Posttraumatic Stress in Children Exposed to Family

Violence and Single-Event Trauma

LAURA ANN McCLOSKEY, PH.D., AND MARLA WALKER, M.A.

ABSTRACT
Objective: To examine posttraumatic stress disorder (PTSD) and other comorbid forms of psychopathology in a sample of
children exposed to chronic abuse and single-event trauma. Method: School-age children ( N = 337) were assessed for expo-
sure to traumatic events (family violence, violent crime, death or illness of someone close to child, accidents) and posttrau-
matic stress symptoms. Children and mothers received structured diagnostic interviews to assess child psychopathology.
Results: Children from violent households were no more likely to report an extrafamilial traumatic stressor than children from
nonviolent homes. Among the children reporting a traumatic event, 24.6% met the diagnostic criteria for PTSD. The leading
precipitating event for PTSD symptoms was death or illness of someone close to the child (log odds = 4.3). Family violence,
violent crime, but not accidents also resulted in PTSD. Children with PTSD displayed comorbidity across different symptom
classes, most notably phobias and separation anxiety. Conclusions: Both type I and type II trauma can result in PTSD in
about one quarter of children. Children with posttraumatic stress symptoms had many other forms of comorbid psychopathol-
ogy, indicating a global and diffuse impact of trauma on children. J. Am. Acad. CbildAdolesc.Psycbiafrl: 2000, 39(1):108-115.
Key Words: posttraumatic stress, family violence, child psychopathology, trauma.

Posttraumatic stress disorder (PTSD) was formally sequelae. Studies of children resettled in the United States
acknowledged and classified in 1980 (American Psychi- who fled political violence have shown that even across
atric Association, 1980). The disorder is unique because cultures and experiences, about half of the young survivors
a key feature of classification is anchored to the external meet the diagnostic criteria for PTSD (McCloskey and
world: the person must have been through a potentially Southwick, 1996). In the United States, researchers need
life-threatening event or one that threatened his or her not restrict their studies to war refugees to locate children
basic physical integrity. There are 3 putative symptom cat- exposed to violence. A large proportion of American urban
egories: (1) intrusive feelings of reexperiencing the trauma, children experience community and criminal violence and
(2) avoidance of event-related stimuli, and (3) increased have concomitant symptoms of distress (Martinez and
arousal and attentional problems. Richters, 1993).
According to epidemiologicalstudies, PTSD diets 3% While witnessing community violence has deleterious
to 4% of U.S. adults (Heltzer et al., 1987), with higher effects on children, violence within the family is also a
lifetime prevalence rates (Kessler et al., 1995).Although no source of trauma. Children who are physically or sexually
national prevalence statistics are available for children, they abused, for instance, display elevated traumatic stress
also face numerous threats to personal safety worldwide, symptoms (Kiser et al., 1991). Children witnessing mari-
and various studies have reported PTSD as among the tal violence in the home also show elevated symptoms
(McCloskey et al., 1995). According to one study, chil-
dren of battered women met the criteria for PTSD in
Accepted Jub 27, 1999. 13% of the cases (Graham-Berman and Levendosky,
Dr. McCloskq is with the Harvard lJniue?sitySchool of Public Health, Boston.
This research was supported by grantsfiom the National Center on Child Abuse
1998). Few studies have compared the impact of domes-
and Neglect and the NIMH. The authors thank Kel4 Stanmorefor her assistance tic violence to other stressors.
in coding these data andlefiq Stuewigfor statistical help. Accidents and natural disasters elicit posttraumatic
Reprint requests to Dr. McCloskq, Department of Maternal and Child Health, stress symptoms as well. Studies of children’s responses to
Haruard School of Public Health, 677 Huntington Avenue, Boston, M A 02115.
0890-8567/00/3901-010802000 by the American Academy of Child earthquakes in California (Nolen-Hoeksema and Morrow,
and Adolescent Psychiatry. 1991) or to hurricanes in Florida (La Greca et al., 1996)

