa et al. (2003)
were used to establish diagnostic status, 10 children (28%) were Fig. 1. ROC curves for three CPCL-PTSD scales predicting PTSD according to DSM-IV
diagnosed as suffering from PTSD. criteria in foster children (N = 36).

3.2. Psychometric evaluation of CBCL-PTSD scales nonsignificant extent with the number of endorsed PTS symptoms.
The Internalizing scale showed no association at all, neither with
In a first step, reliabilities of the various PTSD scales were symptom number or severity. The apparent moderate, but nonsig-
examined. The CBCL-PTSD scale by Wolfe et al. (1989) demonstrated nificant correlation of the Total Problems scale is probably due to the
an acceptable internal consistency (Cronbach's α = .73). The internal included items from the Externalizing scale.
consistencies of the other two scales were somewhat lower, probably Changes in sensitivity and specificity at different cut-off points are
because they contain fewer items, with α = .67 for Dehon and illustrated by ROC curves for each scale predicting PTSD according to
Scheeringa (2006) and α = .63 for the scale by Sim et al. (2005). DSM-IV criteria (see Fig. 1). The area under the ROC curve (AUC) is
None of the correlations between the number of PTS symptoms .51 (95% CI: .19–.82) for the CBCL-PTSD scale according to Sim et al.
endorsed on the CAPS-CA, total symptom severity score in the CAPS- (2005), which means its capacity to predict PTSD is equivalent to
CA, and the three CBCL-PTSD scales reached significance, although the random guessing. The other scales received better results with an
scales of Wolfe et al. (1989) and Dehon and Scheeringa (2006) AUC of .81 (95% CI: .67–.95) for the scale by Dehon and Scheeringa
showed small to moderate correlations with the CAPS-CA's symptom (2006) and .75 (95% CI: .56–.94) for the scale developed by Wolfe et
severity score (see Table 2). Concerning the three CBCL broad-band al. (1989), but only for Dehon and Scheeringa's scale the AUC was
scales, only Externalizing problems correlated significantly with PTS significantly different from .50, p = .044.
symptom severity, r = .39, p = .020, and to a smaller, yet Additionally, we estimated ROC curves based on PTSD diagnosis
according to ICD‐10 and according to the adapted criteria proposed
by Scheeringa et al. (2003), both representing less strict PTSD
Table 2 definitions. Results were even less favorable than using DSM‐IV
Means (SD) of CBCL and CBCL-PTSD scales and correlation with PTS symptoms (CAPS- PTSD definition as outcome and none of the scales showed an AUC
CA). score that was significantly different from chance. With a PTSD
M (SD) ra (CBCL × CAPS-CA)
diagnosis according to ICD-10 as outcome criteria, the AUC statistic
was .37 for Sim et al.'s (2005) scale, .53 for the scale developed by
No. of PTS Symptom
Wolfe et al. (1989), and .53 for the scale by Dehon and Scheeringa
symptomsb severity scorec
(2006). When considering the child-adapted DSM-criteria of
CBCL-PTSD scale
Scheeringa et al. (2003), AUC for the CBCL-PTSD scale by Sim et al.
Wolfe et al. (1989) 8.42 (4.75) .11 .21
Dehon and Scheeringa (2006) 6.58 (3.74) .16 .27 was .46, for Wolfe et al.'s scale .66, and for Dehon and Scheeringa's
Sim et al. (2005) 2.64 (2.49) .00 .12 scale .68. (Because none of these AUC scores was significantly
CBCLd different from .50, the respective CI's are not reported.) Thus, only
Internalizing 57.86 (9.08) .06 .04 when applying the very strict PTSD criteria of the DSM-IV, two of the
Externalizing 59.94 (11.29) .28 .39⁎
Total Problems 61.69 (9.46) .23 .30
CBCL-PTSD scales—the scale by Wolfe et al. and the scale by Dehon
and Scheeringa—yielded acceptable predictive properties. Because
Note. N = 36. CBCL = Child Behavior Checklist; PTSD = posttraumatic stress disorder;
only the latter's AUC score was significantly different from chance, we
CAPS-CA = Clinician Administered PTSD Scale for Children and Adolescents; PTS =
posttraumatic stress. The CBCL-PTSD scales have different item numbers, with Wolfe et will focus on the 15-item scale by Dehon and Scheeringa, which is
al. (1989): 20 items; Dehon and Scheeringa (2006): 15 items; Sim et al. (2005): 7 appropriate for the preschool form of the CBCL as well. The best cut-
items; off point was a score of ≥8, with a sensitivity of 1 and a specificity of
a
Pearson correlation coefficient. .69. Using this cut-off resulted in the correct classification of all 4
b
Number of PTS symptoms endorsed on the CAPS-CA.
