discussion 2

For the week’s topics of Anxiety and Obsessive-Compulsive Disorders and Trauma: Abuse, Neglect, and Domestic Violence, analyze the primary arguments presented in either one of additional articles posted on Canvas OR  a relevant empirical, peer-reviewed article of your choosing.

Discuss how the author’s perspective contributes to the broader academic conversation on these subjects. Reflect on the strengths and limitations of the author’s arguments, providing specific examples from the text. Include your critical evaluation of the evidence presented and how it supports or contradicts other sources you have encountered or your current knowledge of the study of abnormal child psychology. Ensure you properly cite (APA formatting, 7th edition) the additional articles from Canvas in your discussion.

Feel free to let me know if you need any more assistance.

Behaviour Research and Therapy 41 (2003) 1–10
www.elsevier.com/locate/brat

Cognitive predictors of posttraumatic stress disorder in
children: results of a prospective longitudinal study

A. Ehlersa,*, R.A. Mayoub, B. Bryantb

a Department of Psychology, Institute of Psychiatry, De Crespigny Park, Denmark Hill, London SE5 8AF, UK
b Department of Psychiatry, University of Oxford, Oxford, UK

Accepted 30 October 2001

Abstract

The present study explored whether cognitive factors specified in the Ehlers and Clark model (Behav.
Res. Ther. 38 (2000) 319) of posttraumatic stress disorder (PTSD) predict chronic PTSD in children who
had experienced a road traffic accident. Children were assessed at 2 weeks, 3 months, and 6 months after
the accident. Data-driven processing during the accident, negative interpretation of intrusive memories,
alienation from other people, anger, rumination, thought suppression and persistent dissociation at initial
assessment predicted PTSD symptom severity at 3 and 6 months. On the basis of sex and stressor severity
variables, 14% of the variance of PTSD symptoms at 6 months could be explained. The accuracy of the
prediction increased to 49% or 53% when the cognitive variables measured at initial assessment or 3
months, respectively, were taken into account.
 2003 Elsevier Science Ltd. All rights reserved.

Keywords:Posttraumatic stress disorder; Children; Cognitive predictors; Prospective study; Cognitive model

1. Introduction

There is increasing awareness that posttraumatic stress disorder (PTSD) is common in children
who have experienced traumatic events such as road traffic accidents (RTAs). Recent estimates
suggest that between 14 and 34% of children involved in an RTA will develop PTSD (e.g. Bryant,
Mayou, Wiggs, Ehlers, & Stores, 2001; Canterbury & Yule, 1997; Di Gallo, Barton, & Parry-
Jones, 1997; Ellis, Stores, & Mayou, 1998; Mirza, Bhadrinath, Goodyear, & Gilmore, 1998; Stal-

* Corresponding author. Tel.:+44-20-7848-5033; fax:+44-20-7848-0591.
E-mail address:a.ehlers@iop.kcl.ac.uk (A. Ehlers).

0005-7967/03/$ – see front matter 2003 Elsevier Science Ltd. All rights reserved.
PII: S0005-7967(01)00126-7

2 A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

lard, Velleman, & Baldwin, 1998). Very little is known, however, about factors that contribute
to the development and maintenance of PTSD in children.

The present study explored whether psychological variables that have been shown to predict
PTSD after an RTA in adults also predict PTSD in children. In previous research with adult RTA
survivors, objective indicators of trauma severity such as injury severity were poor predictors of
PTSD symptoms. In contrast, the individuals’ subjective response to the event, in particular their
perceived threat to life, was consistently found to be a significant predictor (e.g. Blanchard et al.,
1995; Ehlers, Mayou, & Bryant, 1998; Pynoos et al., 1987; see also review by March, 1993).
Nevertheless, subjective stressor severity only explained a small proportion of the variance of
chronic PTSD symptoms. The accuracy of the prediction could be substantially improved if main-
taining psychological factors derived from Ehlers and Clark’s (2000) model of PTSD were taken
into account (Ehlers, Mayou et al., 1998; Mayou, Ehlers, & Bryant, in press).

The Ehlers and Clark (2000) model highlights three factors thought to determine the develop-
ment and maintenance of PTSD.

1. Trauma memory deficits. The memory for the traumatic event is poorly elaborated and inad-
equately integrated with other autobiographical information. This leads (together with strong
priming and conditioning for associated cues) to easy triggering of re-experiencing symptoms
when matching cues are present. The deficit in elaboration/integration of the trauma memory
is due to (a) incomplete cognitive processing of the event while it is happening, and (b) cogni-
tive avoidance after the event which prevents a change in memory. Ehlers and Clark (2000)
specified three overlapping indicators of incomplete processing during the event, data-driven
processing (e.g. processing the sensory characteristics of the situation rather than its meaning),
lack of self-referent processing, and dissociation.

2. Appraisals. The individual makes excessively negative appraisals of the traumatic event and/or
its sequelae (including the initial PTSD symptoms), leading to a sense of current threat (see
also Ehlers & Steil, 1995; Steil & Ehlers, 1995).

3. Maintaining behaviours and cognitive strategies. The negative appraisals motivate the individ-
ual to engage in a range of dysfunctional behaviours and cognitive strategies that are intended
to control the perceived current threat, but maintain the problem (see also Ehlers & Steil,
1995; Steil & Ehlers, 1995). Examples include thought suppression, avoidance, rumination, and
persistent dissociation.

Evidence for the role of each of these three factors has been accumulated in a series of cross-
sectional and prospective longitudinal studies of adult trauma survivors. First, indicators of trauma
memory deficits and incomplete processing during trauma (data-driven processing, lack of self-
referent processing and dissociation) predicted PTSD in adult RTA and assault survivors (Murray,
Ehlers, & Mayou, in press; Halligan, Michael, Clark, & Ehlers, 2001). Second, negative appraisals
of the trauma and its sequelae were strongly related to PTSD severity across a range of different
traumas (Dunmore, Clark, & Ehlers, 1999, 2001; Foa, Ehlers, Clark, Tolin, & Orsillo, 1999).
Common negative appraisals of the trauma in PTSD include overgeneralization of danger, global
negative thoughts about the self, preoccupation with unfairness and self-blame (Foa et al., 1999;
Foa, Riggs, Massier, & Yarczower, 1995). Trauma sequelae that are often interpreted negatively
by trauma survivors include the initial symptoms of PTSD, e.g. intrusive memories may be inter-

3A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

preted as a sign of going crazy (Clohessy & Ehlers, 1999; Ehlers & Steil, 1995; Ehlers, Mayou
et al., 1998; Steil & Ehlers, 2000), and the reactions of others in the aftermath of the event that
may be interpreted as signs of alienation (Dunmore et al., 1999, 2001; Ehlers, Clark et al., 1998;
Ehlers, Maercker, & Boos, 2000). Third, several studies of RTA survivors and ambulance staff
have supported the maintaining role of rumination and thought suppression (Clohessy & Ehlers,
1999; Ehlers, Mayou et al., 1998; Murray et al., in press; Steil & Ehlers, 2000).

Dissociation has received special attention in many studies. It represents an indicator of incom-
plete processing during trauma and is thus thought to lead to problematic trauma memories (e.g.
Foa & Hearst-Ikeda, 1996; Spiegel & Cardena, 1990). Indeed, several prospective studies have
found that dissociation during trauma predicts subsequent PTSD (Ehlers et al., 1998; Halligan et
al., 2001; Koopman, Classen, & Spiegel, 1994; Shalev, Peri, Canetti, & Schreiber, 1996; Murray
et al., in press). However, dissociation can also be a more persistent response style thought to
prevent a change in trauma memories. Murray et al. (in press) and Halligan et al. (2001) found
that persistent dissociation was a better predictor of PTSD at 6 months after an RTA or an assault
than dissociation during the trauma.

The present study was designed to explore whether the Ehlers and Clark (2000) model can be
applied to children and adolescents. A prospective longitudinal study assessed children and ado-
lescents who had been involved in an RTA at 2 weeks, 3 months, and 6 months. The assessment
had to be brief to make the study feasible. Therefore, only a few key variables could be chosen
for the investigation:

1. Data-driven processing during the accident, as one of the indicators of incomplete processing.
2. Appraisal measures that are thought to lead to a sense of current threat: negative appraisals of

intrusive memories, alienation from other people, and anger as a measure of preoccupation
with unfairness.

3. Dysfunctional cognitive strategies hypothesized to maintain PTSD: rumination, thought sup-
pression, persistent dissociation, and, as children often do not have control over exposure to
reminders, their parents’ attitude favouring avoidance strategies to deal with the event.

The study investigated whether these variables predict PTSD severity in children and ado-
lescents at 3 and 6 months after an RTA, and whether they predict PTSD over and above what
can be predicted from measures of objective and subjective stressor severity.