108 J . A M . A C A D . C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2 0 0 0

 POSTTRAUMATIC STRESS

document PTSD as among the sequelae. Less information functional families? The research on this topic has been
exists about the impact of more common and widespread confined to adult samples, most of whom are war veterans,
accidents such as automobile wrecks or drownings. and the findings remain inconclusive. In a recent series of
clinical studies, adults exposed to non-war-related trauma
Features of the Event That Enhance the Traumatic Impact
were more resilient if they remembered a close and stable
Various event features might strengthen the likelihood early home life in contrast to those who remembered
that PTSD will result, including whether the trauma is severe neglect (Van der Kolk, 1996). Attachment relations
based on a single or multiple incidents; whether the event and secure home lives might protect children from the
involves interpersonal violence or loss, or an accident; devastating effects of extrafamilial trauma (Breslau and
and whether the person is a target or witness of the event. Davis, 1992). Conversely, a history of psychopathology,
O n the basis of clinical observations, Lenore Terr family violence, childhood victimization, or child behavior
(1991) distinguished between single event-based (type I) problems might predispose a person to develop the mag-
and ongoing trauma (type 11) in childhood. According to nified composite of PTSD symptoms when faced with
Terr, the psychological response to a single event (type I new extreme stressors. Obviously it is important to exam-
trauma) results in the PTSD. She distinguishes type I1 dis- ine the role of family factors in children’s responses to trau-
order, on the other hand, by coping mechanisms (e.g., matic events to illuminate this “vulnerability hypothesis.”
massive denial, numbing) which develop into enduring
characterological and mental health problems. Terr con- Comorbidity in Posttraumatic Stress Disorder
tends that children enduring ongoing abuse have more
There seems to be a range of psychological maladies
problems in general and that the classic symptoms of
following trauma, aside from or in addition to PTSD,
PTSD are less characteristic of this group. There has been
including adjustment and panic disorders, major depres-
little empirical support offered for the distinction between
sion, and closed head injury among adults (Lyons, 1987).
single- and multiple-event trauma. A recent comparison of
Some authors argue against the unique outcome of PTSD
the effects on children of a dog attack (type I) and domes-
as a result of acute stress and point out that trauma precip-
tic violence (type II), however, revealed similar PTSD pro-
itates a wide range of different symptomatic outcomes
files (Rossman et d., 1997).
(e.g., Keane and Wolfe, 1990).
Another aspect of the event that might exacerbate the
traumatic impact is whether the trauma originated from Goals of This Study
intentionally inflicted personal injuries, as in physical
attacks, or unmotivated impersonal injuries, as in natural There are a number of shortcomings in the psycholog-
disasters or car accidents. Despite the ill effects of disasters, ical research on PTSD and childhood trauma. With the
interpersonal traumas might lead to a more acute response exception of studies in psychiatric epidemiology, the
because they shake the foundations of human trust and research participants are often patients presenting to psy-
charity that could cast a long shadow on social adjustment. chiatric venues. Some researchers neglect to recruit con-
Whether one witnesses or is a direct target of a terrifying trol groups. Because research has been limited to small
event might be relevant in PTSD etiology. One group of and unique samples, the complex array of questions sur-
researchers found that children who had been both targets rounding the etiology of PTSD in school-age children
and witnesses of adult aggression and anger in the home remains unanswered. For instance, do children respond
reacted with heightened fear and sensitivity to simulated differently to different types of traumatic events, from
scenes of interadult anger (Hennessy et al., 1994), in ’ con- accidents to interpersonal violence? How do the event fea-
trast to children who were targets only, or control group tures influence children (e.g., witness versus target, inter-
children. These authors propose a “sensitization hypothe- personal and intentional harm versus impersonal, etc.)?
sis,” predicting heightened sensitivity to conflict or stress of Does a family history of violence pose a risk for PTSD
children who have been repeatedly exposed to anger and after encounters with new sources of trauma?What is the
family discord. pattern of comorbid symptoms with PTSD in children?
The goal of this study is to address each of these ques-
Vulnerability to Posttraumatic Stress Disorder
tions systematically with a large sample of children
Are victims from abusive backgrounds more vulnerable exposed to a panoply of risks, including family violence.
to PTSD after a trauma than victims from secure and We document the expression of posttraumatic stress