c
Total symptom severity score on the CAPS-CA.
positive cases of the present sample and yielded no false negatives. It
d
Scores are T-scores (Döpfner et al., 1998). did, however, yield 10 false positives, which means it is erring on the
⁎ p b .05. side of sensitivity.
 R. Rosner et al. / Children and Youth Services Review 34 (2012) 1689–1694
4. Discussion
Specifically developed CBCL-PTSD subscales have been used in
several studies on trauma in children; however, it is questionable
whether those subscales assess the genuine response to a traumatic
event or rather general distress. We found that two of the CBCL
derived subscales, namely the CBCL-PTSD scales by Wolfe et al.
(1989) and by Dehon and Scheeringa (2006), might prove useful as
screening instruments for PTSD in traumatized children. But,
because we used a very conservative way to diagnose PTSD
(DSM-IV plus CAPS-CA plus exclusion of children with unclear
memories plus exclusion of dissociating children), our sample
included only 4 cases of PTSD. Numbers this small somewhat
restrict the interpretation of our results. Although the AUC scores of
both scales were substantial, confidence intervals were large, and
only the CBCL-PTSD scale by Dehon and Scheeringa (2006) yielded
a significant result, i.e., proved to be reliably better than chance in
identifying possible PTSD cases. For statistically stronger results, a
much larger sample would have been necessary, including more
genuine PTSD cases.
Therefore, the optimal cut-off score of 8 that we propose for
Dehon and Scheeringa's (2006) scale awaits further corroboration. In
any case, such a cut-off would yield a very high sensitivity which is
preferable for a screening instrument. At the same time, an overly
sensitive screening instrument might lead to many false positives
and thereby tempts to overdiagnose PTSD. This is especially
important in the light of Sim et al.'s (2005) concern regarding
specificity. The authors concluded that the CBCL-PTSD scales
probably reflect generic distress rather than trauma-related distress,
which means that the items in these scales might tap PTS symptoms,
but not specifically so. Children with general distress and arousal
might be rated as if they were as strongly affected as actually
traumatized children concerning these items. In our sample, the
broad-band scale Externalizing problems correlated significantly
with PTS symptom severity, and even more so than the CBCL-PTSD
scales, perhaps because of the much higher item number (33 items).
However, if the carefully selected items of the CBCL-PTSD scales
would tap PTS symptoms and only PTS symptoms, they should fare
better in predicting PTSD than a broad-band scale summarizing
externalizing symptoms. In a similar vein, Ayer et al. (2009) argue
that the Child Behavior Checklist-Juvenile Bipolar Disorder (CBCL-
JBD; Biederman et al., 1995) does not reliably predict bipolar
disorder in youth, but rather reflects a general dysregulatory
syndrome that simply comprises the three syndromes Aggressive
Behavior, Anxious/Depressed, and Attention Problems, analogous to
the respective syndrome scales that were used for the CBCL-JBD.
Furthermore, Ayer et al. (2009) showed that the CBCL-JBD overlaps
with Achenbach and Rescorla's (2007) own CBCL subscale for
Posttraumatic Stress Problems (CBCL-PTSP; Achenbach & Rescorla,
2007) in regard to an underlying latent variable. The authors
conclude that both the CBCL-JBD and the CBCL-PTSP—which in turn
shares items with the three CBCL-PTSD scales evaluated in the
present study—measure a single syndrome related to severe
psychopathology and disordered self-regulation, and thus are not
useful screening instruments for a diagnosis of PTSD or bipolar
disorder.
In this light, and although our results attest to the usefulness of
two of the herein evaluated CBCL-PTSD scales, we do not recommend
them as screening instruments for PTSD at this time. Further studies
with larger and more representative samples are needed and should
not only examine possible correlations between these scales and a
PTSD diagnoses. Future studies should also concentrate on specificity
to avoid mixing overall distress with trauma symptoms.