2. Method

The study was part of an investigation into the prevalence of children’s psychological symptoms
such as posttraumatic stress symptoms, travel anxiety, sleep disturbance, and behavioural problems
in the aftermath of an RTA. The prevalence data are presented in Bryant et al. (2001).

2.1. Participants

Children resident in Oxfordshire, UK, and aged 5–16 years who were passengers, pedestrians
or cyclists involved in an RTA and who were taken to the emergency department of the John

4 A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

Radcliffe Hospital, Oxford, in July 1997–July 1998 were recruited into a prospective study. A
total of 150 children from 137 families were invited to take part in the study. Of these, 86 children
(58%) from 80 families agreed to participate. For the 3 and 6 months assessments, data from 81
(94%) and 82 (95%) of the children who participated in initial assessment were available. At 2
weeks after the accident, 15% of the children met diagnostic criteria for acute stress disorder,
and at 3 and 6 months, 25 and 18% of the children met criteria for PTSD, respectively (see
Bryant et al., 2001, for a full description).

Non-participation was not related to the age or sex of the child, nor to the type of accident.
Those with less severe injury were less likely to take part. Telephone conversations with the
parents who declined participation suggested that the proportion of acute stress disorder in non-
participants was comparable to that of the participants (see Bryant et al., 2001).

Fifty-five per cent of the participants were boys. Mean age was 12.3 years, SD 2.86. About
half of the participants were teenagers, and one-fifth were under 10 years old. Most of the children
had contracted soft tissue injuries (73%), 23% had bony injuries, and 4% remained uninjured; 21
had been admitted to hospital.

2.2. Measures

2.2.1. Symptoms of post-traumatic stress disorder
The dependent variable was the severity of PTSD symptoms as defined by DSM-IV (American

Psychiatric Association, 1994). When the study was planned, the best validated measures of PTSD
symptoms in children (see review by McNally, 1996) were the children’s version of the Impact
of Event Scale (IES, Horowitz et al., 1979; children’s version by Yule and colleagues, e.g. Yule &
Williams, 1990) and the Child Post-traumatic Stress Reaction Index (RI, Pynoos et al., 1987). As
items of these scales have been shown to be appropriate for children, we used them to represent
the DSM-IV symptoms whenever possible. Participants were instructed to rate the symptoms on
a scale from 0 ‘no’ to 3 ‘yes, often’ . The total PTSD severity score was the sum of the scores
for the 17 DSM-IV symptoms. If none of the IES or RI items measured a DSM-IV symptom,
the authors constructed a new item; for example, symptom C6 was assessed with the questions:
“ Is it difficult for you to have strong feelings? For example, do you find it hard to get really
excited or happy, or do you find it hard to cry when you are sad?” . To avoid duplication of
questions, sleep disturbance was scored from the Sleep Behavior Questionnaire (Simonds & Par-
raga, 1982). Some DSM-IV items were represented by two items, for example the IES items “Do
pictures of the accident pop into your mind” and “Do you think about the accident even if you
don’ t mean to” both represented symptom B1. The maximum score for these items was used in
the overall severity score. For young children, the parent attending the interview (usually the
mother) complemented the child’s answers to the items and also provided information on symp-
toms of repetitive play and reenactment.

2.3. Predictor variables

2.3.1. Stressor severity measures
The Injury Severity Score (ISS) of the Abbreviated Injury Scale (AIS, American Association

for Automotive Medicine, 1985) assessed injury severity. Information was taken from the hospital

5A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

case notes. On the AIS, each injury is coded on a six-point scale from 1 ‘minor’ to 6 ‘maximum’ .
The ISS is the sum of the squares of the highest AIS score in each of the three most injured ISS
body regions (head or neck, face, chest, abdominal, extremities). In addition, for comparability,
two measures of injury severity as used in our previous study of adult survivors of RTAs were
included. First, we assessed whether children had no injury, soft issue injuries, or bony injuries
to distinguish between forms of minor injuries. Second, we assessed whether or not children were
admitted to hospital.

Two measures of subjective stressor severity were taken. First, children indicated whether or
not they thought they were going to get hurt or die (perceived threat to life/physical integrity).
Second, they indicated the extent to which they felt scared/frightened during the accident and
while in hospital on a scale from 1 ‘not scared’ to 3 ‘a lot’ . The fear response score was the
maximum of these two answers.

As presented elsewhere, characteristics of the accident and the children’s age and previous
psychological problems were not related to PTSD (Bryant et al., 2001).

2.3.2. Data-driven processing of the RTA
Participants indicated the extent to which they were muddled/confused during the accident on

a scale from 1 ‘not muddled’ to 3 ‘a lot’ .

2.3.3. Appraisal measures
These measures were rated on a scale from 0 ‘no’ to 3 ‘yes, often’ . Negative interpretation of

intrusive memories was measured as the response to the question “Do you ever think that some-
thing is wrong with you because you cannot forget the accident, for example, do you ever feel
you are going mad?” . Alienation from other people was measured as the response to the question
“Do you feel like other people really don’ t understand what you went through?” . As an indirect
measure of appraisals relating to unfairness, children were asked to rate “Do you get angry when
you think about the accident?” .

2.3.4. Maintaining cognitive strategies
These measures were rated on a scale from 0 ‘no’ to 3 ‘yes, often’ . As in Ehlers, Mayou et

al. (1998), rumination was scored as the mean of two items. Participants rated whether they kept
going over the accident over and over again and whether they kept thinking again and again about
why the accident happened to them. Thought suppression was measured as the response to the
question “ If pictures of the accident pop into your mind do you try to stop them and push them
out of your mind again?” . As a measure of persistent dissociation, three symptoms not included
in the symptoms of PTSD were used, i.e. feeling in a daze, feelings of unreality, and feelings of
depersonalisation. Parental attitude favouring avoidance was measured by asking the parent to
rate how helpful they thought it would be to avoid reminders of the accident, for the child to push
the memories of the accident out of his/her mind, and to act as if the accident had not happened.

3. Results

Table 1 shows the correlations of the predictor variables and PTSD symptom severity at 3 and
6 months after the accident. Sex and measures of injury severity were not significantly related to

6 A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

Table 1
Prediction of children’s PTSD symptom severity at 3 and 6 months after a motor vehicle accident

Predictor PTSD severity

3 months 6 months

Female gender �0.01 0.13
Indices of objective accident severity

Type of injury 0.07 0.00
ISS score 0.01 0.12
Admission to hospital 0.03 0.15

Indices of subjective accident severity
Perceived threat to life/physical 0.37** 0.31**
integrity
Maximum fear during accident 0.28* 0.25*
or in hospital

Cognitive factors assessed at 2 weeks
Trauma memory measures

Data-driven processing during 0.30* 0.22(*)

accident
Appraisal measures

Negative interpretation of 0.36** 0.35**
intrusions
Alienation 0.37** 0.41***
Anger 0.30* 0.30*

Maintaining cognitive strategies
Rumination 0.31* 0.22(*)

Thought suppression 0.29* 0.26*
Persistent dissociation 0.51*** 0.42***
Parental avoidant attitude 0.02 0.21(*)

Cognitive factors assessed at 3 months
Appraisal measures

Negative interpretation of 0.27*
intrusions
Alienation 0.55***
Anger 0.41***

Maintaining cognitive strategies
Rumination 0.55***
Thought suppression 0.35**
Persistent dissociation 0.61***

(*)p�0.10, *p�0.05, **p�0.01, ***p�0.00.

PTSD symptoms. The children’s perceived threat to life/physical integrity and their degree of fear
during the accident and while in hospital showed small significant correlations with subsequent
PTSD. As expected, the cognitive variables predicted subsequent PTSD symptoms. Correlations
were small to moderate.

Hierarchical multiple regression analyses tested whether the cognitive factors predict PTSD
symptom severity over and above what can be predicted on the basis of sex and stressor severity.

7A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

In the first step, sex, hospital admission (as the injury measure with the highest correlation with
PTSD symptoms), perceived threat to life and fear during the accident/time in hospital were
entered into the regression function. These variables predicted 14% of the variance, R2=0.136,
F(4,57)=2.234, p=0.077. In the second step, we entered the cognitive predictors measured at 2
weeks into the equation, with the exception of data-driven processing as there were some missing
data for this variable. The accuracy of the prediction increased significantly to 49% variance
explained, R2 change=0.356, F change (7,50)=4.993, p=0.001, R2=0.491, F (11,50)=4.388,
p�0.001.

The analysis was repeated using the cognitive variables measured at 3 months in the second
step of the hierarchical regression. Again, the cognitive variables significantly improved the pre-
diction of PTSD severity at 6 months, and together with sex and stressor severity explained 53%
of the variance, R2 change=0.385, F change (5,51)=8.441, p�0.001, R2=0.534, F (9,51)=6.501,
p�0.001.