J . AM. ACAD. C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2 0 0 0 109

McCLOSKEY AND WALKER

symptoms in children who have been exposed to an array Measures

of hazards both within and outside of the home, as either Coding ofStressor Reponses. The first criteria to be met when assess-
witness, target, or both. We expand on former studies by ing PTSD is that a precipitating traumatic, usually violent and threat-
including a matrix of stressor events across dimensions of ening event must have occurred. The children were asked the
(I) personal versus impersonal, (2) target versus witness, following question: “Some children have really bad things happen to
them, or they have seen other people getting hurt. Have you ever been
and (3) a history of abuse in the family, from witnessing in a situation where you were really, really scared that you or someone
the battering of their mothers to directly experiencing else would get seriously hurt or die?” Children provided brief narra-
physical punishment and abuse. In addition, we assess tives describing events that were summarized in writing by the inter-
child psychopathology with multiple instruments, from viewers as well as tape-recorded. It should be noted that with only a
few exceptions, children provided only a single traumatic episode in
both the mothers’ and children’s perspectives, to obtain response to this question. This does not preclude the likelihood that
psychological profiles beyond only PTSD. Our aim is to children were exposed to multiple sources of trauma, but questioning
widen the lens by which to interpret posttraumatic stress was restricted only to the most recent of those events. All events
and psychopathology in children exposed to family vio- described occurred within the previous several months.
The traumatic events described by the children were coded and
lence and other stressors and to offer information that divided into one of the following categories: (1) domestic violence
could be valuable in developing a theory of the roots of against mother or child, (2) violent crimes, (3) accidents, (4) death or
psychological trauma in children. illness. These events were then coded for whether the child was a wit-
ness of the event, a target, or both. Coders agreed on the events in 90%
of the instances.
METHOD Domestic Kooknce Episode. Children described a single unique, salient,
and often terrifying scene of domestic violence. Children might have
Participants been chronically exposed to domestic violence as well, but they
Participants were 337 children (51% female), drawn from a total responded to the traumatic stress question with a description of an
sample of 363, between the ages of 6 and 12 years and their mothers, extremely traumatic episode.
interviewed in a mid-size Southwestern city during 1990-1991. The Kooknt Crime. Children reported exposure to a violent crime. These
children’s mean age was 9.3 years (SD = 1.9). Because the study focused crimes included mugging, rape and sexual abuse, homicide, threatened
on the impact of domestic violence on children, mothers who had expe- homicide, and assault.
rienced relationship violence were oversampled from shelters and com- Accidents. Our third category of traumatic events covered accidents.
munity advertisements (n = 170). Our comparison group (n = 167) Car accidents or drownings and near-drownings made up nearly all of
was recruited from posters at the same addresses as our index (battered) the events in this category.
group, except for shelters. Most of the children exposed to marital vio- Death or Illness. A proportion of children also had encounters with
lence had been exposed for a long period of their lives, predating any death or serious illness among family members or close friends.
further traumatic events they recounted for this study. Family History of W$e or Paternal Child Abuse. Items from the
Although most of the families in the study were low-income, one- Conflict Tactics Scale (Straus, 1979) and additional items were adrnin-
way analyses of variance revealed that family income in the comparison istered to mothers and the target child. Responses were on a scale of 0
group was significantlyhigher than in either the shelter or community
battered groups = 13.7,= ~ .0001). In addition, women from the
comparison group had on average 1 year of college in contrast to the
index group, who had slightly less than 12 years on average, indicating a TABLE 1
lower high school graduation rate among the battered women. Table 1 Sample Demographics
presents the demographics for the families.

Procedure Child’s age (yr) 9.3 (1.9)

No. of children in home 2.9 (1.3)
During the initial telephone intake, a child was selected for partic-
Mother education (yr) 12.45 (2.7)
ipation by project staff, alternating the sex of the child chosen. This
Family income ($) 1,460 (957)
rule eliminated parental choice of the child to be interviewed. When
Times moved past year 0.95 (1.4)
2 siblings of the same sex were available, the one closest to the age of
8 was selected. Proportion
Mothers provided informed consent and children provided signed, Partner employed 75.4
informed “assent” to participate. Children who displayed signs of Mother employed 40.8
trauma or were ongoing victims of abuse were seen by our in-house Mother married 56.5
counselor and were referred to the appropriate clinical or government Ethnic group membership
agencies with the permission and cooperation of their mothers. Hispanic 34.4
Interviewers were blind to whether children were from violent Anglo 53.7
homes or not (except in the obvious case of shelters) until the last OtheP 11.8
questions in the interview that asked about family violence. Psycho-
pathology, therefore, was assessed first, then posttraumatic stress a African-American (5.6%), Native American (4.4%), and Asian
symptoms, and finally exposure to abuse. or Pacific Islander (1.8%).