Apart from the small sample size, there are further limitations to
our study. First, the current sample included only children and
adolescents aged between 10 and 18, which means that the results
1693
cannot be generalized to younger children. Secondly, the sample is
rather specific. Foster children often have a very troubled past with
more than one distinct traumatic event (Oswald, Heil, & Goldbeck,
2010). In our sample, some of the children experienced severe
interpersonal trauma. Thus, it may be difficult to generalize these
findings to other populations, especially to children having experi-
enced single-event trauma. It is possible that some children in our
sample manifested a more complex trauma syndrome that was not
adequately captured with the DSM-IV concept of PTSD (cf. DeJong,
2010). Our diagnostic procedure may have been too conservative for
severe trauma of that kind. Thirdly, in our study, foster parents
instead of the birth parents completed the CBCL. One might argue
that foster parents are not reliable informants of their foster
children's behavior problems. However, one could also argue that
foster parents have a unique and authoritative perspective on their
foster children because they have as much opportunities to observe
the children as birth parents, but they might be less biased
concerning their psychological problems, because they do not share
biological vulnerabilities or the often traumatizing experiences that
led to foster placement. Yet research on this question is scarce
(Tarren-Sweeney, Hazell, & Carr, 2004). In a comparison between
birth parents, foster parents, and teachers, Randazzo, Landsverk, and
Ganger (2003) found some evidence that parental depressive
symptoms distorted CBCL reports, concerning internalizing behavior
in particular. Overall, although foster parent's reports were more
strongly correlated to teachers' reports, they did not differ signifi-
cantly from birth parents' reports; that is, there was no sign of
symptom under- or overreporting in foster parents. Compared to the
teachers, foster parents showed less agreement concerning internal-
izing behavior than externalizing behavior. This was also the case in
Tarren-Sweeney et al.'s (2004) study. Therefore, we do not believe
that foster parents are less accurate than birth parents when
reporting on behavior problems of the children in their care and
consider this limitation as a minor one. Finally, our results are not
directly comparable to the results of those studies that relied on
caretaker report on PTS symptoms when evaluating predictive
power of the CBCL-PTSD scales. In our study, the children
themselves were questioned about their PTS symptoms, which
means, we correlated information from two different sources,
which should weaken the possible association in any case, but
especially in light of research showing that parent–child agreement
on the child's PTS symptoms is poor (Meiser-Stedman et al., 2007;
Shemesh et al., 2005). Nevertheless, we view the use of a well-
validated structured interview (the CAPS-CA) rather as a methodo-
logical strength. Shemesh et al. (2005) showed in a sample of
children and adolescents (aged 8–19) that using their ratings in a
screening tool for PTSD provided better fit to a “best estimate
diagnostic procedure” (p. 584) than parents' ratings. In case of acute
stress disorder in children, parental report failed to predict later
child or parental report of PTSD and the authors conclude that
children and adolescents should be directly interviewed (Meiser-
Stedman et al., 2007). If the CBCL-PTSD scales are to be shown as
valid predictors of PTS symptoms, they have to be evaluated against
a gold standard criterion appropriate for the respective age group.
This means, in children and adolescents, their predictive value
should be reflected in substantial correlations with child report as
well, going beyond correlation of the same source, that is, with
parental report.
In summary, we recommend caution in using CBCL-PTSD scales for
screening for PTSD in children and adolescents. Instead, we
recommend applying specific PTSD screening instruments. The
AACAP practice parameter (Cohen et al., 2010) suggests a number
of instruments and during the last years further promising screening
instruments have been developed (Kenardy, Spence, & Macleod,
2006; Nixon, Ellis, Nehmy, & Ball, 2010; Winston, Kassam-Adams,
Garcia-España, Ittenbach, & Cnaan, 2003).
1694 R. Rosner et al. / Children and Youth Services Review 34 (2012) 1689–1694
Role of the funding source
This research was supported by scholarships from Deutsche
Gesellschaft für Verhaltenstherapie (DGVT; German Society for
Behavior Therapy) and friendly support from Deutsches Jugendin-
stitut (DJK, German Youth Institute). Neither of these was involved in
design, analysis or interpretation.