4. Discussion

4.1. Do the cognitive variables derived from Ehlers and Clark’s model predict PTSD?

The results support the role of cognitive predictors of chronic PTSD in children. Our prospec-
tive longitudinal study showed that cognitive factors measured soon after an RTA predict PTSD
symptom severity at 3 and 6 months after the accident. Nearly all of the cognitive variables
showed significant correlations with PTSD severity.

Data-driven processing during the accident was related to subsequent PTSD symptoms at 3
months and showed a trend for a correlation at 6 months. This pattern of results replicates findings
in adult survivors of trauma (Halligan et al., 2001) and is in line with the hypothesis that data-
driven processing (like other indicators of incomplete processing) during trauma is involved in
the initial development of PTSD, and that its influence on PTSD in the long term depends on the
presence of maintaining factors.

The evidence for such maintaining factors in the present children sample was strong. All indi-
cators of negative appraisals of the trauma and its sequelae, i.e. negative interpretation of intrusive
memories, perceived alienation from others, and anger (as an indicator of appraisals relating to
unfairness) were significant predictors of PTSD at 3 and 6 months. The results replicate those
found with adult survivors of a range of traumas (Clohessy & Ehlers, 1999; Dunmore et al., 1999,
2001; Ehlers, Mayou et al., 1998; Ehlers et al., 2000; Steil & Ehlers, 2000), and are in line with
two recent studies by Steil, Hempt, & Deffke, 2001) who found that negative appraisals of intrus-
ive memories and the trauma correlated highly with PTSD symptom severity in children and
adolescents after RTAs. Ehlers and Clark (2000) propose that these appraisals lead to a sense of
current threat and prevent the trauma survivor from putting the trauma behind them. They also
motivate the use of dysfunctional behaviours and cognitive strategies that maintain PTSD (see
also Steil & Ehlers, 2000).

As expected, such dysfunctional cognitive strategies were also correlated with subsequent PTSD
severity in the children sample, i.e. rumination, suppression of intrusive memories, and persistent
dissociation. For rumination, the relationship with subsequent PTSD appeared to become stronger

8 A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

with time. It is possible that some degree of rumination is quite common in the initial weeks
after a traumatic event, and that only persistent rumination strongly predicts PTSD. Similar results
were obtained in adults by Murray et al. (in press). The results for suppression of intrusive memor-
ies parallel those of Steil and colleagues in a cross-sectional and a prospective study of children
and adolescents after RTAs (Steil, Gundlach et al., 2001; Steil, Hempt et al., 2001).

Parental attitude favouring avoidance strategies only showed a trend for a correlation with the
children’s PTSD severity at 6 months. It is possible that measures of parental behaviour rather
than attitude would have been more predictive. It is also possible that a measure of the child’s
perception of the parents’ behaviour would have been more predictive, as indicated by recent
data by Steil, Gundlach et al. (2001). Alternatively, in the present sample, the influence of the
parents on their children’s way of coping with the accident and the symptoms arising from it
may have been limited, especially since about 50% of the sample comprised teenagers.

4.2. Do the cognitive variables predict PTSD over and above other predictors?

The power of the cognitive variables in predicting subsequent PTSD symptoms has to be inter-
preted against a background of other variables that are potential predictors. Sex, injury severity
and other accident and sample characteristics did not significantly predict PTSD symptom severity
(Bryant et al., 2001). In line with other studies (reviewed by March, 1993), indicators of the
subjective stressor severity, i.e. perceived threat to life/physical integrity and the children’s fear
response were significant predictors. However, the cognitive predictors derived from the Ehlers
and Clark (2000) model predicted PTSD severity over and above what could be predicted from
subjective stressor severity, increasing the accuracy of the prediction from 14% to about 50%
explained variance.

4.3. Limitations and conclusions

The present study had several strengths and weaknesses. Among the strengths was the use of
a prospective longitudinal design and the recruitment from a consecutive sample of patients.
Among the weaknesses was a modest participation rate and the use of few or single items to
measure the constructs. The analysis of responses of parents who declined participation suggested,
however, that the remaining sample was not biased towards a higher or lower PTSD rate. Similar
to studies with adult RTA survivors, children who had contracted minor injuries were less likely
to participate than children with more severe injuries. It cannot be determined whether this affected
the patterns of correlations reported in the study, although it seems unlikely, given that injury
severity was unrelated to PTSD. However, the correlations of the cognitive predictors and PTSD
severity in the present study are likely to underestimate the true relationship as the use of single
items rather than multi-item scales will have introduced some error variance due to measurement
error. Future studies are warranted that use standardized questionnaires of established reliability
to measure the cognitive constructs. Such questionnaires have already been developed for adult
populations, and with adaptions in wording may prove to be useful in the prediction of chronic
PTSD in children (see also Steil, Gundlach et al., 2001; Steil, Hempt et al., 2001).

9A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

Acknowledgements

The study was funded by a grant from the Wellcome Trust. Anke Ehlers is a Wellcome Principal
Research Fellow. We thank Luci Wiggs and Ann Day for their help with the study. We are
grateful for the collaboration of the John Radcliffe Accident and Emergency Services. We would
also like to thank the parents and children who participated in the study.

References

American Association for Automotive Medicine (1985). The abbreviated injury scale. Des Plaines, IL: American
Association for Automotive Medicine.

American Psychiatric Association (1994). Diagnostic and statistical manual of mental disorders (DSM-IV) (4th ed.).
Washington, DC: APA.

Blanchard, E. B., Hickling, E. J., Mitnick, N., Taylor, A. E., Loos, W., & Buckley, T. C. (1995). The impact of severity
of injury and perception of life threat in the development of posttraumatic stress disorder after motor vehicle acci-
dents. Behaviour Research and Therapy, 33, 529–534.

Bryant, B., Mayou, R.A., Wiggs, L., Ehlers, A., & Stores, G. (2001). Psychological consequences of road traffic
accidents in children and their mothers. Submitted for publication.

Canterbury, R., & Yule, W. (1997). The effects in children of road traffic accidents. In M. Mitchell (Ed.), The aftermath
of road traffic accidents. London: Routledge.

Clohessy, S., & Ehlers, A. (1999). PTSD symptoms, response to intrusive memories, and coping in ambulance service
workers. British Journal of Clinical Psychology, 38, 251–265.

Di Gallo, A., Barton, J., & Parry-Jones (1997). Road traffic accidents: early psychological consequences in children
and adolescents. British Journal of Psychiatry 170, 358–362.

Dunmore, E., Clark, D. M., & Ehlers, A. (1999). Cognitive factors involved in the onset and maintenance of PTSD.
Behaviour Research and Therapy, 37, 809–829.

Dunmore, E., Clark, D. M., & Ehlers, A. (2001). A prospective study of the role of cognitive factors in persistent
posttraumatic stress disorder after physical or sexual assault. Behaviour Research and Therapy, 39, 1063–1984.

Ehlers, A., & Clark, D. M. (2000). A cognitive model of posttraumatic stress disorder. Behaviour Research and Therapy,
38, 319–345.

Ehlers, A., Clark, D. M., Dunmore, E. B., Jaycox, L., Meadows, E., & Foa, E. B. (1998). Predicting response to
exposure in PTSD: the role of mental defeat and alienation. Journal of Traumatic Stress, 11, 457–471.

Ehlers, A., Maercker, A., & Boos, A. (2000). PTSD following political imprisonment: the role of mental defeat, alien-
ation, and permanent change. Journal of Abnormal Psychology, 109, 45–55.

Ehlers, A., Mayou, R. A., & Bryant, B. (1998). Psychological predictors of chronic PTSD after motor vehicle accidents.
Journal of Abnormal Psychology, 107, 508–519.

Ehlers, A., & Steil, R. (1995). Maintenance of intrusive memories in posttraumatic stress disorder: a cognitive approach.
Behavioural and Cognitive Psychotherapy, 23, 217–249.

Ellis, A., Stores, G., & Mayou, R. (1998). Psychological consequences of road traffic accidents in children. European
Child and Adolescent Psychiatry, 7, 61–68.

Foa, E. B., Ehlers, A., Clark, D. M., Tolin, D., & Orsillo, S. (1999). The Post-traumatic Cognitions Inventory (PTCI).
Development and validation. Psychological Assessment, 11, 303–314.

Foa, E. B., & Hearst-Ikeda, D. (1996). Emotional dissociation in response to trauma: an information-processing
approach. In L. K. Michelson, & W. J. Ray (Eds.), Handbook of dissociation: theoretical, empirical, and clinical
perspectives. New York: Plenum Press.

Foa, E. B., Riggs, D. S., Massier, E. D., & Yarczower, M. (1995). The impact of fear activation and anger on the
efficacy of exposure treatment for PTSD. Behavior Therapy, 26, 487–500.

Halligan, S.L., Michael., T, Clark, D.M., & Ehlers, A. (2001). Posttraumatic stress disorder following assault: the role
of cognitive processing, trauma memory, and appraisals. Submitted for publication.