110 J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2 0 0 0

 POSTTRAUMATIC STRESS

(never) to 6 (20 times or more). Items about the partner‘s abuse of the RESULTS
mother included (I) pushed or grabbed; (2) slapped; (3) kicked, bit, or
hit; (4) hit or tried to hit her with something; ( 5 ) beat her for several Exposure to Potentially Traumatic Events
minutes; (6) choked or tried to kill her; (7) threatened her with a knife
or gun. The final item asked mothers about injuries incurred, or asked Sixty percent of the children described a stressor inci-
children, “How often does your mother get hurt when your parents dent in response to the question asking when they had felt
fight and argue?”If the child or mother reported at least 2 of 8 wife “really, really scared that you or someone else would get
abuse items or 2 of 3 child abuse items, the child was coded as coming
from an abusive home. This categorical abuse variable, therefore, rep-
seriously hurt or die?” Only a few children ( 5 ) described
resented the presence of either wife or child abuse in the home accord- more than one episode, and the most recent was selected.
ing to either the mother or the child. Reports of mothers and children Of 205 responses, 40% recounted an episode of domestic
on the presence of abuse were correlated ( r = 0 . 4 1 , ~= .001).
violence, followed by accidents (36%), violent crime
Mothers and children were both asked about the fathers’ escalated
abuse toward the target child or other siblings in the home. Items (l6%), and death or serious illness of a family member or
about escalated paternal abuse directed toward the children were (1) close friend (8%). Of the accidents, 61% were automo-
“kicked or hit you or your sibs with a fist”; (2) “hit you or sibs with an bile accidents and 39% were drowning. Many more chil-
object, like a belt, hairbrush, electrical cord, etc.”; and (3) “burned you
or sibs with something hot, like an iron, cigarette, or really hot water.” dren were witnesses to these events than were targets. Of
Posttraumatic Stress Symptom Clusters. To measure posttraumatic the children reporting any traumatic event, 87% were
stress symptoms, an original 12-item inventory derived from the witnesses only, 10% were targets only, and 3% were both.
criteria established in the DSM-111-R (American Psychiatric Asso-
ciation, 1987) was administered to the child. Our brief inventory Exposure to Wife and Child Abuse
covered all 4 categories of symptoms, but not all possible symptoms
classified under PTSD. The inventory also had never been used or Of our sample of 337 children, 49% saw their fathers
standardized on other populations, and to preserve quality of mea-
physically attack their mothers and 12% reported receiving
surement we also relied on a standardized psychiatric diagnostic
interview schedule, the Child Assessment Schedule, to supplement escalated physical abuse from their father in the past year.
our assessment of PTSD (Hodges and Saunders, 1990). Items from When both the mother and child reports were combined
the Child Assessment Schedule relevant to the classification of PTSD across both forms of wife and child abuse, a total of 191
included those from the following clinical categories: depression,
attention deficit disorder with hyperactivity, and overanxious dis- children, or 54% of the entire sample, were categorized as
order. This method of supplementing our brief inventory has the exposed to at least one of these forms of domestic abuse.
benefit of repeated questioning about symptoms throughout the
interview, particularly important when relying on the responses of Posttraumatic Stress Symptoms
children. Our approach relies exclusively on the children’s reports of
their own posttraumatic symptoms, although the mother was also Posttraumatic stress was assessed exclusively by inter-
interviewed about the child’s mental health. Some information was viewing the child. Although about two thirds of the entire
missing from our assessment, most critically the duration of symp-
toms, and we therefore might classify some children as meeting the sample of children reported at least 2 or 3 symptoms con-
criteria when perhaps they would fall into the acute stress category sistent with PTSD, only 15% ( n = 52) met the cutoff.
instead. Children who met the criteria for PTSD scored an average
Child P%ychopathology. Mothers and children were administered
parent and child versions of the Child Assessment Schedule, which is
of 15.1 (SD = 5.15) of a possible 31 assessment items. The
a standardized, 1-hour psychiatric diagnostic instrument to assess range of scores for these children was 7 to 29. Children
children’s mental health. The authors of this instrument report high who did not meet criteria because their symptoms were
convergent validity and internal reliability (Hodges and Saunders, not distributed across the putative categories (e.g., numb-
1990). We computed interrater (or interviewer) reliabilities for 5
training videotapes provided through Hodges’ laboratory. Across ing, reexperiencing, etc.) had an average of 4.2 (SD = 3.4)
diagnostic categories, and controlling for the variance accounted for symptoms, with a lower overall range of 0 to 17. The sub-
by subscales and tapes, for our 13 interviewers we obtained a correla- sample of children who met the symptom criteria for
tion coefficient of 0.89.
Mothers also completed the Child Behavior Checklist (Achenbach PTSD was similar on important demographic character-
and Edelbrock, 1983), a standardized instrument to assess symptoms istics (i.e., age, gender) to the rest of the children.
of psychopathology or emotional disturbances. Overall 24.6% of the children reporting any stressor
Children? Self-ReportedFeelings of Fear When Parents Argue. We
event met the symptom criteria for PTSD. We antic-
obtained information about the children’s emotional responses to
scenes of interparental conflict or violence that they might witness, in ipated that different types of stressors would be more
particular their fear response. After asking about forms of marital vio- likely to engender PTSD. There were, however, few dif-
lence in the home, children were asked, “How scared do you get when ferences between the type of exposure and the devel-
your parents argue or fight?”Their responses were on a 4-point scale:
(1) “not very scared,” (2) “a little bit scared,” (3) “pretty scared,” (4) opment of symptoms, although accidents were the least
“reallyvery scared.” likely catalysts.