References
Achenbach, T. M., & Edelbrock, C. (1983). Manual for the Child Behavior Checklist and
revised child behavior profile. Burlington: University of Vermont, Department of
Psychiatry.
Achenbach, T. M., & Rescorla, L. A. (2007). Multicultural supplement to the manual for the
ASEBA school-age forms & profiles. Burlington, VT: University of Vermont, Research
Center for Children, Youth, & Families.
Andrews, G., Slade, T., & Peters, L. (1999). Classification in psychiatry: ICD 10 versus
DSM-IV. The British Journal of Psychiatry, 174, 3–5.
Ayer, L., Althoff, R., Ivanova, M., Rettew, D., Waxler, E., Sulman, J., et al. (2009). Child
Behavior Checklist‐Juvenile Bipolar Disorder (CBL‐JBD) and CBCL Posttraumatic
Stress Problems (CBCL‐PTSP) scales are measures of a single dysregulatory
syndrome. Journal of Child Psychology and Psychiatry, and Allied Disciplines, 50,
1291–1300.
Biederman, J., Wozniak, J., Kiely, K., Ablon, S., Faraone, S., Mick, E., et al. (1995). CBCL
clinical scales discriminate prepubertal children with structured interview derived
diagnostic of mania from those with ADHD. Journal of the American Academy of
Child and Adolescent Psychiatry, 34, 464–471.
Cohen, J. A., Bukstein, O., Walter, H., Benson, H. R., Chrisman, A., Farchione, T. R., et al.
(2010). Practice parameter for the assessment and treatment of children and
adolescents with posttraumatic stress disorder. Journal of the American Academy of
Child and Adolescent Psychiatry, 49, 414–430.
De Young, A. C., Kenardy, J. A., & Cobham, V. E. (2011). Diagnosis of posttraumatic stress
disorder in preschool children. Journal of Clinical Child and Adolescent Psychology,
40, 375–384.
Dehon, C., & Scheeringa, M. (2006). Screening for preschool posttraumatic stress
disorder with the Child Behavior Checklist. Journal of Pediatric Psychology, 31,
431–435.
DeJong, M. (2010). Some reflections on the use of psychiatric diagnosis in the looked
after or “in care” child population. Clinical Child Psychology and Psychiatry, 15,
589–599.
Döpfner, M., Plück, J., Bölte, S., Lenz, K., Melchers, P., & Heim, K. (1998). Elternfragebo-
gen über das Verhalten von Kindern und Jugendlichen. Deutsche Bearbeitung der Child
Behavior Checklist (CBCL/4–18). Einführung und Anleitung zur Handauswertung.
Köln: KJFD, Arbeitsgruppe Kinder-, Jugend- und Familiendiagnostik.
Kenardy, J. A., Spence, S. H., & Macleod, A. C. (2006). Screening for posttraumatic stress
disorder in children after accidental injury. Pediatrics, 118, 1002–1009.
Levendosky, A. A., Huth-Bocks, A. C., Semel, M. A., & Shapiro, D. L. (2002). Trauma
symptoms in preschool-age children exposed to domestic violence. Journal of
Interpersonal Violence, 17(2), 150–164.
Loeb, J., Stettler, E. M., Gavila, T., Stein, A., & Chinitz, S. (2011). The Child Behavior
Checklist PTSD scale: Screening for PTSD in young children with high exposure to
trauma. Journal of Traumatic Stress, 24, 430–434.
Meiser-Stedman, R., Smith, P., Glucksman, E., Yule, W., & Dalgleish, T. (2007). Parent
and child agreement for acute stress disorder, post-traumatic stress disorder and
other psychopathology in a prospective study of children and adolescents exposed
to single-event trauma. Journal of Abnormal Child Psychology, 35, 191–201.
Nader, K., Kriegler, J. A., Blake, D. D., Pynoos, R. S., Newman, E., & Weathers, F. W.
(1996). Clinician Administered PTSD Scale, Child and Adolescent Version. White River
Junction, VT: National Center for PTSD.
Nixon, R. D., Ellis, A. A., Nehmy, T. J., & Ball, S. (2010). Screening and predicting
posttraumatic stress and depression in children following single incident trauma.
Journal of Clinical Child and Adolescent Psychology, 39, 588–596.