10 A. Ehlers et al. / Behaviour Research and Therapy 41 (2003) 1–10

Horowitz, M., Wilner, N., & Alvarez, W. (1979). Impact of event scale: a measure of subjective stress. Psychosomatic
Medicine, 41, 209–218.

Koopman, C., Classen, C., & Spiegel, D. (1994). Predictors of posttraumatic stress symptoms among survivors of the
Oakland/Berkeley Calif., firestorm. American Journal of Psychiatry, 151, 888–894.

March, J. S. (1993). What constitutes a stressor? The ‘A’ criterion issue. In R. T. Davidson, & E. B. Foa (Eds.),
Posttraumatic stress disorder and beyond (pp. 37–56). Washington, DC: American Psychiatric Press.

Mayou, R.A., Ehlers, A., & Bryant, B. (in press). Posttraumatic stress disorder after motor vehicle accidents: 3-year
follow-up of a prospective longitudinal study. Behaviour Research and Therapy.

McNally, R. J. (1996). Assessment of posttraumatic stress disorder in children and adolescents. Journal of School
Psychology, 34, 147–161.

Mirza, K. A. H., Bhadrinath, B. R., Goodyear, I. M., & Gilmour, C. (1998). Post-traumatic stress disorder in children
and adolescents following road traffic accidents. British Journal of Psychiatry, 172, 443–447.

Murray, J., Ehlers, A., & Mayou, R.A. (in press). Dissociation and posttraumatic stress disorder: two prospective studies
of motor vehicle accident survivors. British Journal of Psychiatry.

Pynoos, R. S., Frederick, C., Nader, K., Arroyo, W., Steinberg, A., Eth, S., Nunez, F., & Fairbanks, L. (1987). Life
threat and posttraumatic stress in school-age children. Archives of General Psychiatry, 44, 1057–1063.

Shalev, A., Peri, T., Canetti, L., & Schreiber, S. (1996). Predictors of PTSD and injured trauma survivors: a prospective
study. American Journal of Psychiatry, 153, 219–225.

Simonds, J. F., & Parraga, H. (1982). Prevalence of sleep disorders and sleep behaviors in children and adolescents.
Journal of the American Academy of Child Psychiatry, 21, 383–388.

Spiegel, D., & Cardena, E. (1990). Dissociative mechanisms in posttraumatic stress disorder. In M. E. Wolf, & A. D.
Mosnaim (Eds.), Posttraumatic stress disorder: etiology, phenomenology, and treatment. Washington: American
Psychiatric Press.

Stallard, P., Velleman, R., & Baldwin, S. (1998). Prospective study of post-traumatic stress disorder in children involved
in road traffic accidents. British Medical Journal, 317, 1619–1623.

Steil, R., & Ehlers, A. (2000). Dysfunctional meaning of posttraumatic intrusions in chronic PTSD. Behaviour Research
and Therapy, 38, 537–558.

Steil, R., Gundlach, P., Müller, S., & Schier, F. (2001). Posttraumatic stress disorder in child and adolescent survivors
of motor vehicle accidents—is a cognitive model applicable to children and adolescents? Submitted for publication.

Steil, R., Hempt, A., & Deffke, I. (2001). PTSD in children and adolescents. Paper at the World Conference of the
Association for the Advancement of Behavior Therapy, Vancouver, Canada, 21.7.2001.

Yule, W., & Williams, R. M. (1990). Post-traumatic stress reactions in children. Journal of Traumatic Stress, 3,
279–295.

• Cognitive predictors of posttraumatic stress disorder in children: results of a prospective longitudinal study

Introduction

Method

Participants

Measures

Symptoms of post-traumatic stress disorder

Predictor variables

Stressor severity measures

Data-driven processing of the RTA

Appraisal measures

Maintaining cognitive strategies

Results

Discussion

Do the cognitive variables derived from Ehlers and Clark’s model predict PTSD?

Do the cognitive variables predict PTSD over and above other predictors?

Limitations and conclusions

Acknowledgements

References

The triad of symptoms following a traumatic experience
has been termed posttraumatic stress disorder (PTSD)
(American Psychiatric Association 1994). Approximately
20% of individuals exposed to a significant traumatic
event will develop PTSD (Breslau et al., 1998, 1999b),
and children may be at even higher risk (Breslau et al.,
1999a). Depending on the sample being studied and the
methodology used, prevalence rates have ranged widely
in at-risk child populations (American Academy of Child
and Adolescent Psychiatry, 1998). For example, 34% of
urban youths exposed to community violence have been
reported to develop PTSD (Berman et al., 1996), chil-
dren exposed to physical injury have had rates of 23%
(Aaron et al., 1999), and children exposed to maltreat-

ment (i.e., physical and or sexual abuse) have been shown
to have rates of 36% (Ackerman et al., 1998) or higher
(see McLeer et al., 1998).

These prevalence rates highlight the need for under-
standing the progression and expression of PTSD in chil-
dren. Children’s initial response to trauma is often
characterized by physiological and behavioral hyperarousal,
and when the trauma is ongoing, the response may become
complicated by dissociation (Carrion and Steiner, 2000;
Perry et al., 1995). Moreover, research suggests that indi-
viduals who experience chronic trauma have lower rates
of recovery from PTSD (Famularo et al., 1996; Green,
1985; Terr, 1991). With evidence suggesting that chil-
dren’s response to trauma can be enduring and detrimental,
it has become increasingly important to ensure develop-
mentally appropriate classification criteria.

Research has demonstrated that developmentally sen-
sitive assessment of symptoms after trauma may be more
valid than the DSM-IV criteria in very young children,
because symptoms of PTSD may differ substantially
between children and adults (Scheeringa et al., 1995,
2001). For example, reenactments are more likely to be
expressed in play rather than through verbalizations. In
addition, children are less likely to have flashbacks (Terr,
1983; Terr et al., 1999).

Accepted September 12, 2001.
From Stanford University School of Medicine, Stanford, CA.
This research was supported by an NIMH Minority Research Supplement

Award to Dr. Victor G. Carrion and Dr. Allan L. Reiss (RO1 MH50047). The
authors thank Elana Newman, Ph.D., Natalie Pageler, Jessica Letchemanan,
and the California San Mateo and San Francisco counties for their participa-
tion in this project.

Reprint requests to Dr. Carrion, Division of Child and Adolescent Psychiatry
and Child Development, Stanford University, Stanford, CA 94305-5719;
e-mail: vcarrion@stanford.edu.

0890-8567/02/4102–0166�2002 by the American Academy of Child and
Adolescent Psychiatry.

Toward an Empirical Definition of Pediatric PTSD:
The Phenomenology of PTSD Symptoms in Youth

VICTOR G. CARRION, M.D., CARL F. WEEMS, PH.D., REBECCA RAY, M.A., AND ALLAN L. REISS, M.D.

ABSTRACT

Objective: To examine the frequency and intensity of posttraumatic stress disorder (PTSD) symptoms and their relation

to clinical impairment, to examine the requirement of meeting all DSM-IV symptom cluster criteria (i.e., criteria B, C, D),

and to examine the aggregation of PTSD symptom clusters across developmental stages. Method: Fifty-nine children

between the ages of 7 and 14 years with a history of trauma and PTSD symptoms were assessed with the Clinician-

Administered PTSD Scale for Children and Adolescents. Results: Data support the utility of distinguishing between the

frequency and the intensity of symptoms in the investigation of the phenomenology of pediatric PTSD. Children fulfilling

requirements for two symptom clusters did not differ significantly from children meeting all three cluster criteria with

regard to impairment and distress. Reexperience (cluster B) showed increased aggregation with avoidance and numb-

ing (cluster C) and hyperarousal (cluster D) in the later stages of puberty. Conclusions: Frequency and intensity of

symptoms may both contribute to the phenomenology of pediatric PTSD. Children with subthreshold criteria for PTSD demon-

strate substantial functional impairment and distress. J. Am. Acad. Child Adolesc. Psychiatry, 2002, 41(2):166–173. Key

Words: posttraumatic stress disorder, trauma, phenomenology, development.

166 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002

Frequency of symptoms has also been studied in older
adolescents. This research has indicated that distressing
recollections (reexperience), efforts to avoid thoughts and
feelings (avoidance/numbing), and efforts to avoid activ-
ities that facilitate recollections (avoidance/numbing) are
the most common symptoms in 16- to 22-year-olds
(Cuffe et al., 1998). However, data not only on the fre-
quency but also on the intensity of the specific DSM-IV
symptoms are needed. Previous research on pediatric anx-
iety has shown that the intensity of worry is important in
discriminating normative from clinical worry (Weems
et al., 2000). These authors found that the intensity of
worry differentiates children referred for anxiety disorder
treatment (n = 119, 6 to 16 years old) from a large nor-
mative comparison sample of nonreferred children, whereas
frequency of worry did not differentiate the samples.