J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2000 111

McCLOSKEY A N D WALKER

Our next analyses address whether there are differences TABLE 2

between witnessing an event or being a target of the event. Logistic Regressions of Traumatic Stressors Predicting
Posttraumatic Stress
Across our traumatic event analyses, those who were solely
Predictor Events Parameter Estimate (B) Odds Ratio
witnesses developed PTSD symptom profiles 21% of the
time and those who were targets met criteria 38% of the Domestic abusea*+ 1.11 3.03
time. If the child was both a witness and a target, he or she Violent crimes’; 0.939 2.55
Accidents 0.185 1.20
met the PTSD symptom criteria 100% of the time. In Death or illness++ 1.45 4.27
other words, being both a witness and a target of the same
event seemed to carry the most serious and elevated risk for Note: Final model x 2 4 = 2 1 . 0 7 , ~= .0003.
a Domestic abuse refers to partner-to-mother and partner-to-
developing PTSD. Upon further inspection of the chil- child abuse.
dren who were coded as both targets and witnesses, 5 of * p < .05; “ “ p< .01.
the 6 incidents entailed interpersonal violence in the
home. Since there were so fay children (3%) who reported failed to yield any further variance to our overall model.
being both witness and target, and these were dispropor- None of the regression equations testing for the impact of
tionately children from violent home backgrounds, these interactions on PTSD was statistically significant, and
must be considered only tentative and descriptive findings. therefore they are not represented in Table 2. In summary,
Family Violence and Posttraumatic Stress Disorder our model provides evidence for the link between domes-
tic abuse and PTSD, but little direct support for the
There was a relationship between children’sprior expe- notion of an elevated risk for PTSD in the aftermath of
riences with abuse in the family and posttraumatic symp- further traumatic stressors. One problem with our anal-
toms, regardless of the stressor identified by the child in yses, however, is the ceiling effect of domestic abuse
the PTSD inventory. Among those 52 children classified among children afflicted with PTSD. With so little range,
with PTSD, 83% were categorized as coming from a vio- it is difficult to discern compounded effects. It is neverthe-
lent home in contrast to 55% of the children without less clear that children from type I1 backgrounds develop
PTSD (x21= 1 6 . 9 , <~ .OOOl). PTSD symptom profiles.
We conducted a series of logistic regressions to test the
influence of a background of domestic abuse and 3 of the Children’s Reports of Fear and Domestic Abuse
coded traumatic events (violent crime, death or illness, In response to questions about how scared they were
accidents) on the expression of PTSD (Table 2). The co- when parents argued, 78% of children with PTSD from
variates were entered simultaneously, and each covariate violent homes endorsed the highest rating for fear dur-
was entered controlling for all others in the equation. The ing domestic quarrels: “really very scared.”
strongest predictor of PTSD was the death or illness of a Although exposure to domesticviolence was a significant
family member or friend, with these children 4.3 times risk factor for heightened PTSD symptoms (x’~= 3 . 8 9 , ~ =
more likely to develop PTSD than others. An abusive .048), the impact of direct victimization by the father was
home background and exposure to crime also predicted even more pronounced (x21
= 12.53,~ = .OOO). While 19%
PTSD. of the children who witnessed the abuse of their mothers
After testing for the main effects of these different stres- had PTSD, 34% of children who were targets of their
sors, we tested the interactions of domestic abuse with any fathers’ or stepfithers’physical escalated abuse had the syn-
of the other stressor events to determine whether the pres- drome. This finding is especially striking in light of the fact
ence of domestic abuse combined with a reported trau- that only 12% of the children were apparently targeted by
matic event made the child even more vulnerable to their fathers for escalated abuse. Being a target of abuse,
developing PTSD. This is an important test of the vulner- therefore, appears to be even more deleterious than witness-
ability hypothesis because it is linked to the notion that ing violence against other family members.
prior family dysfunction and abuse would render a child
more sensitive to the impact of subsequent traumatic Comorbidity With Other Forms of Psychopathology
events than children without such a history. The findings To explore further the effects of domestic violence on
did not support such a conclusion, however. The combi- children’s mental health, we looked at the psychological
nation of domestic abuse with any of the 3 stressor events symptoms according to maternal and child reports on the