Oswald, S. H., Heil, K., & Goldbeck, L. (2010). History of maltreatment and mental
health problems in foster children: A review of the literature. Journal of Pediatric
Psychology, 35, 462–472.
Randazzo, K., Landsverk, J., & Ganger, W. (2003). Three informants' reports of child
behavior: Parents, teachers, and foster parents. Journal of the American Academy of
Child and Adolescent Psychiatry, 42, 1343–1350.
Rosner, R., Groh, E.-M., Arnold, J., & Hagl, M. (in preparation). PTBS bei Pflegekindern:
Häufigkeit und Prädiktoren.
Rosner, R., & Powell, S. (2009). Does ICD-10 overestimate the prevalences of PTSD?
Effects of differing diagnostic criteria on estimated rates of posttraumatic stress
disorder in war zone exposed civilians. Trauma & Gewalt, 3(2), 1–8.
Ruggiero, K. J., & McLeer, S. V. (2000). PTSD scale of the Child Behavior Checklist:
Concurrent and discriminant validity with non-clinic-referred sexually abused
children. Journal of Traumatic Stress, 13, 287–299.
Saigh, P. A., Yasik, A. E., Oberfield, R. A., Halamandaris, P. V., & McHugh, M. (2002). An
analysis of the internalizing and externalizing behaviors of traumatized urban
youth with and without PTSD. Journal of Abnormal Psychology, 111, 462–470.
Scheeringa, M. S., Peebles, C. C., Cook, C. A., & Zeanah, C. H. (2001). Toward establishing
procedural, criterion, and discriminiant validity for PTSD in early childhood.
Journal of the American Academy of Child and Adolescent Psychiatry, 40, 52–60.
Scheeringa, M. S., Wright, M. J., Hunt, J. P., & Zeanah, C. H. (2006). Factors affecting the
diagnosis and prediction of PTSD symptomatology in children and adolescents. The
American Journal of Psychiatry, 163, 644–651.
Scheeringa, M. S., Zeanah, C. H., Drell, M. J., & Larrieu, J. A. (1995). Two approaches to
the diagnosis of posttraumatic stress disorder in infancy and early childhood.
Journal of the American Academy of Child and Adolescent Psychiatry, 34, 191–200.
Scheeringa, M. S., Zeanah, C. H., Myers, L., & Putnam, F. W. (2003). New findings on
alternative criteria for PTSD in preschool children. Journal of the American Academy
of Child and Adolescent Psychiatry, 42, 561–570.
Shemesh, E., Newcorn, J. H., Rockmore, L., Shneider, B. L., Emre, S., Gelb, B. D., et al.
(2005). Comparisons of parent and child reports of emotional trauma symptoms in
pediatric outpatient settings. Pediatrics, 115, 582–589.
Sim, L., Friedrich, W. N., Davies, W. H., Trentham, B., Lengua, L., & Pithers, W. (2005).
The Child Behavior Checklist as an indicator of Posttraumatic Stress Disorder and
dissociation in normative, psychiatric, and sexually abused children. Journal of
Traumatic Stress, 18, 697–705.
Steil, R., & Füchsel, G. (2006). IBS-KJ (Interviews zu Belastungsstörungen bei Kindern und
Jugendlichen). Göttingen: Hogrefe.
Tarren-Sweeney, M. J., Hazell, P. L., & Carr, V. J. (2004). Are foster parents reliable
informants of children's behaviour problems? Child: Care, Health and Development,
30, 167–175.
WHO [World Health Organisation] (1992). ICD 10 classification of mental and
behavioural disorders: Clinical descriptions and diagnostic guidelines. Geneva: WHO.
Winston, F. K., Kassam-Adams, N., Garcia-España, J. F., Ittenbach, R., & Cnaan, A. (2003).
Screening for risk of persistent posttraumatic stress in injured children and their
parents. JAMA: The Journal of the American Medical Association, 290, 643–649.
Wolfe, V. V., Gentile, C., & Wolfe, D. A. (1989). The impact of sexual abuse on children:
A PTSD formulation. Behavior Therapy, 20, 215–228.
Zero to Three (2005). Diagnostic classification of mental health and developmental
disorders of infancy and early childhood (re. ed.). Washington DC: Author.