DSM-IV took a step forward with the introduction of
developmental modifications to criterion A (experience
of trauma) and the symptom cluster B criteria (reexperi-
ence). Although symptom clusters C (avoidance and numb-
ing) and D (hyperarousal) have not had developmental
modifications, each of the symptom cluster criteria must
be met before the diagnosis may be assigned in children
and adolescents. There is little empirical evidence that the
tripartite clustering of symptoms that depict adult PTSD
appropriately characterize pediatric PTSD. In fact, to our
knowledge, no studies have examined the clinical impor-
tance of the co-occurrence of the symptom clusters in a
pediatric population. Research has indicated that partial
symptomatology is common (Aaron et al., 1999; Cuffe
et al., 1998) and may be disabling to the point of requir-
ing treatment even when full criteria are not met (see
Pfefferbaum, 1997). Information on the impairment result-
ing from partial symptomatology and the aggregation of
the clusters may help provide future developmental mod-
ifications to the current diagnostic criteria.

The aims of this study were thus to (1) examine the
frequency and intensity of all 17 DSM-IV PTSD symp-
toms and evaluate the relation of these symptoms to clin-
ical impairment, (2) examine the requirement of meeting
all DSM-IV symptom cluster criteria (i.e., criteria B, C,
and D) in a pediatric sample, and (3) examine the aggre-
gation (i.e., co-occurrence) of PTSD symptom clusters
across developmental stages.

METHOD

Participants

The sample was recruited from local social service departments and
mental health clinics. All of the children in this sample were referred

to the project because of exposure to interpersonal trauma. Therapists
and caseworkers were the referring sources. We recruited only chil-
dren who (1) had at least one episode of exposure to trauma, as defined
by DSM-IV criterion A1 (American Psychiatric Association, 1994);
(2) had undergone the trauma episode or episodes for which the indi-
vidual was referred at least 6 months before referral; (3) had no known
history of neurological disorders; and (4) had no known history of
alcohol or drug abuse/dependence.

Sixty children were referred to this study. Consent was obtained
from the participating counties’ courts for those subjects in foster place-
ment (n = 27). Most cases had prior child protective services involve-
ment (n = 35). A procedure was in place to report any suspected ongoing
maltreatment; however, no cases were identified. The principal inves-
tigator (V.G.C.) presented all subjects and their caretakers, regardless
of prior court consent, with a written institutional review board–approved
informed consent at a scheduled visit. All referred children underwent
screening with the Child PTSD Reaction Index and were assessed with
the Clinician-Administered PTSD Scale for Children and Adolescents
(CAPS-CA). Only one subject was not able to complete the CAPS-
CA because of scheduling problems, and he was not included in this
analysis. The final sample consisted of 34 boys and 25 girls for a total
sample of 59 children. The mean age of the children was 10.6 years,
with a range of 7 to 14 years. Most children (55%) had experienced
multiple traumatic events. Traumatic events included separation and
loss (55%), witnessing violence (40%), physical abuse (37%), sexual
abuse (20%), physical neglect (12%), and emotional abuse (7%). In
terms of family income, 48.4% reported incomes between 0 and
$31,000, 15.0% reported incomes between $31,000 and $76,000, and
14.9% of the families reported incomes over $76,000; 21.7% did not
report income data (because the children were in foster care, residen-
tial treatment, or other nontraditional rearing environment). In terms
of education, caregivers reported partial high school education (3.3%),
a high-school education (21.7%), partial college (18.3%), college
(11.7%), or graduate school education (16.7%). Educational back-
ground was not available for 28.3% of the sample. Ethnic composi-
tion was Euro-American (n = 25), African-American (n = 26), Hispanic
(n = 5), Asian (n = 2), and other (n = 1).

Instruments

The CAPS-CA (Nader et al., 1996), a structured clinical interview,
is a developmentally sensitive counterpart to the CAPS for adults
(Blake et al., 1995). It facilitates assessment of exposure to criterion
A1 events and the individuals’ experience of these events (A2), fre-
quency and intensity for each of the 17 symptoms for PTSD clus-
tered in DSM-IV (i.e., criteria B, C, and D), and the 1-month duration
requirement (criterion E). Additional features to increase the utility
of this instrument with children include iconic representations of the
behaviorally anchored 5-point frequency and intensity rating scales,
opportunities to practice with the format before questions, and a stan-
dard procedure for identification of the critical 1-month time frame
for current symptoms. The CAPS-CA also helps evaluate the impact
of symptoms (i.e., impairment; criterion F) on functioning and the
overall distress related to PTSD symptoms. In this sample, the CAPS-
CA total score was also significantly correlated with the Reaction Index
(r = 0.51, p < .01). A certified child psychiatrist (V.G.C.) who was trained on the administration of the instrument conducted the CAPS- CA interview. Moreover, an intraclass correlation coefficient of 0.97 was established on a subsample of the interviews in the present sam- ple with one of the originators of the CAPS-CA (Dr. Elana Newman), who rated videotaped recordings of 10 interviews.

The Child PTSD Reaction Index is a 20-item self-report instru-
ment used to assess PTSD symptoms after exposure to violence

PEDIATRIC PTSD

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002 167

(Pynoos et al., 1987). It is a widely used instrument and has been shown
to be a valid and reliable measure of PTSD symptoms in pediatric sam-
ples (Nader et al., 1990). This instrument was administered as a ques-
tionnaire. Subjects were instructed to ask questions if they did not
understand any item or if they could not read the questionnaire. When
this was the case, the questionnaire was read to them. All subjects
underwent interview assessment with the CAPS-CA after screening
with the Reaction Index.

The Schedule for Affective Disorders and Schizophrenia for School-
Age Children–Present and Lifetime Version (K-SADS-PL) is a semi-
structured clinical interview designed to identify Axis I DSM-IV disorders
(Kaufman et al., 1997). This instrument was administered to the par-
ticipants by a certified child psychiatrist (V.G.C.) to assess comorbidity.

Participants’ pubertal development was determined by self-report.
Participants selected from drawings with written descriptions repre-
senting the five Tanner stages (Marshal and Tanner, 1970) of pubic hair
development and genital development for boys and breast development
for girls. Children’s median pubic hair Tanner stage was 2, for girls
median breast Tanner stage was 3, and for boys median genital Tanner
stage was 2. For developmental analysis, the sample was split accord-
ing to combined Tanner stage 2 and below 2 (53%) or above 2 (47%).

The Wechsler Abbreviated Scales of Intelligence (WASI) was used
to determine intelligence (Psychological Corporation, 1999). The WASI
is a nationally standardized test that yields IQ scores that correlate with
subscales of the Wechsler Intelligence Scale for Children, Third Edition
(WISC-III). Full-scale IQs ranged from 65 to 142, average score 90
(5 subjects scored below 70 but were included in the sample because
they did not meet criteria for mental retardation because of higher
adaptive behavioral functioning and their ability to participate in the
clinical evaluation).

Current caretakers completed the Child Behavior Checklist (CBCL)
(Achenbach, 1991), a 113-item rating scale that assesses children’s
behavioral and social problems. The CBCL provides scores for both
Internalizing and Externalizing subscales. CBCL scaled scores and clin-
ical cut-points have been found to discriminate between clinic-referred
and nonreferred children, and normative data are available (Achenbach,
1991). The subscales were used as a cross-informant measure of chil-
dren’s symptoms. The CBCL has good reliability and has been exten-
sively validated (Achenbach, 1991). Because some children were in
foster care, residential treatment, or other nontraditional rearing envi-
ronment, 48 children had complete CBCL data.

Hypotheses and Data Analyses

To examine the relation of each symptom’s frequency and inten-
sity to meeting full diagnostic criteria, we addressed the following
questions: (1) which symptoms are most associated with full diagno-
sis? and (2) which symptoms are most associated with impairment?
SPSS was used for all data analyses, including descriptive statistics,
and employed an α level of p < .05 (two-tailed). We used simple lin- ear regression to examine the association between each symptom’s fre- quency (i.e., how often) and each symptom’s intensity (i.e., how distressing, how uncontrollable) and PTSD diagnosis and also with overall clinical impairment. Diagnostic status was coded (1 for full PTSD, 0 for not full PTSD). Because of the exploratory nature of the analyses and our interest in the size of the effect, we retained an α level of p < .05 (two-tailed) as recommended for such analyses (see Cohen, 1994; Jensen at al. 2001) and focused on variables that accounted for at least 10% of the variance as a way of emphasizing the size of the association as opposed to the α significance of the test statistic.