112 J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2000

 POSTTRAUMATIC STRESS

Child Assessment Schedule, and mothers’ reports only on PTSD, followed by ongoing domestic abuse. Violent
the Child Behavior Checklist. When correlations were crime was a significant risk, but accidents failed to predict
computed between PTSD (meeting criteria only) and the PTSD. The finding that death or loss precipitated PTSD
number of symptoms in various other diagnostic cat- symptoms is in contrast to one epidemiological study of
egories from the children’s reports in the Child Assessment adults responding to personal loss (grief) or violence
Schedule, we found remarkably high correlations, mostly (Breslau et al., 1998). Adults responding to this survey
greater than 0.30 (Table 3). Children meeting criteria on were more immediately traumatized by violence than by
the PTSD classification were most likely to show high loss, although loss was more widely experienced. O n the
levels of symptoms characteristic of phobias ( r = 0.41), other hand, our finding is consistent with claims made by
with high rates of separation anxiety ( r = 0.37) and oppo- l‘ynoos et al. (1996) that grief and loss are characteristically
sitional disorder ( r = 0.36). associated with the symptoms of PTSD. It is certainly pos-
Correlations between child-reported PTSD symptoms sible that children are more extremely affected by threat-
and other mother-reported symptoms of psychopathol- ened loss than violence, in part because of the high
ogy were highest among those classes of internalizing dis- dependency needs of childhood. Clearly, hrther research
orders one might expect to resemble the symptoms of needs to be conducted with children to clarify the distinc-
PTSD, including overanxious disorder ( r = 0.14), depres- tions between the effects of traumatic interpersonal vio-
sion ( r = O.ZO), and separation anxiety ( r = 0.13). These lence and nonviolent threatened loss of a loved one.
correlations indicate that children with PTSD have O n one hand, the finding that ongoing exposure to
heightened symptomatology in other areas, but not nec- family violence results in posttraumatic stress symptoms
essarily in all, as reported by the children. seems to challenge Terr’s (1991) dichotomy of family-
based and single-event crises. The high comorbidity of
PTSD with other childhood disorders, however, is con-
DISCUSSION
sistent with her argument that ongoing stressors impose
Our results indicate that background family risk more enduring psychological costs on children, resulting
factors-in the present case, wife and child abuse- in high levels of general psychopathology. Our study
engender posttraumatic stress symptoms in a significant does not disentangle this problem completely because
proportion of children. However, death or illness of some- most of the children who developed PTSD at all had
one close to the child posed the strongest risk factor for been exposed to domestic abuse.
PTSD symptoms are also more likely to follow an inter-
TABLE 3 personal event or series of events than to follow imper-
Correlations of Posttraumatic Stress With Symptoms in Other sonal, unmotivated disasters such as car accidents or
Diagnostic Categories as Reported by the Mother or the Child near-drownings. Children are less affected when they wit-
Form of Psychopathology‘ Child Mother ness evidence of mortality or near-mortality of strangers,
Child Assessment Schedule as in car accidents. In addition, we found the children who
0bsessive-compulsive 0.23** 0.07 were victims of parental abuse were almost twice as likely
Phobia 0.41** 0.10 to show signs of PTSD as children who only witnessed the
Separation anxiety 0.37** 0.13*
Oppositional 0.36** 0.12*
abuse of their mothers.
Conduct disorder 0.19** 0.04 One question addressed in this study is whether a his-
Depression - 0.20** tory of witnessing interparental violence or experiencing
Overanxious disorder - 0.14*
child abuse placed children more at risk for PTSD if they
Attention deficit disorder - 0.09
Child Behavior Checklist also reported other potentially traumatic events, such as
Total - 0.16** loss of a family member or violent crime. According to
Internalizing - 0.16** this vulnerability hypothesis, or the related sensitization
Externalizing - 0.14**
hypothesis advanced by Hennessy et al. (1994), children
Correlations between posttraumatic stress disorder (PTSD) and with a history of abuse should be more susceptible to
the child’s report of symptoms of depression, overanxious disorder,
developing PTSD than children without such a history.
and attention deficit disorder were excluded from the table because
items from these categories were used to classify PTSD. There were, however, no significant statistical inter-
* p < .05; * * p < .01. actions between abuse in the home and any of the 3 addi-

J. AM. ACAD. CH I LD ADOLESC. PSYCHIATRY, 39:1, JANUARY 2 0 0 0 113

McCLOSKEY A N D WALKER

tional stressors (death, violent crime, accidents). In other The high comorbidity for children with PTSD indi-
words, a history of abuse did predict PTSD indepen- cates the need for a broad clinical approach to traumatized
dently, but it did not fuel PTSD symptoms in the after- children. It is insufficient to diagnose PTSD and end the
math of these other adverse events. This finding was assessment. Rather, a diagnosis of PTSD would require
surprising in light of other research demonstrating further assessment and customized treatment for what will
heightened vulnerability in adults (Breslau et al., 1998). probably surface as a range of disorders, with potential
It is likely that there was a ceiling effect in abuse history candidates for both somatic and behavioral treatment. The
within this sample. In a sample with abuse more nor- challenge for clinicians is to effectively treat children show-
mally distributed, the interaction between abuse history ing symptoms that are orthogonal, as in both numbing and
and crisis event might reach statistical significance. hyperarousal. The common distinction in child psychopa-
It is also possible that children with PTSD profiles were thology of internalizing and externalizing disorders
experiencing ongoing abuse that was so severe and debil- (Achenbach and Edelbrock, 1983) is of little service in
itating that witnessing other single incidents was unlikely assessing and treating traumatized children; they are likely
to add to their symptom profile. By the same token, chil- to show comorbid symptoms from both categories.
dren without traumatic family histories appear resilient Children in this study with PTSD were especially likely
psychologically. Finally, the assessment of traumatic events to show elevated symptoms of phobias, separation anx-
outside the family was circumscribed in our protocol. iety, and oppositional disorder by their own reports, and
Instead of asking for an exhaustive list of life-threatening depression by their mothers’ reports. Given this profile, it
events, we asked about only one that was the most severe. is likely they will be impaired in various aspects of psycho-
It is possible that children who reported one were exposed social functioning, and especially in those domains of
to multiple events, which would account for the increase childhood activity that are so important for growth: aca-
in symptoms. demic skills and friendships. It is crucial to extend services
Children’s self-reported PTSD symptoms correlated to children of battered women to thwart the potentially
with other symptoms of psychopathology. In fact, if chil- serious consequences of exposure to domestic abuse.
dren met the criteria for PTSD, they had extensive symp-
REFERENCES
toms in virtually all other domains assessed by their own
Achenbach TM, Edelbrock C (1983), Manualfor the Child Behavior Checklist
reports. O n the other hand, the mothers’ reports of child and Revised Child Behavior Profile. Burlington: University of Vermont
psychopathology did distinguish those children with (self- Department of Psychiatry
American Psychiatric Association (1980), Diagnostic and StatisticalManual of
reported) PTSD and a more specific array of symptoms. Mental Disorders, 3rd edition (DSM-III). Washington, DC: American
Mothers’ reports of their children’s symptoms in the areas Psychiatric Association
American Psychiatric Association (1987), Diagnostic and Statistical Manual of
of depression, separation anxiety, and general anxiety dis- Mental Disorders, 3rd edition-revised (DSM-III-R). Washington, DC:
orders are correlated to the children’s self-reported PTSD American Psychiatric Association
Breslau N, Davis GC (1992), Posttraumatic stress disorder in an urban popula-
profile, but other categories of mental health problems tion of young adults: risk factors for chronicity.Am JPychiatry 149:671-675
(e.g., conduct or oppositional disorders)were unrelated to Breslau N, Kessler RC, Chilcoat HD, Schultz LR, Davis GC, Andreski P
(1998), Trauma and posttraumatic stress disorder in the community: the
the child’s self-report of posttraumatic stress. 1996 Detroit area survey of trauma. Arch Gen Psychiatty 55:626-632
Graham-Berman S , Levendosky A (1998),Traumatic stress symptoms in chil-
Clinical Implications dren of battered women.]Fam Violence 13:lll-128
Heltzer JE, Robins LN, McEvoy L (1987), Posttraumatic stress disorder in the
In summary, a background of abuse is a risk factor for general population: findings of the Epidemiologic Catchment Area sur-
vey NEnglJMed 317:1630-1634
posttraumatic stress and other psychological disorders in Hennessy KD, Rabideau GJ, Cicchetti D, Cummings EM (1994), Responses
childhood. It is unclear whether additional traumatic of physically abused and nonabused children to different forms of inter-
adult anger. ChildDev 66315-828
events exert a profound additive psychological effect on Hodges K, SaundersW (1990), Internal consistency of a diagnostic interview
children already traumatized by events at home, at least in for children: the Child Assessment Schedule. JAbnorm Child Psycho1
17:691-701
the relative short term. Children who are exposed to vio- Keane TM, Wolfe J (1990), Comorbidity in post-traumaticstress disorder: an
lent events outside of their family appear fairly resilient; analysis of community and clinical studies.JApp Soc Pychol20:1776-1788
Kessler R, SonnegaA, Bromet E, Hughes M (1995),Posrrraumaticstress hsorder
relatively few develop PTSD in the absence of family vio- in the National ComorbiditySurvey.Arch Gen Pychiatry 5 2 1048-1060
Kiser LJ, Heston J, Millsap PA, Pruitt DB (1991), Physical and sexual abuse
lence, a finding consistent with studies of adults (Breslau in childhood: relationship with post-traumatic stress disorder. ] A m Acad
and Davis, 1992). Child Adolesc Psychiatry 30:776-783