To examine the diagnostic requirement of meeting all DSM-IV
symptom cluster criteria (i.e., criteria B, C, and D), the cumulative
importance of the symptom clusters to impairment and distress was

examined by dividing the sample into three groups. Children meet-
ing all three DSM-IV PTSD symptom cluster criteria (PTSD-3) were
compared with children meeting two (PTSD-2) and children meet-
ing one (PTSD-1) of the symptom cluster criteria. On the basis of
our clinical observation of functional impairment in children mani-
festing only some of the symptoms of PTSD, we hypothesized that
children in the PTSD-1 group would show less clinical impairment
than youths in the PTSD-2 and PTSD-3 groups but that there would
be no difference in terms of clinical impairment between PTSD-2
and PTSD-3 children. ANOVAs or χ2 analyses were used to compare
the subgroups in terms of demographics, the time since the traumatic
experiences, types of traumas, comorbidity, distress, impairment, and
pubertal status. Multiple comparisons between group means across
the three subgroups were implemented with the Fisher least-signifi-
cant-difference procedure.

We hypothesized that the symptom clusters would show increased
aggregation with the emergence of puberty, because recent research
indicates that the transition to puberty may be an important time in
the development of internalizing symptoms, such as panic (Hayward et al.,
2000). Because the reexperience criterion (criterion B) does have spe-
cific developmental modifications, we expected this symptom cluster
to be reported at the highest rates and that avoidance/numbing and
hyperarousal (C and D, respectively) would be more common in later
stages of puberty. We also hypothesized that symptom clusters C and
D would show increased aggregation with cluster B with the emer-
gence of puberty. We used χ2 analyses to examine the convergence of
the symptom clusters.

RESULTS

Fourteen of the 59 children (24%) met full diagnos-
tic criteria for PTSD on the CAPS-CA (criteria A through
F). Children within the sample demonstrated a variety
of comorbid psychiatric conditions as assessed by the K-
SADS. The top six individual comorbid DSM-IV con-
ditions were depressive disorder not otherwise specified
(NOS) (12%), major depressive disorder (11%), atten-
tion-deficit/hyperactivity disorder (ADHD) (11%), spe-
cific phobia (9%), separation anxiety disorder (7%), and
social phobia (7%). Individual disorders were collapsed
into the following categories: “mood” (23%; children
with either depressive disorder NOS or major depressive
disorder), “externalizing” (17%; children meeting crite-
ria for ADHD, conduct disorder, or oppositional defiant
disorder), and “anxiety” disorders (48%; children meet-
ing criteria for DSM-IV anxiety disorders other than
PTSD). Although DSM-IV does not include ADHD as
a disruptive behavior disorder, these conditions were com-
bined in the “externalizing” category to differentiate them
from anxiety and mood disorders. In addition, to pro-
vide cross-informant confirmation on externalizing diag-
noses, we examined the association between the CBCL
Externalizing subscale and inclusion in the Externalizing
disorders group. The correlation was r = 0.39, p < .05.

CARRION ET AL.

168 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002

The frequency and intensity of each of the DSM-IV
PTSD symptoms are presented in Table 1. As can be seen
in the table, the top five most frequent symptoms were
avoidance of thoughts, feelings, and conversations asso-
ciated with the trauma (DSM-IV cluster C, symptom
number 1 [C1], 83.0%); distressing recollections (A1)
and the inability to recall important aspects of the event
(C3), both 70.0%; and distressing dreams (B2) and dif-
ficulty concentrating (D3), both 64.0%. The three most
intense symptoms were irritability/anger (D2), distress-
ing dreams (B2), and detachment from others (C5).

Frequency, Intensity, and Impairment

Separate regression analyses were conducted for the
frequency and intensity of each symptom on the CAPS-
CA and PTSD diagnosis. Results are summarized in Table
1. As the table shows, the intensity of some symptoms
predicted PTSD diagnosis or functional impairment inde-
pendently of frequency. Overall, full PTSD diagnosis was
most strongly associated with detachment from others,
hypervigilance, and exaggerated startle for both frequency
and intensity of symptoms. Separate regression analyses
were conducted using the reported frequency and inten-
sity of each symptom and clinical impairment (assessed
by the CAPS-CA composite clinical impairment index).
The results are presented in Table 1. Overall impairment
was most strongly associated with distressing recollec-

tions, distressing dreams, and the inability to recall impor-
tant aspects of the event for both frequency and inten-
sity of the symptoms. Finally, the intensity of some
symptoms predicted PTSD diagnosis or functional impair-
ment when frequency did not, and the frequency of some
symptoms predicted PTSD diagnosis or functional impair-
ment when intensity did not.

The Three-Clusters Requirement

To examine the importance of meeting each of the
symptom cluster criteria, ANOVAs with Fisher post hoc
tests were conducted to compare children meeting all three
DSM-IV PTSD symptom cluster criteria (PTSD-3; n =
14), children meeting criteria for all but one of the symp-
tom clusters (PTSD-2; n = 23), and children meeting cri-
teria for one of the symptom clusters (PTSD-1; n = 13)
on continuous variables. Nine children did not meet cri-
teria for any one cluster and thus were not included in
these analyses. Likelihood χ2 tests were used to compare
the groups on categorical variables.

Results of analyses comparing the groups on demo-
graphics, comorbidity, and pubertal status are presented
in Table 2. The groups did not differ with regard to gen-
der, ethnicity, age, Tanner stage, or comorbid diagnoses
(see Table 2). To ensure that groups did not differ in the
number of traumatic experiences, time since the first
trauma, time since the most recent trauma, or types of

PEDIATRIC PTSD

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002 169

TABLE 1
Frequency and Intensity of DSM-IV PTSD Symptoms and Their Relation to PTSD Diagnosis and Overall Impairment

Frequency Intensity PTSD Diagnosis Overall Impairment

DSM IV Criteria No. (%) Mean (SD) R2: F R2: I R2: F R2: I

B) Reexperience
1. Distressing recollections 41 (69.5)2 1.98 (1.3)11 0.128** 0.065 0.230*** 0.260***
2. Distressing dreams 38 (64.4)3 2.37 (0.9)2 0.79* 0.111* 0.203** 0.281***
3. Feeling of reoccurrence 24 (40.7)10 1.67 (1.1)15 0.011 0.138** 0.144** 0.160**
4. Distress at exposure to cues 34 (57.6)5 2.26 (1.0)5 0.008 0.041 0.044 0.084*
5. Physiological reactivity to cues 19 (32.2)12 1.95 (1.1)12 0.017 0.059 0.014 0.017

C) Avoidance and numbing
1. Avoid thoughts, feelings, & conversations 49 (83.1)1 2.18 (1.2)6 0.004 0.065 0.017 0.303***
2. Avoid places & people 35 (59.3)4 1.86 (1.1)13 0.016 0.081* 0.014 0.044
3. Inability to recall important aspects of event 41 (69.5)2 2.00 (1.1)10 0.100* 0.111* 0.194** 0.230***
4. Diminished interest 22 (37.3)11 1.68 (0.9)14 0.064 0.060 0.117* 0.185**
5. Detachment from others 30 (50.8)6 2.30 (1.0)3 0.346*** 0.215*** 0.123** 0.109*
6. Restricted range of affect 30 (50.8)6 2.07 (0.9)8 0.054 0.010 0.109* 0.032
7. Foreshortened future 18 (30.5)13 2.28 (1.0)4 0.028 0.024 0.078* 0.044

D) Hyperarousal
1. Sleep problems 35 (59.3)4 1.86 (1.0)13 0.075* 0.087* 0.160** 0.058
2. Irritability/anger 26 (44.1)9 2.73 (1.0)1 0.004 0.012 0.026 0.006
3. Difficulty concentrating 38 (64.4)3 2.11 (0.9)7 0.034 0.111* 0.096* 0.102*
4. Hypervigilance 29 (49.2)7 2.03 (0.9)9 0.380*** 0.270*** 0.102* 0.078*
5. Exaggerated startle 28 (46.6)8 2.07 (0.9)8 0.196** 0.256*** 0.008 0.017

Note: PTSD = posttraumatic stress disorder; F = frequency of symptoms; I = intensity of symptoms. Superscript numbers represent rank.
* p < .05; ** p < .01; *** p < .001.

CARRION ET AL.

170 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002

traumas experienced, the three groups were compared,
and results indicated that the groups did not differ on
any of these variables. Results from comparisons on impair-
ment and distress are presented in Table 3. As Table 3
demonstrates, the PTSD-2 and PTSD-3 groups did not
differ significantly on any of these variables, but both dif-
fered significantly from the PTSD-1 group in terms of
distress of symptoms, social impairment, school impair-

ment, overall impairment, and the percentage meeting
criterion F (i.e., clinically significant impairment). Ratings
on the PTSD Reaction Index and on developmental
impairment did not differ across the groups.