114 J . AM. A C A D . C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2000

 POSTTRAUMATIC STRESS

La Greca A, Silverman WK, Vernherg EM, Prinstein MJ (1996), Symptoms matic Stress: The Effects of Overwhelming Experience on Mind, Body, and
of posttraumatic stress in children after Hurricane Andrew: a prospective Society, Van der Kolk BA, McFarlane AC, Weisaeth L, eds. New York:
study.J Consult Clin Pychol64:712-723 Guilford, pp 331-358
Lyons JA (19871, Post-traumaticstress disorder in children and adolescents: a Rossman BBR, Bingham RD, Emde RN (1997), Symptomatologyand adap-
review of the literature.J Dev Behuv Pediatr 8:349-356 tive functioning for children exposed to normative stressors, dog attack,
Martinez P, Richters JE (1993),The NIMH Community Violence Project, 11: and parental violence.] A m Acad ChildAdolesc Pychiahy 36: 1089-1097
children’s distress symptoms associated with violence exposure. Pychiuhy Straus MA (1979), Measuring intrafamily conflict and violence: the Conflict
56:22-35 Tactics Scale.jMuviuge Fum 41:75-88
McCloskey LA, Figueredo AJ, Koss MP (1995),The effects of systemic family Terr L (1991), Childhood traumas: an outline and overview. AmJPychiuhy
violence on children’s mental health. Child Deu 66:1239-1261 148:10-20
McCloskey LA, Southwick K (1996), The psychosocial impact of war for Van der Kolk BA (1996),The complexityof adaptation to trauma: self-regulation,
Central American refugee children. Pediatrics 97:294-297 stimulus discrimination, and characterological development. In: Traumatic
Nolen-Hoeksema S, Morrow J (1991), A prospective study of depression and Stress: The Efick of OverwhelmingEqmience on Mind,Body, undSociey, Van
posttraumatic stress symptoms after a natural disaster: the 1989 Loma der Kolk BA, McFarlane AC, Weisaeth L, eds. New York: Guilford, pp
Prieta earthquake.jPers Soc Pychol61:115-121 182-213
Pynoos RS, Steinberg AM, Goenjian A (1996), Traumatic stress in childhood
and adolescence: recent developmentsand current controversies. In: Euu-

Low Literacy andviolence Among Adolescents in a Summer Sports Program. Terry C. Davis, PhD, Robert S. Byrd, MD, MPH,
Connie L. Arnold, PhD, Peggy Auinger, MS, Joseph A. Bocchini, Jr, M D
Pulpose: To investigate the relationship between inadequate literacy and violent behavior among adolescents. Method: This descriptive
study involved a convenience sample of 386 adolescents who participated in a summer track and field and literacy program serving
youths in low-income neighborhoods in Shreveport, Louisiana, during 1994-1996. Self-reportedviolence was measured using the Youth
Risk Behavior Survey (YRBS) and reading grade levels were measured by the Slosson Oral Reading Test-Revised (SORT-R). Results:
Youths ranged in age from 11 to 18 years; 66% were male, and 86% were African-American. Forty-three percent of adolescents tested
had below-grade reading levels (22 grades). Participants with below-grade reading skills had higher rates of self-reported violent behav-
iors compared with those reading at grade level. When gender, race, and age were controlled for, adolescents reading below grade level
were significantlymore likely to report carrying weapons [odds ratio (OR) = 1.9; 95% confidence interval (CI) 1.1-3.51, carrying guns
(OR = 2.6; CI 1.1-6.2), to have been in a physical fight at school (OR = 1.7; CI 1.l-2.6), and to have been in a physical fight resulting
in injuries requiring treatment (OR = 3.1; CI 1.6-6.1). In addition, youths reading below grade level were significantlymore likely to be
threatened at school with a weapon (OR = 2.1; CI 1.2-3.7) and to report missing days of school in the previous 30 days because they felt
unsafe at school (OR = 2.3; CI 1.3-4.3). In characterizing the violence related behaviors, we found that low reading-level adolescents
were more likely to be both aggressor/perpetratorand victim (44% vs. 3 2 % ; ~= .02) and less likely to be only a victim (6% vs. 12%;p =
.04) compared to adolescents with grade-appropriate reading skills. Conclusions: Below-grade-level reading was significantly related to
violence behaviors among adolescentswho volunteered for a summer track and field program. Longitudinal studies are needed to hrther
investigate the relationship of below-grade-level reading and aggressive/perpetrator and victim behaviors. J Adolesc Health 1999;
24:403-411.

J . A M . ACAD. C H I L D A D O L E S C . PSYCHIATRY, 39:1, JANUARY 2 0 0 0 115