Developmental Issues in Cluster Aggregation

The most common symptom cluster was B, with 76.0%
of the sample meeting the DSM-IV criteria, followed by

TABLE 2
Comparison of Three PTSD Symptom Groups on Demographics and Comorbidity

Met Criteria for

One Symptom Two Symptom Three Symptom
Cluster Clusters Clusters

Measures (n = 14) (n = 23) (n = 13) F or χ2 p

Age (years) 10.29 (1.8) 10.55 (2.1) 10.74 (2.0) 0.77 .838
% Female 35.7 47.8 35.7 0.77 .682
Ethnicity 9.6 .289

% White 42.9 39.1 57.1
% African American 50.0 43.5 21.4
% Hispanic 0.0 13.0 14.3
% Asian 7.1 4.3 0.0
% Other 0.0 0.0 7.1

Duration of trauma
Time since most recent 20.07 (22.1) 31.70 (31.9) 38.00 (34.8) 1.23 .302
Time since first 58.79 (43.4) 48.08 (38.9) 64.62 (37.5) 0.79 .462

No. of traumatic experiences 2.07 (1.0) 1.83 (1.0) 2.07 (0.8) 0.42 .658
% Other anxiety disorder 41.7 56.5 50.0 0.71 .702
% Mood disorder 25.0 26.1 28.6 0.05 .977
% Disruptive disorder 14.3 26.1 14.3 1.11 .573
% Tanner stage > 2 30.8 52.2 35.7 1.89 .389
CBCL Internalizing 57.82 (13.0) 65.32 (11.6) 65.40 (9.8) 1.69 .199
CBCL Externalizing 59.09 (18.0) 69.63 (9.7) 70.10 (13.3) 2.59 .088

Note: Means with standard deviations in parentheses or percentage. PTSD = posttraumatic stress disorder; CBCL = Child Behavior Checklist.

TABLE 3
Comparison of the Three PTSD Symptom Groups in Symptom Distress and Impairment

Met Criteria for

One Symptom Two Symptom Three Symptom
Cluster Clusters Clusters

Measures (n = 14) (n = 23) (n = 13) F or χ2 p

Distress of symptomsa 1.54 (0.9) 2.39 (1.0) 2.57 (1.1) 4.30 .019
Social impairmenta 0.54 (0.8) 1.52 (1.1) 1.79 (1.1) 5.67 .006
School impairmenta 0.69 (0.9) 1.76 (1.3) 2.14 (1.2) 5.38 .008
Developmental impairment 0.38 (0.8) 0.91 (1.2) 1.29 (1.1) 2.36 .106
Overall impairmenta 3.15 (2.0) 6.64 (3.4) 7.79 (3.0) 9.06 <.001 Reaction Index score 26.14 (9.2) 35.59 (12.7) 33.86 (18.1) 2.16 .126 % Meeting criterion Fa 64.3 91.3 100.0 8.96 .011

Note: Means with standard deviations in parentheses or percentage. PTSD = posttraumatic stress disorder.
a Significant differences (p < .05) on Fisher post hoc test or paired χ2 with individuals meeting criteria for one symptom cluster significantly

less than individuals meeting for two or three. No significant differences were found between the group meeting criteria for two and the group
meeting criteria for three symptom clusters.

cluster C (51.0%) and then cluster D (46.0%). As noted,
we divided the sample between two pubertal groups (i.e.,
Tanner stage 2 and below versus above 2), and prelimi-
nary analyses indicated that the two groups did not dif-
fer in ethnicity, number of traumatic experiences, time
since the first trauma, time since the most recent trauma,
or types of traumas experienced. There were more girls
in the group in the later stages of puberty.

The group in earlier stages of pubertal development and
the older group did not differ with regard to the propor-
tion of children meeting all criteria for either cluster B, C,
or D. There was evidence, however, of increased aggrega-
tion of the clusters with pubertal development (i.e., there
was evidence that children who met one of the symptom
cluster criteria tended to meet additional symptom clus-
ter criteria in the older group). Specifically, the likelihood
ratio χ2 test statistic indicated significant relations between
cluster B and clusters C (χ2

1 = 3.83, p = .050) and D
(χ2

1 = 8.97, p = .003) in the older group (i.e., Tanner stages
3, 4, and 5). In contrast, in the younger group (i.e., Tanner
stages 1 and 2), the relation between cluster B and clusters
C (χ2

1 = 0.19, p = .657) and D (χ2
1 = 0.01, p = .930) were

not significant. Because there were more girls in our group
of children in the later stages of puberty, we next tested
whether gender was responsible for the developmental
findings. Girls and boys did not differ with regard to the
proportion meeting full criteria for either cluster B, C, or
D. In addition, there was no evidence of differential aggre-
gation of the clusters with gender (i.e., there was no evi-
dence that girls who met one of the symptom cluster
criteria were more likely to meet additional symptom clus-
ter criteria than boys).

DISCUSSION

This study contributes to the literature on the phe-
nomenology of pediatric PTSD in three areas. First, data
supported the utility of distinguishing between the fre-
quency and the intensity of symptoms. Second, findings
supported the hypothesis that children with subthresh-
old criteria did not differ significantly from children meet-
ing all three cluster criteria with regard to impairment
and distress. Third, there was more symptom cluster aggre-
gation in the later stages of puberty. In terms of the fre-
quency of the symptoms, there were commonalties in
symptom frequency in our sample and previous findings
with older adolescents (i.e., Cuffe et al., 1998).

Our data support the utility of distinguishing between
the frequency and the intensity of symptoms in the inves-

tigation of the phenomenology of pediatric PTSD. For
example, the intensity of some symptoms predicted PTSD
diagnosis or functional impairment independently of fre-
quency. For instance, as indicated in Table 1, the inten-
sity of the avoidance of feelings, thoughts, and conversations
(symptom C1) and distress at exposure to cues (B4) pre-
dicted functional impairment, whereas the frequency of
these symptoms was not predictive of impairment. Intensity
of the difficulty concentrating (D3), avoiding places and
people (C2), and feelings of recurrence (B3) predicted
PTSD diagnosis, whereas frequency of these symptoms
did not. One reason for this finding may be that these
particular symptoms are less intrinsically distressing and/or
simply more common than symptoms whose frequency
is predictive of impairment (e.g., restricted range of affect
[C6], sense of foreshortened future [C7]). Overall, our
findings point to the possibility that frequency and inten-
sity may both contribute to the phenomenology of pedi-
atric PTSD. Further understanding of their respective
contributions may help clarify why the same degree of
clinical impairment can be found in different clinical pre-
sentations.

Certain symptoms predicted DSM-IV PTSD diagno-
sis but not functional impairment. For example, exag-
gerated startle was highly associated with PTSD diagnosis
but did not correlate with impairment. This may indi-
cate that certain PTSD criteria may be describing traits
associated with vulnerability to the development of PTSD
in childhood but are not necessarily markers of disorder
impairment. Moreover, some symptoms, such as dimin-
ished interest, were highly associated with impairment
but not predictive of diagnosis. This may be because of
a lack of developmentally appropriate descriptions of
PTSD symptoms for children or alternatively, existing
comorbidity, such as depression.

Our findings supported the hypothesis that children
with subthreshold criteria did not differ significantly from
children meeting all three cluster criteria with regard to
impairment and distress. Supporting the statistical find-
ings, children fulfilling diagnostic requirement for two
clusters had elevated Internalizing T scores on the CBCL,
and these scores were very similar to the T scores of chil-
dren meeting criteria for three clusters. In addition, our
findings suggest that the impairment found on sub-
threshold (PTSD-2) children is not due to comorbidity
but rather is specific to the posttraumatic symptoms pre-
sent, because there were no significant differences among
the groups in terms of comorbidity or other demographic

PEDIATRIC PTSD

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002 171

variables. Taken together, our results suggest that rather
than seeking a threshold number of symptoms, a more
precise diagnosis of pediatric PTSD could be developed
by evaluating the intensity of symptoms and their rela-
tion to functional impairment.

Finally, our findings suggest that developmental mod-
ifications to symptom clusters C (Avoidance and Numbing)
and D (Hyperarousal) as done with B (Reexperience) may
be useful. As expected, there was evidence of increased
aggregation of the clusters with pubertal development
(i.e., there was evidence that children who meet the symp-
tom cluster B criteria tended to meet criteria for additional
symptom clusters in the developmentally older group but
not in the younger group). This may indicate that the cur-
rent diagnostic criteria may not be appropriate for chil-
dren. In other words, our results suggest that the absence
of this triad does not indicate a lack of posttraumatic stress
problems in children but may be due to developmental
differences in symptom expression. The significant seque-
lae of early trauma, as indicated by the levels of impair-
ment, distress, and comorbidity, should be recognized
early to provide appropriate interventions.

Limitations

Our findings are limited by the multiple group com-
parisons made, a relatively small sample size, and a cross-
sectional design. It will therefore be important to replicate
these findings in larger samples with additional informants
and longitudinal designs. Our failure to demonstrate age
differences in the prevalence of symptom clusters might
be a function of age range as well as sample size. Specifically,
because the age range was narrow, we compared only two
groups of Tanner stage children. Although we examined
the time since trauma, number of traumas, and types of
trauma, we did not specifically measure severity of trauma
or duration of trauma. Future research is thus needed to
examine the role of severity and duration of trauma in
pediatric PTSD symptom expression. Other characteris-
tics of the trauma, such as proximity and number of per-
petrators, may also have an impact on PTSD symptom
expression. Although previous investigations have shown
children to be as valid informants of internalizing symp-
toms, such as fear, as their parents (Weems et al., 1999),
researchers may wish to use additional informants to assess
externalizing symptoms in PTSD samples. The fact that
the K-SADS was given only to the children may have lim-
ited our evaluation of externalizing disorders; however,
the CBCL Externalizing subscale and inclusion in the

externalizing disorders group were significantly associated
in our sample.

Clinical Implications

Children with subthreshold PTSD should be evalu-
ated for functional impairment and distress and be given
appropriate recommendations for treatment even when
they do not fulfill DSM-IV criteria. Results suggest that
certain items were strongly associated with full diagno-
sis and/or impairment. Clinicians may wish to use these
items when screening children who have experienced
traumatic stress. Early recognition of the sequelae of
trauma in children may help prevent disturbance on the
acquisition of cognitive, social, and emotional milestones.
Trauma can interrupt, either temporarily or permanently,
the attainment of these skills, compromising the indi-
vidual’s abilities to learn, interact with others, and regu-
late mood (Perry, 1994; Pfefferbaum, 1997).

Behaviors such as hyperactivity, learning difficulties,
hypervigilance, and emotional dysfunction that begin
after the experience of a traumatic event or events sug-
gest neural mechanisms requiring study. Behavioral obser-
vations of children who experience trauma should guide
these investigations. Our results suggest that the behav-
ioral description of pediatric PTSD should be improved.
The data from this study may facilitate a better descrip-
tion of pediatric PTSD phenomenology.

REFERENCES

Aaron J, Zaglul H, Emery RE (1999), Posttraumatic stress in children fol-
lowing acute physical injury. J Pediatr Psychol 24:335–343

Achenbach TM (1991), Manual for the Child Behavior Checklist and Revised
Child Behavior Profile. Burlington: University of Vermont Department of
Psychiatry

Ackerman PT, Newton JEO, McPherson WB, Jones JG, Dykman RA (1998),
Prevalence of post traumatic stress disorder and other psychiatric diag-
noses in three groups of abused children (sexual, physical, and both). Child
Abuse Negl 22:759–774

American Academy of Child and Adolescent Psychiatry (1998), Practice para-
meters for the assessment and treatment of children and adolescents with
posttraumatic stress disorder. J Am Acad Child Adolesc Psychiatry 37:4S–26S

American Psychiatric Association (1994), Diagnostic and Statistical Manual of
Mental Disorders, 4th edition (DSM-IV). Washington, DC: American
Psychiatric Association

Berman SL, Kurtines WK, Silverman WK, Sarafini LT (1996), The impact of
exposure to crime and violence in urban youth. Am J Orthopsychiatry
66:329–336

Blake DD, Weathers FW, Nagy LM, Kaloupeck DG (1995), The develop-
ment of a clinician-administered PTSD scale. J Trauma Stress 8:75–90

Breslau N, Chilcoat HD, Kessler RC, Davis GC (1999a), Previous exposure
to trauma and PTSD effects of subsequent trauma: results from the Detroit
area survey of trauma. Am J Psychiatry 156:902–907

Breslau N, Chilcoat HD, Kessler RC, Peterson EL, Lucia VC (1999b),
Vulnerability to assaultive violence: further specification of the sex differ-
ence in post-traumatic stress disorder. Psychol Med 29:813–821

CARRION ET AL.

172 J . AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002

PEDIATRIC PTSD

J. AM. ACAD. CHILD ADOLESC. PSYCHIATRY, 41 :2 , FEBRUARY 2002 173

Breslau N, Kessler RC, Chilcoat HD, Schultz LR, Davis GC, Andreski P
(1998), Trauma and posttraumatic stress disorder in the community: the
1996 Detroit area survey of trauma. Arch Gen Psychiatry 55:626–632

Carrion VC, Steiner H (2000), Trauma and dissociation in delinquent ado-
lescents. J Am Acad Child Adolesc Psychiatry 39:353–359

Cohen J (1994), The earth is round (p < .05). Am Psychol 49:997–1003 Cuffe SP, Addy CL, Garrison CZ et al. (1998), Prevalence of PTSD in a com-

munity sample of older adolescents. J Am Acad Child Adolesc Psychiatry
37:147–154

Famularo R, Fenton T, Augustyn M, Zuckerman B (1996), Persistence of pedi-
atric posttraumatic stress after two years. Child Abuse Negl 20:1245–1248

Green BL (1985), Children traumatized by physical abuse. In: Posttraumatic
Stress Disorder in Children, Eth S, Pynoos RS, eds. Washington, DC:
American Psychiatric Press, pp 133–154

Hayward C, Killen JD, Kraemer HC, Taylor C (2000), Predictors of panic
attacks in adolescents. J Am Acad Child Adolesc Psychiatry 39:207–214

Jensen PS, Hinshaw SP, Kraemer HC et al. (2001), ADHD comorbidity find-
ings from the MTA study: comparing comorbid subgroups. J Am Acad
Child Adolesc Psychiatry 40:147–158

Kaufman J, Birmaher B, Brent D et al. (1997), Schedule for Affective Disorders
and Schizophrenia for School-Age Children-Present and Lifetime Version
(K-SADS-PL): initial reliability and validity data. J Am Acad Child Adolesc
Psychiatry 36:980–988

Marshal WA, Tanner JM (1970), Variations in the pattern of pubertal changes
in girls. Arch Dis Child 45:13–23

McLeer SV, Dixon JF, Henry D et al. (1998), Psychopathology in non–clini-
cally referred sexually abused children. J Am Acad Child Adolesc Psychiatry
37:1326–1333

Nader K, Pynoos RS, Fairbanks L, Frederick C (1990), Childhood PTSD reac-
tions one year after a sniper attack. Am J Psychiatry 147:1526–1530

Nader KO, Kriegler JA, Blake DD, Pynoos RS, Newman E, Weather FW
(1996), Clinician Administered PTSD Scale, Child and Adolescent Version.
White River Junction, VT: National Center for PTSD

Perry BD (1994), Neurobiological sequelae of childhood trauma: posttrau-
matic stress disorder in children. In: Catecholamine Function in Posttraumatic
Stress Disorder: Emerging Concepts, Murburg MM, ed. Washington, DC:
American Psychiatric Press, pp 233–255

Perry BD, Pollard RA, Blakley TL, Baker WL, Vigilante D (1995), Childhood
trauma, the neurobiology of adaptation, and “use-dependent” develop-
ment of the brain: how “states” become “traits.” Infant Ment Health J
16:271–291

Pfefferbaum B (1997), Posttraumatic stress disorder in children: a review of
the past 10 years. J Am Acad Child Adolesc Psychiatry 36:1503–1511

Psychological Corporation (1999), Wechsler Abbreviated Scale of Intelligence.
San Antonio, TX: Harcourt Brace

Pynoos RS, Frederick C, Nader K et al. (1987), Life threat and posttraumatic
stress in school-age children. Arch Gen Psychiatry 44:1057–1063

Scheeringa MS, Peebles CD, Cook CA, Zeanah CH (2001), Toward estab-
lishing procedural, criterion, and discriminant validity for PTSD in early
childhood. J Am Acad Child Adolesc Psychiatry 40:52–60

Scheeringa MS, Zeanah CH, Drell MJ, Larrieu JA (1995), Two approaches
to the diagnosis of posttraumatic stress disorder in infancy and early child-
hood. J Am Acad Child Adolesc Psychiatry 34:191–200

Terr LC (1983), Chowchilla revisited: the effects of psychic trauma four years
after a school bus kidnapping. Am J Psychiatry 140:1543–1550

Terr LC (1991), Childhood traumas: an outline and overview. Am J Psychiatry
148:10–20

Terr LC, Bloch DA, Michel BA, Shi H, Reinhardt JA, Metayer S (1999),
Children’s symptoms in the wake of Challenger: a field study of distant-
traumatic effects and an outline of related conditions. Am J Psychiatry
156:1536–1544

Weems CF, Silverman WK, La Greca AM (2000), What do youth referred for
anxiety problems worry about? Worry and its relation to anxiety and anx-
iety disorders in children and adolescents. J Abnorm Child Psychol 28:63–72

Weems CF, Silverman WK, Saavedra LM, Pina AA, White-Lunkin (1999),
Discrimination of children’s phobias using the Revised Fear Survey Schedule
for Children. J Child Psychol Psychiatry 40:941–952

• Toward an Empirical Definition of Pediatric PTSD: The Phenomenology of PTSD Symptoms in Youth

METHOD

Participants

Instruments

Hypotheses and Data Analyses

RESULTS

Frequency, Intensity, and Impairment

Developmental Issues in Cluster Aggregation

DISCUSSION

Limitations

Clinical Implications

REFERENCES